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

Write a stakeholder update

Can the model turn messy project status into an honest update that leads with what the reader needs to know or decide?

Task type v1.2 · 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. 56% were usable with at most a quick edit.

Reliably right

  1. Reports outcomes and learning, not just output100% pass
    Leads with outcome metrics (ARR, active teams, SLA, pipeline) and what the team learned (pipeline redefinition, invoice ticket surge), not just activities.
    Sonnet 5.5 · API · On track, mostly
  2. Leads with the honest headline100% pass
    First slide headline includes the retention drop alongside the new-ARR beat, not just 'record quarter'.
    GPT-6 Astra · ChatGPT · A quarterly business review deck, in brand
  3. A real 7-slide deck100% pass

Where it slips

  1. Owns the problem55% pass
    Does not state what the team has already done about the payments reliability miss or the NPS drop; only reports the problems.
    GPT-6 Astra · ChatGPT · A quarterly business review deck, in brand
  2. Identifies material uncertainty61% pass
    The output identifies the overlap uncertainty for decision A but does not name any material unknown that could change the call for decision B, such as the reliability of the Payments lead's estimate or the impact of pausing activation experiments.
    Opus 5.5 · Claude · A quarterly business review deck, in brand
  3. Flags the redefined pipeline63% pass
    It names the redefinition but does not give the falling like-for-like trend (e.g., pipeline fell from $2.41M to $2.23M on the old definition).
    GPT-6.1 Sol · API · On track, mostly

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. Asks leaders only for the big calls

    Does the update separate the decisions it needs from its readers (hard to reverse, high stakes) from the ones the team will make itself, and ask clearly for the former?

    Passes when Names the decisions it needs from the readers, with a recommendation and a date, and says which calls the team is making itself.

    Geoff CharlesLenny’s Podcast

  2. Reports outcomes and learning, not just output

    Does the update report progress as movement on the outcomes the team is accountable for and what it has learned, rather than as features shipped or dates hit?

    Passes when Leads with the outcome metrics and what changed in the team's understanding, with shipping as supporting detail.

    Melissa PerriLenny’s PodcastChristian IdiodiLenny’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 · Owns the problem · Ready for the reader as delivered · and 2 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 Head of Product at Cobalt. Write this fortnight's update for the CEO and the exec team, in no more than 450 words, from the weekly metrics and the team notes below. It should say where we stand, what's wrong, what we're doing about it, and what you need from them.

What the model was given3 items: About Cobalt, weekly_metrics_weeks_1_to_8.csv, team_notes.md
About CobaltTeam scheduling software, sold per seat.
weekly_metrics_weeks_1_to_8.csv8 rows · Download
week,new_arr_signed_usd,qualified_pipeline_usd,weekly_active_teams,support_backlog_tickets,support_sla_met_pct
1,182000,2410000,8210,140,96
2,205000,2380000,8260,152,95
3,168000,2350000,8240,171,94
4,214000,2290000,8300,198,91
5,199000,2230000,8280,260,86
6,221000,3140000,8310,344,79
…
team_notes.md7 lines · Download
# Notes from the teams

- Sales ops: from week 6, "qualified pipeline" includes active free trials over 10 seats (they used to be excluded).
- Data: weekly active teams for week 7 is missing because of a pipeline outage; it's being backfilled.
- Support: the backlog grew after the September pricing change (tickets about invoices doubled). SLA target is 90% answered within one business day.
- Quarter target: $2.4M new ARR in 12 weeks ($200,000 a week). Weeks 1 to 8 total: see the file.
- Support hiring: two agents approved, start date not set. Decision needed: whether to pause the planned price change for existing customers in week 10 until the backlog is cleared.
What a strong answer doesThe answer key the graders mark against

Leads with the takeaway: new ARR is ahead of plan ($1.67M in eight weeks against $1.6M), but support is breaking and pipeline is weaker than it looks. Flags that the jump in qualified pipeline in week 6 (from about $2.23M to $3.14M) is the redefinition to include large trials; on the old definition pipeline fell every week from $2.41M to $2.23M, a risk for the rest of the quarter. Reports the support problem plainly: backlog from 140 to 455 tickets and SLA from 96% to 72% against a 90% target since the September pricing change, with what's being done (two agents approved). Notes week 7's active teams are missing and being backfilled rather than hiding it. Asks for the decisions that need the execs: pause the week-10 price change for existing customers until the backlog clears, and set the start date for the support hires. Separates what the team is handling from what needs them.

