Humanity’s Last PRD
Latest drop: Haiku 5.5 · 8 Oct
Tasks / Strategy

Call product-market fit

Can the model tell real product-market fit from a blended survey, a launch spike or a flattering metric, and say where to scale?

Task type v1.1 · 2 tasksLast changed 6 Oct 2026 · ChangelogDifficulty

What AI gets right here, and what you’ll still have to catch

From 16 graded outputs by 8 models. 50% were usable with at most a quick edit.

Reliably right

  1. Respects explicit constraints100% pass
    The output is a memo addressed to Sam, within 1,200 words, and respects the form and length requirements.
    GPT-6 Astra · ChatGPT · 38%, and the retention that isn't
  2. Finds who it's a must-have for100% pass
    It identifies recruiters as the must-have segment, noting what they value (candidate summaries, ATS scorecards, focus on the candidate) and builds the plan around them.
    GPT-6 Astra · ChatGPT · 38%, and the retention that isn't
  3. Spots false fit100% pass
    The memo calls out the August promotional MRR spike, the active-definition change, and the early erosion as misleading signs of fit.
    GPT-6 Astra · ChatGPT · 38%, and the retention that isn't

Where it slips

  1. Uses the supplied evidence correctly28% pass
    The claim that the cohort jumps line up with the August 3 definition change is not supported; the jumps occur in the week of July 27–August 2, one week before the change, so the evidence is misused.
    GPT-6.1 Sol · API · 38%, and the retention that isn't
  2. Gets the base of every number right44% pass
    The month-6 retention percentages are computed from incorrect sums and denominators, so the base is wrong.
    GPT-6 Luna · API · Do we have fit, or a spike?
  3. Unpicks the 38%53% pass
    It separates by persona but does not break down the 38% by tenure into ~11% very disappointed for users under two weeks and ~47% for established users, which the criterion explicitly requires.
    Haiku 5.5 · API · 38%, and the retention that isn't

How it’s graded

Each check is built on how world-class product and growth leaders do this work, as they’ve described it on Lenny’s Podcast, and names the leader it comes from.

  1. Judges fit by retention, not sign-ups

    Does the call on product-market fit rest on whether retention curves flatten, segment by segment, rather than on sign-ups, revenue growth or a survey alone?

    Passes when Reads retention by segment from the data and lets it decide where fit exists, with the survey as supporting evidence.

    Todd JacksonLenny’s PodcastSean EllisLenny’s Podcast

  2. Finds who it's a must-have for

    Does the output profile the users who would be very disappointed without the product, and aim the plan at them and at the users closest to them?

    Passes when Names the segment or persona where the product is a must-have, what they value, and builds the plan around them.

    Rahul VohraLenny’s Podcast

  3. Spots false fit

    Does the output catch evidence that looks like fit but isn't, such as revenue won through custom work, a launch spike, a promotion, or a metric whose definition changed?

    Passes when Names each piece of flattering evidence in the material and explains why it isn't fit.

    Product and growth leadersOn Lenny’s Podcast

  4. Doesn't scale before fit

    Does the recommendation avoid pouring spend or headcount into segments that haven't shown fit, and say what would justify scaling them later?

    Passes when Scales only where fit is shown, and sets the evidence that would justify widening.

    Todd JacksonLenny’s Podcast

Plus our standard checks

Uses the supplied evidence correctly · Addresses the actual decision · Respects explicit constraints · Identifies material uncertainty · Avoids unsupported claims · Produces the required deliverable · Gets the base of every number right · and 3 written for each task, which you’ll see in the tasks below.

The tasks

Read the brief, then compare up to three outputs, each with the LLM judge’s verdict on every check. Highlights mark what a PM had to fix.

The brief

You're the VP Product at Minutely. Our CEO, Sam Okafor, plans to send the draft investor update below next week to open our Series A, and to hire six account executives with the money. Sam has asked you to check it. Write a memo for Sam of no more than 1,200 words: do we have product-market fit, and for whom; what in the draft is right and what isn't; what we should tell investors instead; and where the new money should go. Everything we have is attached. The files are as the teams exported them.

What the model was given8 items: About Minutely, draft_investor_update.md, pmf_survey.csv (in-app survey, September), weekly_active_by_cohort.csv (product analytics dashboard export), core_action_by_persona.csv (share of each sign-up month who shared a summary in week 4 and week 8), mrr.csv, analytics_changelog.md, market_notes.md
About MinutelyAI meeting notes: joins your calls, writes the summary, and pushes it to your tools. Free plan, Pro ($18 per user per month) and Team ($30). 41,000 weekly active users.
draft_investor_update.md5 lines · Download
# Draft investor update (CEO)

We've hit product-market fit. 38% of users say they'd be very disappointed without Minutely, right at the 40% bar, and retention is improving fast: our newest cohorts are far stickier than June's. MRR grew 28% in August.