Critical failures (cap the score at 40)
  • Presents qualified pipeline as having grown about 40%
Task

Staff-level · v1.0

The outputs

8 setups answered this brief.

Sonnet 5.5API98Usable with a quick editGPT-6 AstraChatGPT96Usable with a quick editGPT-6.1 SolAPI92Usable with a quick editHaiku 5.5API87Needs more workOpus 5.5Claude79Needs more workGPT-6 LunaAPI77Usable with a quick editGemini 3.8 FlashAPI63Needs more workGemini 3.5 Flash-LiteGemini58Needs more work
Sonnet 5.5API

Usable with a quick edit12 of 12 checks passed

The output is accurate, complete, and well-structured. A PM could use it with only minor wording tweaks, without re-checking evidence or redoing any reasoning.

Output

Cobalt product update: weeks 7-8

Bottom line: New ARR is on track, but support is deteriorating and our pipeline number is misleading. I need a decision on the week 10 price change.

Where we stand - New ARR: $1.673M signed over 8 weeks, against a pro-rata target of $1.6M. We're $73k ahead. Hitting $2.4M needs about $182k a week over the last four, and we've averaged $209k. - Active teams: 8,190 in week 8, down from a peak of 8,310 in week 6 and roughly flat since week 1. Week 7 is missing because of a data outage and is being backfilled, so I'm not reading a trend into it.

What's wrong 1. Support is failing its SLA. Backlog has risen from 140 to 455 tickets. The SLA hit rate fell from 96% to 72% against a 90% target, and we've been under target since week 5. Support attributes this to the September pricing change, which doubled invoice tickets. Weekly backlog growth slowed in week 8 (+43 versus +68), but it's still growing. 2. Pipeline looks healthier than it is. Reported qualified pipeline jumped from $2.23M to $3.14M in week 6, but that's when Sales Ops started counting active free trials over 10 seats. Before that, pipeline had drifted down for five straight weeks ($2.41M to $2.23M). Until it's restated, I can't tell you whether underlying pipeline is growing, and that matters for next quarter's ARR.

What we're doing - Invoice tickets: We're working with Support to identify the top invoice-confusion causes, then ship clearer invoice explanations and canned responses to cut ticket volume at the source. - Pipeline: I've asked Sales Ops and Data to restate weeks 1-8 on a consistent definition, showing trials separately. I'll bring that next update. - Active teams: Data will backfill week 7, and I'll confirm whether the dip since week 6 is real.

What I need from you 1. Decision: pause the week 10 price change for existing customers until the SLA is back at 90%. The last pricing change drove a ticket surge we haven't absorbed, and a second one on top of a 455-ticket backlog risks churn and more SLA misses. New ARR has a cushion, so I think the delay is affordable. I'd like Finance to confirm the revenue impact of waiting. 2. Set start dates for the two approved support agents. They're approved but have no date, and the backlog is growing now. Two agents may not clear it alone, so I'd also like to discuss temporary cover. 3. Agree that we report pipeline on the restated basis (trials shown separately) from the next update.

Grades and run details

Decision model 96 · LLM judge 100
Decision model checks
  • passUses the supplied evidence correctly37%
  • passAddresses the actual decision64%
  • passRespects explicit constraints57%
  • passIdentifies material uncertainty76%
  • partialAvoids unsupported claims50%
  • passProduces the required deliverable92%
  • passFlags the redefined pipeline100%
  • passReports the support breach plainly100%
  • passOwns the problem68%
  • passReady for the reader as delivered57%
  • passAsks leaders only for the big calls76%
  • passReports outcomes and learning, not just output99%
Run
Run
#1
API response time
17 s
Submitted
6 Oct 2026
GPT-6.1 SolAPI

Usable with a quick edit11 of 12 checks passed

The output is complete and evidence-based; a PM could add the like-for-like pipeline trend in one sentence without re-checking anything else.

Output

Product update — Weeks 7–8

Bottom line: New ARR is ahead of plan, but support is deteriorating and active-team usage is flat to slightly down. I recommend pausing the week-10 price change for existing customers until support recovers.