Plan for the round: hire six account executives to sell Team plans into enterprise engineering organisations, where the seat counts are biggest.
pmf_survey.csv (in-app survey, September)response_id,persona,plan,weeks_active,answer,would_use_instead,what_they_value M2001,Engineer,Free,14,somewhat disappointed,Zoom's notes,Standup summaries M2002,Recruiter,Pro,5,very disappointed,Back to typing notes,Candidate summaries straight into our ATS M2003,Sales,Free,8,somewhat disappointed,Gong,Saves an hour a day M2004,Recruiter,Free,1,not disappointed,Our ATS notes,Scorecards fill themselves M2005,Founder,Team,12,not disappointed,Zoom's notes,Nice to have M2006,Recruiter,Pro,10,very disappointed,Nothing does this as well,I can focus on the candidate M2007,Recruiter,Free,1,not disappointed,Back to typing notes,I can focus on the candidate M2008,Sales,Pro,6,very disappointed,Back to typing notes,CRM notes after every call M2009,Founder,Free,7,somewhat disappointed,"Nothing, I'd manage",Board meeting summaries M2010,Recruiter,Team,10,very disappointed,Nothing does this as well,I can focus on the candidate M2011,Recruiter,Team,21,very disappointed,Our ATS notes,I can focus on the candidate M2012,Recruiter,Pro,4,very disappointed,Back to typing notes,Scorecards fill themselves M2013,Engineer,Pro,18,somewhat disappointed,I don't really need notes,Standup summaries M2014,Recruiter,Pro,3,very disappointed,Nothing does this as well,Scorecards fill themselves M2015,Sales,Pro,4,very disappointed,Our CRM's notes,CRM notes after every call M2016,Recruiter,Team,6,very disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2017,Founder,Pro,13,not disappointed,Zoom's notes,Board meeting summaries M2018,Recruiter,Free,18,very disappointed,Back to typing notes,Candidate summaries straight into our ATS M2019,Recruiter,Team,15,very disappointed,Back to typing notes,I can focus on the candidate M2020,Recruiter,Free,21,very disappointed,Back to typing notes,Candidate summaries straight into our ATS M2021,Engineer,Free,18,very disappointed,Zoom's notes,Not much M2022,Sales,Free,1,somewhat disappointed,Our CRM's notes,Follow-up emails drafted M2023,Engineer,Free,5,very disappointed,Zoom's notes,Not much M2024,Recruiter,Pro,20,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2025,Recruiter,Pro,10,very disappointed,Our ATS notes,I can focus on the candidate M2026,Sales,Team,14,not disappointed,Our CRM's notes,CRM notes after every call M2027,Sales,Free,12,very disappointed,Our CRM's notes,Saves an hour a day M2028,Recruiter,Team,12,somewhat disappointed,Nothing does this as well,I can focus on the candidate M2029,Recruiter,Pro,3,somewhat disappointed,Nothing does this as well,Scorecards fill themselves M2030,Founder,Pro,8,very disappointed,"Nothing, I'd manage",Investor call notes M2031,Recruiter,Free,18,very disappointed,Back to typing notes,Scorecards fill themselves M2032,Recruiter,Team,16,somewhat disappointed,Back to typing notes,I can focus on the candidate M2033,Founder,Pro,20,somewhat disappointed,"Nothing, I'd manage",Board meeting summaries M2034,Recruiter,Free,1,somewhat disappointed,Nothing does this as well,I can focus on the candidate M2035,Sales,Team,6,somewhat disappointed,Our CRM's notes,Saves an hour a day M2036,Engineer,Pro,13,not disappointed,I don't really need notes,Not much M2037,Recruiter,Free,19,very disappointed,Our ATS notes,Scorecards fill themselves M2038,Recruiter,Pro,5,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2039,Founder,Team,21,somewhat disappointed,"Nothing, I'd manage",Board meeting summaries M2040,Recruiter,Pro,3,very disappointed,Nothing does this as well,Scorecards fill themselves M2041,Sales,Pro,2,somewhat disappointed,Our CRM's notes,CRM notes after every call M2042,Sales,Pro,8,very disappointed,Gong,Saves an hour a day M2043,Recruiter,Free,11,very disappointed,Our ATS notes,I can focus on the candidate M2044,Sales,Pro,20,very disappointed,Our CRM's notes,Saves an hour a day M2045,Sales,Pro,5,not disappointed,Our CRM's notes,Follow-up emails drafted M2046,Founder,Free,1,not disappointed,"Nothing, I'd manage",Investor call notes M2047,Sales,Team,2,very disappointed,Back to typing notes,CRM notes after every call M2048,Engineer,Team,15,somewhat disappointed,I don't really need notes,Standup summaries M2049,Sales,Team,4,somewhat disappointed,Back to typing notes,Follow-up emails drafted M2050,Sales,Free,5,somewhat disappointed,Gong,Follow-up emails drafted M2051,Sales,Pro,13,not disappointed,Back to typing notes,Saves an hour a day M2052,Recruiter,Pro,5,very disappointed,Our ATS notes,Candidate summaries straight into our ATS M2053,Founder,Free,12,not disappointed,Zoom's notes,Investor call notes M2054,Founder,Pro,14,somewhat disappointed,Zoom's notes,Board meeting summaries M2055,Recruiter,Free,0,not disappointed,Nothing does this as well,I can focus on the candidate M2056,Recruiter,Team,4,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2057,Founder,Pro,16,very disappointed,Otter,Investor call notes M2058,Recruiter,Pro,17,very disappointed,Nothing does this as well,Scorecards fill themselves M2059,Engineer,Free,1,somewhat disappointed,Nothing,Standup summaries M2060,Sales,Team,3,somewhat disappointed,Our CRM's notes,Saves an hour a day M2061,Engineer,Pro,18,somewhat disappointed,Nothing,Occasionally useful M2062,Founder,Free,10,very disappointed,"Nothing, I'd manage",Board meeting summaries M2063,Sales,Free,0,not disappointed,Our CRM's notes,CRM notes after every call M2064,Recruiter,Pro,18,very disappointed,Back to typing notes,Candidate summaries straight into our ATS M2065,Recruiter,Free,19,very disappointed,Nothing does this as well,I can focus on the candidate M2066,Founder,Free,19,very disappointed,Zoom's notes,Board meeting summaries M2067,Sales,Pro,9,not disappointed,Gong,Saves an hour a day M2068,Founder,Team,6,somewhat disappointed,Zoom's notes,Nice to have M2069,Sales,Free,11,very disappointed,Our CRM's notes,Follow-up emails drafted M2070,Sales,Pro,16,very disappointed,Back to typing notes,Saves an hour a day M2071,Sales,Free,14,very disappointed,Our CRM's notes,Follow-up emails drafted M2072,Founder,Pro,17,very disappointed,"Nothing, I'd manage",Board meeting summaries M2073,Sales,Free,1,somewhat disappointed,Back to typing notes,Follow-up emails drafted M2074,Engineer,Pro,18,very disappointed,Nothing,Standup summaries M2075,Sales,Free,1,not disappointed,Our CRM's notes,Saves an hour a day M2076,Recruiter,Free,17,very disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2077,Recruiter,Free,10,very disappointed,Our ATS notes,Scorecards fill themselves M2078,Engineer,Free,0,not disappointed,Zoom's notes,Standup summaries M2079,Founder,Pro,10,not disappointed,Zoom's notes,Board meeting summaries M2080,Recruiter,Free,1,not disappointed,Back to typing notes,Candidate summaries straight into our ATS M2081,Recruiter,Free,7,very disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2082,Founder,Pro,17,very disappointed,Otter,Board meeting summaries M2083,Engineer,Team,12,somewhat disappointed,I don't really need notes,Standup summaries M2084,Sales,Pro,15,very disappointed,Back to typing notes,Saves an hour a day M2085,Recruiter,Free,0,very disappointed,Our ATS notes,Candidate summaries straight into our ATS M2086,Sales,Pro,21,very disappointed,Back to typing notes,Saves an hour a day M2087,Founder,Pro,13,very disappointed,Otter,Nice to have M2088,Engineer,Free,20,very disappointed,Nothing,Not much M2089,Recruiter,Free,0,very disappointed,Our ATS notes,I can focus on the candidate M2090,Sales,Team,5,very disappointed,Gong,Follow-up emails drafted M2091,Engineer,Pro,6,not disappointed,I don't really need notes,Not much M2092,Recruiter,Free,18,very disappointed,Back to typing notes,I can focus on the candidate M2093,Sales,Pro,6,somewhat disappointed,Gong,CRM notes after every call M2094,Recruiter,Pro,4,very disappointed,Back to typing notes,I can focus on the candidate M2095,Sales,Free,0,somewhat disappointed,Our CRM's notes,Saves an hour a day M2096,Recruiter,Free,15,very disappointed,Our ATS notes,Candidate summaries straight into our ATS M2097,Recruiter,Pro,2,very disappointed,Nothing does this as well,I can focus on the candidate M2098,Engineer,Pro,13,not disappointed,Zoom's notes,Standup summaries M2099,Recruiter,Pro,5,somewhat disappointed,Back to typing notes,I can focus on the candidate M2100,Recruiter,Pro,16,very disappointed,Our ATS notes,I can focus on the candidate M2101,Sales,Pro,19,very disappointed,Back to typing notes,Saves an hour a day M2102,Sales,Free,1,very disappointed,Gong,CRM notes after every call M2103,Engineer,Pro,4,somewhat disappointed,Nothing,Standup summaries M2104,Recruiter,Free,0,somewhat disappointed,Our ATS notes,Candidate summaries straight into our ATS M2105,Founder,Free,0,not disappointed,Zoom's notes,Investor call notes M2106,Recruiter,Team,16,very disappointed,Our ATS notes,I can focus on the candidate M2107,Sales,Free,14,very disappointed,Gong,Follow-up emails drafted M2108,Sales,Free,1,somewhat disappointed,Gong,Follow-up emails drafted M2109,Engineer,Free,1,somewhat disappointed,Nothing,Not much M2110,Founder,Team,19,not disappointed,"Nothing, I'd manage",Nice to have M2111,Sales,Pro,4,not disappointed,Back to typing notes,CRM notes after every call M2112,Sales,Free,0,somewhat disappointed,Gong,Saves an hour a day M2113,Sales,Pro,10,very disappointed,Our CRM's notes,CRM notes after every call M2114,Engineer,Team,14,somewhat disappointed,Nothing,Not much M2115,Founder,Pro,17,not disappointed,"Nothing, I'd manage",Nice to have M2116,Engineer,Free,0,not disappointed,I don't really need notes,Not much M2117,Founder,Free,0,not disappointed,Otter,Nice to have M2118,Founder,Pro,5,very disappointed,Zoom's notes,Nice to have M2119,Sales,Free,1,not disappointed,Gong,Saves an hour a day M2120,Engineer,Pro,13,somewhat disappointed,I don't really need notes,Not much M2121,Recruiter,Team,6,very disappointed,Back to typing notes,I can focus on the candidate M2122,Sales,Pro,8,very disappointed,Back to typing notes,CRM notes after every call M2123,Engineer,Pro,3,not disappointed,I don't really need notes,Standup summaries M2124,Recruiter,Team,12,very disappointed,Our ATS notes,I can focus on the candidate M2125,Recruiter,Free,1,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2126,Sales,Team,21,somewhat disappointed,Back to typing notes,Follow-up emails drafted M2127,Recruiter,Pro,17,not disappointed,Our ATS notes,I can focus on the candidate M2128,Sales,Pro,15,somewhat disappointed,Back to typing notes,CRM notes after every call M2129,Sales,Team,7,very disappointed,Our CRM's notes,CRM notes after every call M2130,Sales,Free,10,very disappointed,Gong,Saves an hour a day M2131,Engineer,Free,0,not disappointed,I don't really need notes,Occasionally useful M2132,Recruiter,Team,11,somewhat disappointed,Our ATS notes,I can focus on the candidate M2133,Recruiter,Pro,5,very disappointed,Our ATS notes,Scorecards fill themselves M2134,Engineer,Pro,4,not disappointed,I don't really need notes,Standup summaries M2135,Founder,Free,1,somewhat disappointed,Otter,Investor call notes M2136,Recruiter,Free,0,not disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2137,Sales,Free,13,very disappointed,Gong,Saves an hour a day M2138,Recruiter,Pro,4,very disappointed,Nothing does this as well,Scorecards fill themselves M2139,Recruiter,Free,13,somewhat disappointed,Back to typing notes,Scorecards fill themselves M2140,Sales,Team,11,very disappointed,Back to typing notes,Follow-up emails drafted M2141,Engineer,Pro,7,not disappointed,Nothing,Standup summaries M2142,Engineer,Free,7,somewhat disappointed,Nothing,Not much M2143,Engineer,Free,0,not disappointed,Zoom's notes,Standup summaries M2144,Founder,Pro,8,not disappointed,Otter,Nice to have M2145,Sales,Free,6,somewhat disappointed,Our CRM's notes,Saves an hour a day M2146,Engineer,Free,2,somewhat disappointed,I don't really need notes,Not much M2147,Sales,Pro,8,somewhat disappointed,Our CRM's notes,Saves an hour a day M2148,Recruiter,Free,0,somewhat disappointed,Back to typing notes,Scorecards fill themselves M2149,Sales,Pro,2,somewhat disappointed,Our CRM's notes,CRM notes after every call M2150,Recruiter,Pro,5,very disappointed,Nothing does this as well,I can focus on the candidate M2151,Founder,Free,1,not disappointed,Otter,Board meeting summaries M2152,Recruiter,Pro,18,very disappointed,Nothing does this as well,Scorecards fill themselves M2153,Sales,Pro,20,very disappointed,Back to typing notes,Follow-up emails drafted M2154,Recruiter,Pro,20,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2155,Engineer,Free,0,not disappointed,Nothing,Standup summaries M2156,Recruiter,Free,1,somewhat disappointed,Nothing does this as well,I can focus on the candidate M2157,Engineer,Free,0,somewhat disappointed,Zoom's notes,Standup summaries M2158,Founder,Free,1,not disappointed,Zoom's notes,Board meeting summaries M2159,Sales,Pro,6,somewhat disappointed,Back to typing notes,CRM notes after every call M2160,Engineer,Free,16,somewhat disappointed,I don't really need notes,Occasionally useful M2161,Founder,Free,10,not disappointed,Zoom's notes,Investor call notes M2162,Sales,Free,1,not disappointed,Gong,Saves an hour a day M2163,Founder,Free,1,somewhat disappointed,Zoom's notes,Investor call notes M2164,Engineer,Team,3,somewhat disappointed,Nothing,Standup summaries M2165,Recruiter,Pro,7,not disappointed,Nothing does this as well,Scorecards fill themselves M2166,Recruiter,Pro,13,very disappointed,Back to typing notes,Scorecards fill themselves M2167,Engineer,Team,5,very disappointed,Zoom's notes,Not much M2168,Sales,Team,18,not disappointed,Our CRM's notes,Follow-up emails drafted M2169,Sales,Free,0,somewhat disappointed,Our CRM's notes,Follow-up emails drafted M2170,Sales,Free,10,very disappointed,Gong,Saves an hour a day M2171,Sales,Pro,17,somewhat disappointed,Gong,Saves an hour a day M2172,Founder,Pro,7,somewhat disappointed,Otter,Board meeting summaries M2173,Founder,Free,7,somewhat disappointed,Zoom's notes,Investor call notes M2174,Founder,Team,16,very disappointed,Zoom's notes,Investor call notes M2175,Sales,Free,0,very disappointed,Our CRM's notes,Saves an hour a day M2176,Sales,Pro,6,very disappointed,Gong,Follow-up emails drafted M2177,Recruiter,Pro,9,very disappointed,Our ATS notes,Scorecards fill themselves M2178,Founder,Free,14,somewhat disappointed,Otter,Board meeting summaries M2179,Recruiter,Pro,12,somewhat disappointed,Back to typing notes,I can focus on the candidate M2180,Sales,Pro,10,very disappointed,Gong,CRM notes after every call M2181,Founder,Team,7,somewhat disappointed,Otter,Nice to have M2182,Engineer,Free,20,not disappointed,Zoom's notes,Not much M2183,Engineer,Free,1,not disappointed,I don't really need notes,Not much M2184,Sales,Pro,9,not disappointed,Our CRM's notes,CRM notes after every call M2185,Recruiter,Free,1,somewhat disappointed,Nothing does this as well,Scorecards fill themselves M2186,Founder,Pro,6,very disappointed,"Nothing, I'd manage",Investor call notes M2187,Engineer,Pro,11,not disappointed,I don't really need notes,Standup summaries M2188,Sales,Pro,11,very disappointed,Gong,Follow-up emails drafted M2189,Founder,Free,16,somewhat disappointed,"Nothing, I'd manage",Nice to have M2190,Recruiter,Pro,4,somewhat disappointed,Our ATS notes,Candidate summaries straight into our ATS M2191,Founder,Pro,5,somewhat disappointed,"Nothing, I'd manage",Nice to have M2192,Recruiter,Pro,4,somewhat disappointed,Nothing does this as well,Scorecards fill themselves M2193,Sales,Free,0,very disappointed,Gong,Follow-up emails drafted M2194,Founder,Team,12,not disappointed,Zoom's notes,Nice to have M2195,Sales,Free,4,somewhat disappointed,Our CRM's notes,Follow-up emails drafted M2196,Sales,Pro,14,very disappointed,Back to typing notes,Saves an hour a day M2197,Founder,Free,1,not disappointed,"Nothing, I'd manage",Nice to have M2198,Founder,Pro,12,very disappointed,Otter,Nice to have M2199,Recruiter,Free,15,very disappointed,Nothing does this as well,I can focus on the candidate M2200,Founder,Free,3,very disappointed,"Nothing, I'd manage",Nice to have M2201,Recruiter,Team,13,not disappointed,Our ATS notes,Candidate summaries straight into our ATS M2202,Recruiter,Free,1,not disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2203,Recruiter,Free,1,somewhat disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2204,Recruiter,Team,3,somewhat disappointed,Nothing does this as well,I can focus on the candidate M2205,Sales,Free,1,somewhat disappointed,Back to typing notes,CRM notes after every call M2206,Sales,Pro,2,not disappointed,Our CRM's notes,Saves an hour a day M2207,Founder,Team,3,not disappointed,"Nothing, I'd manage",Board meeting summaries M2208,Founder,Free,0,somewhat disappointed,Otter,Nice to have M2209,Founder,Free,1,very disappointed,"Nothing, I'd manage",Nice to have M2210,Sales,Pro,14,not disappointed,Gong,Saves an hour a day M2211,Sales,Team,5,very disappointed,Our CRM's notes,CRM notes after every call M2212,Recruiter,Team,8,somewhat disappointed,Back to typing notes,Scorecards fill themselves M2213,Recruiter,Pro,9,very disappointed,Our ATS notes,Scorecards fill themselves M2214,Engineer,Team,2,not disappointed,I don't really need notes,Occasionally useful M2215,Sales,Pro,2,very disappointed,Gong,CRM notes after every call M2216,Sales,Free,1,not disappointed,Our CRM's notes,Follow-up emails drafted M2217,Founder,Free,16,somewhat disappointed,Otter,Nice to have M2218,Engineer,Free,0,not disappointed,I don't really need notes,Occasionally useful M2219,Founder,Free,1,not disappointed,Zoom's notes,Nice to have M2220,Recruiter,Team,7,very disappointed,Back to typing notes,Candidate summaries straight into our ATS M2221,Sales,Team,13,somewhat disappointed,Our CRM's notes,CRM notes after every call M2222,Founder,Free,0,somewhat disappointed,"Nothing, I'd manage",Investor call notes M2223,Founder,Free,6,very disappointed,Otter,Board meeting summaries M2224,Engineer,Pro,13,not disappointed,Zoom's notes,Standup summaries M2225,Recruiter,Free,13,not disappointed,Our ATS notes,I can focus on the candidate M2226,Engineer,Pro,18,not disappointed,I don't really need notes,Not much M2227,Founder,Free,5,very disappointed,Otter,Nice to have M2228,Sales,Pro,4,not disappointed,Gong,CRM notes after every call M2229,Engineer,Pro,9,very disappointed,Zoom's notes,Standup summaries M2230,Sales,Pro,15,very disappointed,Back to typing notes,CRM notes after every call M2231,Founder,Team,21,very disappointed,"Nothing, I'd manage",Nice to have M2232,Founder,Free,1,somewhat disappointed,Zoom's notes,Nice to have M2233,Recruiter,Free,6,somewhat disappointed,Nothing does this as well,I can focus on the candidate M2234,Recruiter,Free,15,very disappointed,Our ATS notes,Candidate summaries straight into our ATS M2235,Sales,Pro,21,somewhat disappointed,Gong,CRM notes after every call M2236,Founder,Team,4,somewhat disappointed,Otter,Nice to have M2237,Recruiter,Free,20,somewhat disappointed,Back to typing notes,I can focus on the candidate M2238,Recruiter,Pro,8,not disappointed,Back to typing notes,Candidate summaries straight into our ATS M2239,Sales,Free,2,very disappointed,Back to typing notes,CRM notes after every call M2240,Recruiter,Pro,4,somewhat disappointed,Back to typing notes,I can focus on the candidate M2241,Founder,Pro,13,somewhat disappointed,Otter,Nice to have M2242,Recruiter,Team,18,very disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2243,Founder,Free,0,very disappointed,Otter,Investor call notes M2244,Founder,Free,10,very disappointed,Otter,Nice to have M2245,Sales,Free,5,somewhat disappointed,Gong,Follow-up emails drafted M2246,Founder,Team,18,very disappointed,"Nothing, I'd manage",Board meeting summaries M2247,Recruiter,Free,17,very disappointed,Our ATS notes,Scorecards fill themselves M2248,Recruiter,Pro,10,somewhat disappointed,Back to typing notes,I can focus on the candidate M2249,Sales,Free,1,not disappointed,Gong,Follow-up emails drafted M2250,Recruiter,Free,1,very disappointed,Our ATS notes,Candidate summaries straight into our ATS M2251,Engineer,Free,0,not disappointed,Zoom's notes,Not much M2252,Sales,Pro,21,very disappointed,Our CRM's notes,Saves an hour a day M2253,Sales,Free,0,not disappointed,Back to typing notes,Saves an hour a day M2254,Sales,Free,1,somewhat disappointed,Our CRM's notes,Follow-up emails drafted M2255,Engineer,Pro,5,not disappointed,Zoom's notes,Not much M2256,Sales,Pro,2,somewhat disappointed,Gong,Saves an hour a day M2257,Sales,Free,2,somewhat disappointed,Our CRM's notes,Follow-up emails drafted M2258,Sales,Pro,8,very disappointed,Back to typing notes,CRM notes after every call M2259,Sales,Free,1,not disappointed,Our CRM's notes,CRM notes after every call M2260,Founder,Free,1,somewhat disappointed,Zoom's notes,Investor call notes M2261,Sales,Free,0,not disappointed,Gong,CRM notes after every call M2262,Sales,Pro,18,somewhat disappointed,Back to typing notes,CRM notes after every call M2263,Founder,Free,18,somewhat disappointed,"Nothing, I'd manage",Investor call notes M2264,Recruiter,Pro,4,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2265,Engineer,Free,1,somewhat disappointed,Zoom's notes,Occasionally useful M2266,Sales,Free,0,somewhat disappointed,Our CRM's notes,Saves an hour a day M2267,Engineer,Free,18,very disappointed,Nothing,Standup summaries M2268,Recruiter,Free,10,very disappointed,Our ATS notes,Scorecards fill themselves M2269,Sales,Free,1,not disappointed,Our CRM's notes,Follow-up emails drafted M2270,Sales,Pro,3,not disappointed,Gong,CRM notes after every call M2271,Sales,Pro,7,very disappointed,Our CRM's notes,Saves an hour a day M2272,Engineer,Free,0,not disappointed,Zoom's notes,Occasionally useful M2273,Founder,Free,1,not disappointed,Otter,Board meeting summaries M2274,Founder,Free,1,not disappointed,Otter,Investor call notes M2275,Sales,Team,19,very disappointed,Gong,Saves an hour a day M2276,Sales,Pro,6,somewhat disappointed,Gong,Follow-up emails drafted M2277,Recruiter,Pro,7,not disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2278,Sales,Free,0,not disappointed,Back to typing notes,Saves an hour a day M2279,Sales,Free,1,not disappointed,Back to typing notes,CRM notes after every call M2280,Recruiter,Free,1,somewhat disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2281,Recruiter,Team,3,somewhat disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2282,Engineer,Free,0,somewhat disappointed,Zoom's notes,Standup summaries M2283,Sales,Free,0,somewhat disappointed,Back to typing notes,CRM notes after every call M2284,Founder,Free,0,not disappointed,Otter,Nice to have M2285,Sales,Pro,5,very disappointed,Back to typing notes,Follow-up emails drafted M2286,Sales,Pro,16,very disappointed,Our CRM's notes,Saves an hour a day M2287,Sales,Pro,18,somewhat disappointed,Back to typing notes,Saves an hour a day M2288,Founder,Free,0,somewhat disappointed,Zoom's notes,Investor call notes M2289,Engineer,Free,1,not disappointed,I don't really need notes,Not much M2290,Sales,Team,21,very disappointed,Back to typing notes,CRM notes after every call M2291,Engineer,Free,1,somewhat disappointed,Nothing,Not much M2292,Sales,Pro,11,somewhat disappointed,Gong,Follow-up emails drafted M2293,Founder,Pro,19,somewhat disappointed,Otter,Nice to have M2294,Engineer,Pro,21,somewhat disappointed,I don't really need notes,Not much M2295,Sales,Free,19,somewhat disappointed,Our CRM's notes,CRM notes after every call M2296,Sales,Free,1,somewhat disappointed,Back to typing notes,CRM notes after every call M2297,Engineer,Pro,8,not disappointed,I don't really need notes,Not much M2298,Sales,Free,1,not disappointed,Gong,CRM notes after every call M2299,Sales,Team,7,very disappointed,Back to typing notes,CRM notes after every call M2300,Founder,Team,9,very disappointed,Zoom's notes,Nice to have M2301,Engineer,Pro,21,very disappointed,Nothing,Not much M2302,Engineer,Free,7,very disappointed,Zoom's notes,Standup summaries M2303,Recruiter,Free,1,somewhat disappointed,Back to typing notes,Scorecards fill themselves M2304,Sales,Free,0,not disappointed,Our CRM's notes,CRM notes after every call M2305,Recruiter,Free,8,not disappointed,Our ATS notes,Scorecards fill themselves M2306,Sales,Pro,10,not disappointed,Back to typing notes,Saves an hour a day M2307,Recruiter,Pro,2,very disappointed,Nothing does this as well,I can focus on the candidate M2308,Recruiter,Pro,13,very disappointed,Nothing does this as well,Scorecards fill themselves M2309,Recruiter,Pro,5,not disappointed,Back to typing notes,I can focus on the candidate M2310,Sales,Team,21,somewhat disappointed,Gong,Saves an hour a day M2311,Engineer,Pro,15,not disappointed,I don't really need notes,Occasionally useful M2312,Sales,Pro,7,somewhat disappointed,Back to typing notes,Saves an hour a day M2313,Engineer,Pro,2,very disappointed,Zoom's notes,Occasionally useful M2314,Engineer,Team,12,not disappointed,Zoom's notes,Not much M2315,Founder,Free,15,very disappointed,"Nothing, I'd manage",Investor call notes M2316,Founder,Free,10,very disappointed,"Nothing, I'd manage",Nice to have M2317,Sales,Free,0,somewhat disappointed,Our CRM's notes,Saves an hour a day M2318,Sales,Pro,17,somewhat disappointed,Gong,Follow-up emails drafted M2319,Engineer,Free,12,somewhat disappointed,I don't really need notes,Standup summaries M2320,Sales,Free,16,somewhat disappointed,Gong,Saves an hour a day M2321,Engineer,Pro,12,very disappointed,Nothing,Not much M2322,Recruiter,Team,7,somewhat disappointed,Our ATS notes,Scorecards fill themselves M2323,Engineer,Pro,3,not disappointed,I don't really need notes,Occasionally useful M2324,Sales,Pro,8,somewhat disappointed,Our CRM's notes,Follow-up emails drafted M2325,Engineer,Pro,15,not disappointed,Nothing,Not much M2326,Sales,Team,14,very disappointed,Back to typing notes,Follow-up emails drafted M2327,Founder,Team,21,somewhat disappointed,Zoom's notes,Investor call notes M2328,Sales,Team,14,very disappointed,Gong,CRM notes after every call M2329,Sales,Free,9,very disappointed,Gong,Saves an hour a day M2330,Founder,Free,14,not disappointed,Otter,Nice to have M2331,Sales,Pro,19,somewhat disappointed,Gong,Saves an hour a day M2332,Engineer,Pro,4,somewhat disappointed,Zoom's notes,Not much M2333,Sales,Free,0,not disappointed,Back to typing notes,Saves an hour a day M2334,Sales,Pro,7,very disappointed,Our CRM's notes,Saves an hour a day M2335,Sales,Free,9,somewhat disappointed,Our CRM's notes,CRM notes after every call M2336,Engineer,Free,1,not disappointed,Nothing,Not much M2337,Sales,Free,3,not disappointed,Our CRM's notes,CRM notes after every call M2338,Founder,Free,4,very disappointed,Zoom's notes,Board meeting summaries M2339,Engineer,Pro,18,not disappointed,I don't really need notes,Occasionally useful M2340,Engineer,Free,0,somewhat disappointed,Zoom's notes,Not much M2341,Recruiter,Free,13,very disappointed,Our ATS notes,Scorecards fill themselves M2342,Sales,Team,5,very disappointed,Back to typing notes,CRM notes after every call M2343,Engineer,Team,4,somewhat disappointed,Zoom's notes,Not much M2344,Founder,Pro,5,very disappointed,Zoom's notes,Nice to have M2345,Founder,Pro,15,somewhat disappointed,"Nothing, I'd manage",Investor call notes M2346,Founder,Team,4,somewhat disappointed,"Nothing, I'd manage",Investor call notes M2347,Engineer,Free,1,very disappointed,I don't really need notes,Not much M2348,Recruiter,Pro,11,very disappointed,Back to typing notes,Scorecards fill themselves M2349,Founder,Free,21,not disappointed,"Nothing, I'd manage",Board meeting summaries M2350,Recruiter,Free,0,not disappointed,Our ATS notes,Candidate summaries straight into our ATS M2351,Recruiter,Pro,17,very disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2352,Recruiter,Pro,12,very disappointed,Back to typing notes,I can focus on the candidate M2353,Sales,Free,5,not disappointed,Gong,Follow-up emails drafted M2354,Engineer,Free,1,not disappointed,I don't really need notes,Standup summaries M2355,Founder,Pro,4,somewhat disappointed,Zoom's notes,Nice to have M2356,Sales,Free,13,very disappointed,Gong,Follow-up emails drafted M2357,Recruiter,Free,0,very disappointed,Back to typing notes,Candidate summaries straight into our ATS M2358,Recruiter,Team,8,very disappointed,Nothing does this as well,Candidate summaries straight into our ATS M2359,Founder,Free,8,not disappointed,Otter,Investor call notes M2360,Founder,Free,0,somewhat disappointed,Zoom's notes,Board meeting summaries
weekly_active_by_cohort.csv (product analytics dashboard export)cohort_week_starting,signups,week_1_active_pct,week_2_active_pct,week_3_active_pct,week_4_active_pct,week_5_active_pct,week_6_active_pct,week_7_active_pct,week_8_active_pct,week_9_active_pct,week_10_active_pct,week_11_active_pct,week_12_active_pct 2026-06-01,1491,62.5,47.4,39.6,35.5,30.9,28.7,27.3,25.7,38.6,39.4,38.2,38.5 2026-06-08,1710,62.5,48.4,40.0,35.1,30.6,28.6,27.4,39.6,39.2,39.5,38.5,38.1 2026-06-15,1424,62.4,48.1,40.6,34.8,30.7,28.9,40.7,39.9,38.5,39.1,37.9,38.1 2026-06-22,1546,61.7,48.3,39.7,34.8,31.1,42.5,40.5,39.7,39.0,38.8,38.3,37.7 2026-06-29,1672,62.6,48.1,39.4,35.0,44.5,42.5,41.1,40.5,38.8,38.8,37.4,38.2 2026-07-06,1773,61.8,48.4,39.9,48.7,44.6,43.4,40.9,40.6,39.1,38.6,37.4,38.2 2026-07-13,1638,61.8,48.5,54.3,49.3,45.3,43.4,40.7,40.5,38.7,38.9,38.4,37.9 2026-07-20,1831,62.3,61.9,54.1,49.4,44.9,43.1,41.0,40.4,38.4,38.9,38.4, 2026-07-27,1522,76.2,62.4,54.3,48.5,44.8,43.4,41.4,40.4,39.4,38.8,, 2026-08-03,1687,75.7,61.8,54.1,49.1,45.1,42.9,40.8,40.0,38.6,,, 2026-08-10,1769,76.4,62.2,53.7,48.4,45.5,43.2,41.0,39.4,,,, 2026-08-17,1731,75.6,61.6,53.9,48.5,44.4,43.4,41.5,,,,, 2026-08-24,1453,76.1,62.3,54.3,49.6,44.9,43.0,,,,,, 2026-08-31,1603,76.5,61.6,54.1,48.5,45.4,,,,,,, 2026-09-07,1765,66.5,54.2,48.3,44.0,,,,,,,, 2026-09-14,1652,67.2,54.5,47.7,,,,,,,,,
core_action_by_persona.csv (share of each sign-up month who shared a summary in week 4 and week 8)persona,signup_month,week_4_shared_a_summary_pct,week_8_shared_a_summary_pct Recruiter,2026-06,66,58 Recruiter,2026-07,65,57 Recruiter,2026-08,66,58 Recruiter,2026-09,57, Sales,2026-06,50,41 Sales,2026-07,49,40 Sales,2026-08,51,42 Sales,2026-09,44, Founder,2026-06,39,30 Founder,2026-07,38,29 Founder,2026-08,40,31 Founder,2026-09,34, Engineer,2026-06,19,13 Engineer,2026-07,18,12 Engineer,2026-08,18,12 Engineer,2026-09,15,
mrr.csv4 rows · Download
month,mrr_usd,new_mrr_usd,churned_mrr_usd,note
2026-06,182000,21000,7400,
2026-07,196000,22100,8100,
2026-08,251000,63800,8800,"Team plan annual prepay promotion, 40% off, ran 1-31 Aug"
2026-09,262000,21500,10500,
analytics_changelog.md5 lines · Download
# Analytics changelog

- 2026-07-14: Added persona to sign-up (self-reported).
- 2026-08-03: "Active user" now includes opening the daily summary email, as well as recording, opening or sharing a summary in the app. Requested by Growth so email engagement counts.
- 2026-09-01: Added "shared a summary" as a tracked core action; backfilled to June from event logs.
market_notes.md5 lines · Download
# Market notes (from the CEO's chief of staff)

- 8 Sep 2026: Zoom announced free AI meeting notes for every paid Zoom plan.
- Two Team-plan customers asked in September whether they still need us "now that Zoom does this".
- Our closest paid competitors (Otter, Fathom) haven't changed pricing.
What a strong answer doesThe answer key the graders mark against

Says Minutely has fit with recruiters and probably sales teams, not with everyone, and that the draft overstates it in three ways. (1) The survey: 38% is a blend that includes 90 people who've used Minutely for under two weeks (11% very disappointed); among established users it's 47%, and by persona 62% of recruiters and 49% of sales users against 24% of engineers. The case for fit is stronger and narrower than the draft says. (2) Retention isn't 'improving fast': from 3 August 'active' includes opening the summary email, so every cohort's curve jumps about 14 points that week, and August cohorts only look stickier because of the definition. On the consistent core action (shared a summary), week-8 retention is flat across June to August for each persona, and September cohorts' week-4 retention fell (recruiters 66 to 57, sales 51 to 44) after Zoom launched free notes on 8 September: an early erosion signal the draft ignores. (3) August's 28% MRR growth is mostly the Team annual prepay promotion at 40% off (new MRR $63,800 against about $21,500 in a normal month), not demand. Recommends not hiring AEs to sell into enterprise engineering, the persona with the weakest fit and the most exposure to Zoom; put the money into recruiters and sales teams (integrations with ATS and CRM, where value is specific). Rewrites the investor message honestly (strong fit in a named segment, with the consistent retention numbers) and fixes the active-user definition.

Critical failures (cap the score at 40)
  • Endorses sending the draft's claim that retention is improving fast
  • Recommends hiring the six AEs to sell into enterprise engineering organisations
Task

Staff-level · v1.0

The outputs

8 setups answered this brief.

GPT-6 AstraChatGPT98Usable with a quick editGPT-6.1 SolAPI85Needs more workHaiku 5.5API85Usable with a quick editOpus 5.5Claude80Usable with a quick editGPT-6 LunaAPI78Needs more workSonnet 5.5API78Usable with a quick editGemini 3.8 FlashAPI58Needs more workGemini 3.5 Flash-LiteGemini50Needs more work
GPT-6.1 SolAPI

Needs more than a quick edit12 of 14 checks passed

The memo contains a factual error about the cohort jump aligning with the August 3 definition change; a PM would need to correct that before use. Otherwise the reasoning is sound, but the error undermines a key piece of evidence.

Output

To: Sam Okafor From: VP Product Subject: Series A: fund the recruiting workflow, not an enterprise-engineering sales expansion

Recommendation

We have credible early evidence of product-market fit among recruiters, promising evidence among sales users, and weak evidence among engineers and founders. We have not established company-wide PMF or a repeatable enterprise sales motion.

Rewrite the update before sending it. Raise to strengthen and commercialise the recruiting workflow—not to hire six account executives into our weakest demonstrated segment.

1. Where we have fit—and what remains unproven

The survey’s headline is mathematically correct: 137 of 360 respondents, or 38.1%, would be “very disappointed” without Minutely. But the aggregate conceals materially different customer experiences.

It also includes 90 respondents with only zero or one week active. As an exposure sensitivity check, excluding those respondents produces:

PersonaAll respondents: very disappointedRespondents with ≥2 weeks active
------:---:
Recruiter54/100 — 54.0%50/80 — 62.5%
Sales47/120 — 39.2%44/90 — 48.9%
Founder24/80 — 30.0%22/60 — 36.7%
Engineer12/60 — 20.0%11/40 — 27.5%
Total137/360 — 38.1%127/270 — 47.0%

This is a disclosed sensitivity analysis, not permission to replace 38% with the most flattering number. Two weeks active does not prove meaningful product exposure, and an in-app survey misses many users who already left. We lack response rates and representativeness checks. The 40% heuristic is a diagnostic, not a certification.

Recruiters are the strongest segment because sentiment and behaviour agree. Across June–August signup months, 65–66% shared a summary in week four and 57–58% in week eight. They value candidate summaries pushed into the ATS, populated scorecards, and being able to focus on the candidate. Their alternatives include manual note-taking and inferior ATS notes. The product’s advantage appears to be completing a recruiting workflow, not merely generating a transcript.

Sales is a plausible second segment: 40–42% shared a summary in week eight, with value around CRM notes, follow-ups and time saved. Validate it without splitting our initial focus.

Engineering is the wrong expansion thesis: only 12–13% shared a summary in week eight. Founders are also weaker, at 29–31%.

These are user-level signals. We cannot yet identify the winning recruiting company size, buyer, acquisition channel or contract structure. Nor do the exports establish paid-account retention, expansion, gross margin or sales efficiency. We should call recruiter PMF credible but provisional—not claim proven enterprise PMF.

2. What the draft gets right—and wrong

“38% … right at the 40% bar”

The calculation is right; the conclusion is not. It is below the heuristic, and averaging across personas obscures the strongest opportunity. We should report the segment evidence and its limitations.

“Retention is improving fast”

This is not supported.

On August 3, “active” expanded to include opening the daily summary email. The cohort increases line up with that calendar change: June 1 jumps in week nine, June 8 in week eight, June 15 in week seven, and so on. This is strong evidence of a measurement discontinuity, not progressively better cohorts.

The consistently defined core-action export shows essentially flat June–August performance within every persona. September week-four sharing is lower across all four segments—for example, recruiters fall to 57% and sales to 44%. September week-eight results are unavailable.

Recompute activity using one definition throughout. Also audit denominator and persona provenance: persona collection began July 14, so the source of historical labels needs explanation. The two exports should not be treated as interchangeable measures of retention.

“MRR grew 28% in August”

Correct: $196,000 to $251,000 is 28.1% growth. But August included a 40%-off annual Team prepay promotion. New MRR rose to $63,800 versus roughly $21,000–$22,000 in surrounding months; September growth slowed to 4.4%, reaching $262,000.

The promotion does not automatically invalidate MRR. Finance should verify that annual contracts are recognised monthly at their actual discounted recurring value, not counted as upfront cash. Disclose the promotion and track those customers’ usage, renewal and subsequent expansion before presenting August as repeatable momentum.

“Enterprise engineering … where seat counts are biggest”

Seat potential is not demand. We have no evidence of enterprise engineering buyers, repeatable wins or attractive economics—and engineering has our weakest behavioural results.

Zoom’s September 8 free-notes announcement, plus two Team customers questioning our necessity, makes generic notes a particularly risky proposition. It may contribute to September weakness, but these data do not establish causation.

3. What we should tell investors instead

Suggested replacement:

> Minutely has 41,000 weekly active users under our current definition and $262,000 in September MRR. Our strongest evidence of product-market fit is in recruiting: 54% of recruiter survey respondents would be very disappointed without us, rising to 62.5% among those with at least two weeks active. June–August recruiter cohorts show 57–58% sharing a summary in week eight. > > Recruiters value interview documentation delivered into their ATS and scorecards—not just meeting notes. Sales is a promising adjacent segment. > > August MRR grew 28.1%, supported by a discounted annual Team promotion; September grew 4.4%. Our headline activity metric changed in August, so we are not claiming a demonstrated retention improvement. We are raising to deepen the recruiting workflow, improve activation and prove repeatable paid retention and acquisition before scaling sales.

Make the metric definitions, survey denominators and promotion treatment available in diligence.

4. Where the new money should go

Do not hire six AEs now. Proposed allocation:

  • 45%: Recruiting workflow and reliability. Prioritise ATS delivery, useful structured scorecards, summary quality, meeting capture and integration reliability. Validate consent, permissions and data-handling requirements with target recruiting buyers.
  • 25%: Activation and retention. Shorten time to the first successful interview-to-ATS outcome. Investigate September’s decline and measure repeated workflow completion, not email opens.
  • 20%: Focused commercial validation and customer success. Founder-led selling, onboarding and a small recruiting-focused commercial team; at most one initial AE after clarifying the buyer and offer.
  • 10%: Measurement and economics. Consistent event definitions, historical-data audits, paid-account cohorts, promotion tracking, inference costs and acquisition economics.

Release further sales hiring only after we demonstrate repeatable wins in a defined recruiting ICP, sustained paid retention, credible gross margins and acceptable acquisition payback.

The Series A should scale a validated recruiting workflow—not purchase six people to search for one.

Claims the judge couldn’t find in the brief

  • The cohort increases line up with that calendar change: June 1 jumps in week nine, June 8 in week eight, June 15 in week seven, and so on.

Grades and run details

Decision model 96 · LLM judge 80
Decision model checks
  • passUses the supplied evidence correctly37%
  • passAddresses the actual decision96%
  • passRespects explicit constraints58%
  • passIdentifies material uncertainty96%
  • passAvoids unsupported claims58%
  • passProduces the required deliverable87%
  • passCatches the redefined metric100%
  • passUnpicks the 38%68%
  • passSees the early erosion82%
  • partialJudges fit by retention, not sign-ups36%
  • passGets the base of every number right26%
  • passFinds who it's a must-have for100%
  • passSpots false fit96%
  • passDoesn't scale before fit99%
Run
Run
#1
API response time
1.7 min
Submitted
5 Oct 2026
Opus 5.5Claude

Usable with a quick edit11 of 14 checks passed

The memo is well-reasoned and actionable; the unsupported churn-rate estimate can be removed with a quick edit, and no other evidence needs re-checking.

Output

Memo: Series A investor update

To: Sam Okafor From: VP Product Re: Draft investor update and use of funds

Bottom line

We have product-market fit with recruiters. We have it plausibly with salespeople. We don't have it with founders, and we're furthest from it with engineers. Two of the draft's three metrics would mislead investors, and both problems would surface in diligence. The hiring plan aims six AEs at our weakest segment, the one most exposed to Zoom's free notes. I recommend a recruiting-led round story and a recruiting-led use of funds.

1. Do we have product-market fit, and for whom?

The survey's 38% is accurate: 137 of 360 respondents said "very disappointed." But the blended number hides the answer. The standard Ellis method also excludes people who have barely used the product, and 90 of our respondents were in their first week or two.

PersonaRespondentsVery disappointed (all)Very disappointed (2+ weeks active)Shared a summary in week 8 (Aug sign-ups)What they'd use instead
Recruiter10054%63% (50/80)58%"Nothing does this as well", typing, ATS notes. None name Zoom
Sales12039%49% (44/90)42%Gong, CRM notes
Founder7532%40% (22/55)31%"Nothing, I'd manage", Zoom, Otter. Value: "nice to have"
Engineer6518%24% (11/45)12%"Don't really need notes"; ~1 in 3 name Zoom
All36038%47% (127/270)n/an/a

Recruiters are our market. They pass the bar comfortably and keep doing the core action. What they value is workflow, not transcription: scorecards that fill themselves, summaries pushed into the ATS, and focus on the candidate. That is defensible against Zoom.

Sales clears the bar among established users. It is a crowded market, though. Respondents name Gong and their CRM as alternatives, and what they value is generic ("saves an hour a day").

Founders sit at the line, and say themselves that it's nice to have.

Engineers clearly don't need us. Only 1 of 8 engineers on Team plans is "very disappointed."

2. The draft: what's right and what isn't

Right: - 38% is the correct blended figure. - MRR did grow 28% in August ($196k to $251k).

Wrong or misleading:

"Right at the 40% bar." 38% is below the bar. The more important point is that the blended number is the wrong one to lead with. The segment story is far stronger and is the one investors will respect.

"Newest cohorts are far stickier than June's." This is a measurement artefact. - On 3 August we started counting "opened the daily summary email" as active (see the analytics changelog). - Every cohort's curve jumps about 13 points in that exact week. For example, the 1 June cohort goes from 25.7% to 38.6% in its week 9, with no product change. - Measured consistently, nothing improved: - Every cohort from 6 July to 31 August is within about a point of 49% at week 4 and about 40% at week 8. - Summary-sharing, which is unaffected by the definition change, is flat June to August for every persona. Recruiters, for example, are at 66% in week 4 and 58% in week 8 each month. - The same change inflates the 41,000 WAU figure. On the old definition it is probably nearer two-thirds of that. We need to re-run it.

"MRR grew 28% in August." This is a one-off. - About $42k of August's $63.8k new MRR came from the 40%-off Team annual prepay promotion. - Underlying net growth is about 7% (July), about 6–7% (August excluding the promo) and 4.4% (September). - The promo revenue is discounted and pulled forward. Its renewal at full price is unproven. - Churned MRR has risen every month, from $7.4k to $10.5k. Against the base that can actually churn monthly (excluding prepay), September churn is about 5%. - Finance should also confirm the prepay was booked at one-twelfth per month, not as a lump.

Missing: Zoom. Zoom made notes free on paid plans on 8 September. September's numbers soften right after: - Week-1 activity fell from about 76% to about 67%. - Week-4 activity fell from about 48.5% to 44%. - Week-4 sharing fell for every persona. Recruiters dropped 66% to 57%, and engineers 18% to 15%. - Two Team customers have asked whether they still need us.

The September data is partial, but investors will ask about Zoom on the first call. We should address it before they do.

The plan: AEs into enterprise engineering. This targets the segment with: - the weakest pull (18% very disappointed, 12% still sharing summaries in week 8), - the most users who say they don't need notes, and - the most users who already treat Zoom as their alternative.

Large seat counts don't help if the seats don't use the product. We would be paying AEs to sell Team plans that churn at renewal, against a free bundle.

3. What we should tell investors instead

A suggested replacement:

> Minutely has found product-market fit with recruiters. Among recruiters active two weeks or more, 63% would be very disappointed without us, and 58% are still sharing summaries every week two months after sign-up. They value what generic note-takers don't do: scorecards that fill themselves and summaries written straight into the ATS. Not one recruiter we surveyed named Zoom as an alternative. Sales users are our second-strongest segment (49%). > > Retention is stable rather than improving. On a consistent definition, cohorts since July hold about 49% weekly activity at week 4 and about 40% at week 8. MRR is $262k. Underlying growth is 5–7% a month, plus a one-off August uplift from an annual prepay promotion. > > Zoom's free notes, launched in September, will commoditise general meeting notes. Early September cohorts are softer, and we expect the impact to concentrate in our generic use cases. This round funds our move into recruiting workflow, where we already win.

This is a smaller headline but a stronger story. It is specific and defensible, and it shows we understand our own data. The cohort-definition change and the promo would both be found in diligence. Being caught on them would cost far more than a lower headline.

4. Where the new money should go

1. Product depth for recruiting (largest share). - More ATS integrations, and deeper ones. - Structured scorecards and interview-kit support. - Hiring-manager debrief summaries. - Candidate consent and data-retention controls, which recruiting teams' legal reviews will require. - This is our moat against Zoom, and it is engineering-heavy. 2. A recruiting go-to-market, staged. - Start with product-led motions: ATS marketplace listings and partnerships, and recruiting-team trials. - Add one or two AEs selling Team plans to talent-acquisition teams and agencies once we've shown a repeatable conversion path. - Add more AEs only when that path works. Six from day one is premature for any segment. 3. Retention for recruiters and sales. Treat the September dip as urgent. Study what separates recruiters who are still sharing at week 8 from those who aren't, and fix activation for the rest. 4. Measurement. - Report a core-action metric (shared a summary) alongside activity. - Restate WAU on both definitions. - Track promo cohorts' renewal separately. 5. Sales as a second bet, founders and engineers as free-plan users. Revisit sales once recruiting is working. Do not build for engineering orgs.

Before anything goes out

  • Check where persona labels for older users came from. Persona was only added at sign-up on 14 July and is self-reported, yet June cohorts and long-tenured respondents carry labels. If those were inferred, the persona cuts are directional until we verify them.
  • Treat the sample as indicative. The survey has 360 in-app responses out of 41,000 users. The pattern is clear, but the precise percentages aren't.
  • Re-pull September once cohorts mature, to size the Zoom effect.

I'm happy to redraft the update with you this week.

Claims the judge couldn’t find in the brief

  • September churn is about 5% against the base that can actually churn monthly (excluding prepay).
  • On the old definition WAU is probably nearer two-thirds of 41,000.

Grades and run details

Decision model 79 · LLM judge 80
Decision model checks
  • failUses the supplied evidence correctly55%
  • passAddresses the actual decision79%
  • passRespects explicit constraints50%
  • passIdentifies material uncertainty84%
  • partialAvoids unsupported claims39%
  • passProduces the required deliverable79%
  • passCatches the redefined metric100%
  • passUnpicks the 38%87%
  • passSees the early erosion96%
  • partialJudges fit by retention, not sign-ups60%
  • failGets the base of every number right47%
  • passFinds who it's a must-have for100%
  • passSpots false fit95%
  • passDoesn't scale before fit93%
Run
Run
#1
Time to output
2.8 min
Submitted
5 Oct 2026
GPT-6 LunaAPI

Needs more than a quick edit11 of 14 checks passed

The memo fails to break down the 38% survey result by tenure and persona, and makes an unsupported claim about the survey being a convenience sample; a PM would need to add that analysis and correct the claim.

Output

Memo

To: Sam Okafor From: VP Product Subject: Series A update: PMF evidence and use of funds

Do we have product-market fit—and for whom?

We have promising, focused evidence of product-market fit in recruiting workflows, but not evidence that Minutely has broad PMF across meeting notes—or specifically in engineering organizations.

Recruiters show the strongest recurring behavior: 65–66% of June–August signup cohorts shared a summary in week 4, and 57–58% did so in week 8. Survey responses also repeatedly describe concrete value: focusing on candidates, filling scorecards, and sending candidate summaries into an ATS. Sales is a plausible second segment, with week-4 sharing around 49–51% and week-8 around 40–42%. Engineering is much weaker: 18–19% shared in week 4 and 12–13% in week 8.

That supports prioritizing recruiting, not declaring company-wide PMF. The survey is an in-app convenience sample, not necessarily representative of users or paying customers. Its 38% “very disappointed” result is below the 40% rule of thumb—not “right at” the bar—and an aggregate across very different use cases can conceal a strong segment and weak ones.

What is right—and what isn’t—in the draft

  • “38%…right at the 40% bar” is wrong. It misses the benchmark, and the blended figure is not a sound verdict on PMF by itself. The useful question is which users depend on Minutely and why.
  • “Newest cohorts are far stickier” is not supported. On August 3, “active” expanded to include opening the daily summary email. The sharp step-up in reported week-1 activity—from about 62% in earlier cohorts to roughly 76% in late-July/August cohorts—coincides with that change. Later-week figures for older cohorts also jump around the same point. These cohorts are not directly comparable on this definition; don’t present the change as a retention improvement. The separate sharing metric is more informative, and it shows stable recruiting and sales behavior through August. September week-4 sharing is lower across personas and merits investigation.
  • “MRR grew 28% in August” is arithmetically true, but misleading without context. MRR rose from $196k to $251k. August included a 40%-off annual-prepay promotion: new MRR was $63.8k, versus $21–22k in June and July. September MRR was $262k, with $21.5k new and $10.5k churned. Describe the August increase as promotion-driven, not a new underlying growth rate.
  • Six AEs selling Team plans to enterprise engineering is unsupported. Engineering has the weakest repeat sharing in the data. Meanwhile, Zoom announced free AI meeting notes for paid plans, and two Team customers have already questioned whether they still need Minutely. Seat count alone is not evidence of demand or a winning sales motion.

What we should tell investors instead

Be direct: “We see our strongest product-market-fit signal in recruiting teams. June–August cohorts consistently share Minutely summaries, and users value candidate-focused workflows, scorecards, and ATS handoff. Sales is a promising adjacent use case; engineering and general-purpose meeting notes are less compelling today. Our aggregate survey result is 38%, below the common 40% benchmark, and our active-user definition changed in August, so we are not claiming broad PMF or a retention step-change. August MRR growth was boosted by a discounted annual-prepay promotion; September growth was more modest. We’re focusing the next phase on proving repeatable value, retention, and willingness to pay in recruiting.”

Before circulating, validate survey counts and segment denominators, and present paid conversion, retention, expansion, and churn by persona and plan if available. Don’t use the cohort active-user chart to claim improvement until it is rebuilt on a consistent definition.

Where the new money should go

Do not hire six AEs now. First invest in the recruiting workflow: ATS integrations, candidate summaries and scorecards, team adoption, and the trust, privacy, and reliability requirements of handling interview data. Pair that with customer research and measurement that ties shared summaries to paid conversion, retention, and expansion.

Run a focused recruiting go-to-market test: identify the buyer, sales cycle, contract size, and repeatable acquisition channel. Consider one experienced vertical seller or solutions lead once that motion is clearer—not a six-person team selling into a segment with weak usage evidence. Keep sales as a measured adjacent experiment, and preserve capital until the recruiting thesis is proven.

Claims the judge couldn’t find in the brief

  • The survey is an in-app convenience sample, not necessarily representative of users or paying customers.

Grades and run details

Decision model 89 · LLM judge 73
Decision model checks
  • passUses the supplied evidence correctly40%
  • passAddresses the actual decision63%
  • passRespects explicit constraints71%
  • partialIdentifies material uncertainty46%
  • passAvoids unsupported claims38%
  • passProduces the required deliverable74%
  • passCatches the redefined metric99%
  • partialUnpicks the 38%81%
  • partialSees the early erosion36%
  • passJudges fit by retention, not sign-ups86%
  • passGets the base of every number right31%
  • passFinds who it's a must-have for99%
  • passSpots false fit92%
  • passDoesn't scale before fit81%
Run
Run
#1
API response time
33 s
Submitted
5 Oct 2026

Check by check

Where the outputs differ comes first. Open a check to read the judge’s reasons. “Mixed” means the two graders disagreed.

Where they differ 4

Uses the supplied evidence correctlyMixedWrongMixed
GPT-6.1 Sol · API

The claim that the cohort jumps line up with the August 3 definition change is not supported; the jumps occur in the week of July 27–August 2, one week before the change, so the evidence is misused.

Opus 5.5 · Claude

The September churn rate of ~5% relies on a denominator (monthly-recurring base ex-prepay) not provided in the supplied data, so the figure is unsupported.

GPT-6 Luna · API

The memo states the survey is an in-app convenience sample, which is not stated in the brief or supplied context and does not follow from arithmetic.

Avoids unsupported claimsMixedWrongMixed
GPT-6.1 Sol · API

The output presents the jump alignment with the calendar change as fact, but the data shows the jump a week earlier, making it an unsupported claim.

Opus 5.5 · Claude

Presents September churn as about 5% without evidence for the denominator, and the WAU estimate is unsupported speculation.

GPT-6 Luna · API

The claim that the survey is a convenience sample is presented as fact without support from the supplied evidence.

Unpicks the 38%RightRightWrong
GPT-6.1 Sol · API

It breaks down the 38% by tenure (under 2 weeks vs. established) and persona, showing recruiters at 62.5% and engineers at 27.5%.

Opus 5.5 · Claude

Separates users with under 2 weeks (90), gives overall very disappointed for established users (47%), and breaks down by persona (recruiters ~63%, engineers ~24%).

GPT-6 Luna · API

The memo does not break down the 38% by tenure (users under two weeks vs. established) or by persona (recruiters ~62%, engineers ~24%).

Gets the base of every number rightRightWrongRight
GPT-6.1 Sol · API

All derived percentages and growth rates are computed correctly from the supplied data, with bases stated where needed.

Opus 5.5 · Claude

The September churn rate of 5% is stated without a clear denominator or derivation, so the base is not checkable from supplied data.

GPT-6 Luna · API

All derived percentages (sharing rates, MRR figures) are taken correctly from the supplied data.

All got right 10

Addresses the actual decisionRightRightRight
GPT-6.1 Sol · API

The output clearly recommends not hiring the six AEs and allocates funds to the recruiting workflow, with conditions for later scaling.

Opus 5.5 · Claude

Commits early to PMF with recruiters and a recruiting-led use of funds, and says what would change the call.

GPT-6 Luna · API

The memo commits early to a clear call (do not hire six AEs, focus on recruiting) and says what would change it (prove the recruiting motion first).

Respects explicit constraintsRightRightRight
GPT-6.1 Sol · API

The memo is addressed to Sam, stays within the 1,200-word limit, and follows the requested form.

Opus 5.5 · Claude

Memo is within 1,200 words, addressed to Sam, and follows the requested format.

GPT-6 Luna · API

The output is a memo to Sam, under 1,200 words, and addresses what is right/wrong, what to tell investors, and where money should go.

Identifies material uncertaintyRightRightRight
GPT-6.1 Sol · API

It names unknowns (ICP, paid retention, gross margins, sales efficiency, Zoom causation) and states what would change the call (repeatable wins, sustained retention, etc.).

Opus 5.5 · Claude

Names the uncertainty around older users' persona labels, the small survey sample, and September data immaturity.

GPT-6 Luna · API

It names unknowns (September drop, survey representativeness, buyer/sales cycle) and says how to resolve them (investigation, go-to-market test, validate survey counts).

Produces the required deliverableRightRightRight
GPT-6.1 Sol · API

The memo is complete, in the right format for the CEO, and actionable with light edits.

Opus 5.5 · Claude

Delivered as a memo under the word limit, directly usable with light edits.

GPT-6 Luna · API

The memo is in the requested form, for the named reader, within the word limit, and a PM could act on it with light edits.

Catches the redefined metricRightRightRight
GPT-6.1 Sol · API

It catches the August 3 redefinition, attributes the apparent retention improvement to it, and uses the consistent core-action data instead.

Opus 5.5 · Claude

Links the 3 August metric change to the jump in every cohort's curve, explains it inflates retention, and uses the unaffected shared-a-summary data instead.

GPT-6 Luna · API

It links the 3 August changelog to the jump in cohort curves, says the 'improvement' is the definition change, and uses the shared-a-summary data instead.

Sees the early erosionRightRightRight
GPT-6.1 Sol · API

It notes the September drop in week-4 core-action retention, ties it to Zoom's launch, and uses it to advise against the engineering expansion.

Opus 5.5 · Claude

Notes the September drop in week-4 sharing across personas, ties it to Zoom's launch, and uses it to steer the plan away from engineering.

GPT-6 Luna · API

It notes the September drop in week-4 core-action retention, ties it to Zoom's launch by context, and uses it to advise against scaling engineering.

Judges fit by retention, not sign-upsRightRightRight
GPT-6.1 Sol · API

The call on fit rests on the flattening retention curves by segment from the core-action data, with the survey as supporting evidence.

Opus 5.5 · Claude

Bases the PMF call on retention curves flattening for recruiters (shared-a-summary) and not on survey or revenue alone.

GPT-6 Luna · API

The PMF call rests on retention curves (shared-a-summary) by segment, not on sign-ups, revenue growth, or the survey alone.

Finds who it's a must-have forRightRightRight
GPT-6.1 Sol · API

It identifies recruiters as the must-have segment, describes what they value (ATS delivery, scorecards), and builds the plan around them.

Opus 5.5 · Claude

Profiles recruiters as the must-have segment, names what they value (ATS integrations, scorecards), and builds the plan around them.

GPT-6 Luna · API

It names recruiters as the must-have segment, describes what they value (candidate summaries, scorecards, ATS handoff), and builds the plan around them.

Spots false fitRightRightRight
GPT-6.1 Sol · API

It flags the August MRR jump from the promotion, the active-user definition change, and the survey blend as misleading signals of fit.

Opus 5.5 · Claude

Catches the false retention improvement from the metric redefinition, the August MRR spike from the promotion, and the early Zoom erosion.

GPT-6 Luna · API

It spots the active-definition change, the August promotion-driven MRR jump, and the blended survey as false-fit signals and explains why.

Doesn't scale before fitRightRightRight
GPT-6.1 Sol · API

It recommends against scaling sales into enterprise engineering and sets clear evidence thresholds for future expansion.

Opus 5.5 · Claude

Advises against hiring AEs into enterprise engineering without fit, and recommends staged expansion for recruiting only after demonstrating conversion.

GPT-6 Luna · API

It recommends not hiring six AEs, not scaling into engineering, and sets the evidence (proven recruiting motion) that would justify widening later.

Results

Every setup we’ve tested on this task type, across all its tasks and repeats, graded on the current checklist. Calibrated.

#Model · HarnessTask scoreDecision modelLLM judgeRunsCritical failures
1GPT-6 AstraChatGPT90.5100.02None
2Sonnet 5.5API77.990.02None
3Opus 5.5Claude79.782.92None
4Haiku 5.5API77.986.02None
5GPT-6.1 SolAPI90.575.72None
6GPT-6 LunaAPI92.772.42None
7Gemini 3.8 FlashAPI65.062.12None
8Gemini 3.5 Flash-LiteGemini65.151.92None

What you get for the cost

No setup scores within 5 points of the leader at half its cost or less here, so the leader is also the best value.

Task score against relative cost for each setup. The table below has the same figures.50607080901001×3×10×30×100×Cost per output, relative to the cheapest setup (log scale) →↑ Task scoreGPT-6 Astra · ChatGPT: task score 96, 118× the cheapestSonnet 5.5 · API: task score 85, 81× the cheapestOpus 5.5 · Claude: task score 82, 118× the cheapestHaiku 5.5 · API: task score 81, 6.1× the cheapestGPT-6.1 Sol · API: task score 80, 25× the cheapestGPT-6 Luna · API: task score 79, the cheapestGemini 3.8 Flash · API: task score 63, 16× the cheapestGemini 3.5 Flash-Lite · Gemini: task score 56, 2.8× the cheapestGPT-6 Astra · ChatGPTSonnet 5.5 · APIOpus 5.5 · ClaudeHaiku 5.5 · APIGPT-6.1 Sol · APIGPT-6 Luna · APIGemini 3.8 Flash · APIGemini 3.5 Flash-Lite · Gemini
Nobody beats these on both score and cost

Cost is each output’s tokens at the model’s list price, shown relative to the cheapest setup. Runs done in an app have no usage meter, so theirs are estimated from the same setup’s measured runs. How we compare cost

The PM job

Deciding whether you have product-market fit, and for whom.

Why it matters

Scaling before fit amplifies waste. Most teams read fit from sign-ups or a blended survey, and miss the one segment where it's real.

What good looks like

  • Judges fit by retention, not sign-ups
  • Breaks the survey down by segment and tenure
  • Spots revenue and metrics that flatter
  • Scales only where fit is proven

Deliberately not measured

  • Fundraising strategy beyond what the evidence supports
Capability tested

Reading fit from evidence

The failure we’re looking for

Calling fit from a blended number or a growth spike

Grading

Decision model and LLM judge, calibrated against a blind PM review