Where we stand

  • New ARR: $484k signed this fortnight, up 15% from weeks 5–6. Through week 8, we’ve signed $1.673M, $73k ahead of the quarter’s pacing target. We need another $727k over four weeks—$181.75k per week—to reach $2.4M.
  • Qualified pipeline: Reported at $3.35M. This is not comparable with weeks 1–5: from week 6, Sales Ops began including active free trials over 10 seats. We should not interpret the step-up as underlying pipeline growth without a like-for-like view.
  • Usage: Weekly active teams were 8,190 in week 8, versus 8,310 in week 6 and 8,210 in week 1. Week 7 is missing due to a data outage and is being backfilled. There is no clear usage growth signal here; active teams also do not tell us how paid seats are trending.

What’s wrong

Support backlog reached 455 tickets, up 32% from week 6. SLA attainment fell from 79% to 72%, against a 90% target. Invoice tickets doubled after the September pricing change. Another pricing change before recovery risks adding pressure to an already overloaded team.

ARR momentum is encouraging, but it does not establish that customer health is improving.

What we’re doing about it

Our recovery plan is to prioritize invoice-ticket triage, identify recurring billing issues, and fix the largest drivers rather than rely on hiring alone. The two approved support agents need confirmed start dates.

Data is backfilling week 7. We’ll work with Sales Ops on a consistently defined pipeline view, separating free trials, and examine paid-seat trends alongside active teams. We’ll track backlog and SLA weekly to assess recovery.

What I need from you

  1. Approve pausing the week-10 existing-customer price change until the backlog is cleared and SLA attainment is back at or above 90%.
  2. Help secure start dates for the two approved agents, and support temporary coverage if those dates cannot materially aid recovery.
  3. Back prioritizing billing fixes and support recovery over lower-priority roadmap work while we stabilize service.

Grades and run details

Decision model 92 · LLM judge 92
Decision model checks
  • passUses the supplied evidence correctly46%
  • passAddresses the actual decision94%
  • passRespects explicit constraints42%
  • passIdentifies material uncertainty47%
  • partialAvoids unsupported claims29%
  • passProduces the required deliverable90%
  • partialFlags the redefined pipeline95%
  • passReports the support breach plainly100%
  • passOwns the problem38%
  • passReady for the reader as delivered69%
  • passAsks leaders only for the big calls41%
  • passReports outcomes and learning, not just output99%
Run
Run
#1
API response time
23 s
Submitted
6 Oct 2026
Gemini 3.5 Flash-LiteGemini

Needs more than a quick edit6 of 12 checks passed

The ARR total is wrong ($1.73M instead of $1.673M), a factual error that would require re-checking the evidence before use.

Output

To: CEO & Exec Team From: Head of Product, Cobalt Subject: Fortnightly Update (Weeks 7–8)

Here is where we stand for the last two weeks, the critical issues we face, our action plan, and where I need your input.

Where We Stand

We are currently outperforming our top-line targets. We’ve brought in $1.73M in new ARR across 8 weeks against our $1.6M pro-rata goal ($200k/week), closing Week 8 at a strong $248k. Pipeline looks healthy on paper, jumping to $3.35M. However, caveat emptor: Sales Ops adjusted the pipeline definition in Week 6 to include active free trials over 10 seats, masking underlying organic pipeline velocity. Furthermore, weekly active teams dipped to 8,190 in Week 8 (Week 7 data is currently missing due to a data pipeline outage, currently being backfilled).

What’s Wrong

Customer Support is in a critical state. Our support backlog has exploded to 455 tickets, and SLA compliance has plummeted to 72% (target: 90%).

This was triggered by the September pricing change. Invoicing-related tickets have doubled, and our current support bandwidth is completely overwhelmed. While two new hires have been approved, we do not yet have confirmed start dates.

What We’re Doing About It

  1. Stabilizing Support: We are exploring short-term triage options, including engineering-led documentation updates to deflect self-service billing questions.
  2. Data Recovery: Data engineering is actively backfilling the Week 7 active teams metric to restore full visibility.
  3. Pipeline Audit: Sales Ops is recalibrating the pipeline reporting to separate true enterprise/mid-market opportunities from the newly included free trials so we can accurately forecast Q3 close rates.

What I Need From You

We face an immediate strategic decision regarding Week 10:

  • Decision Needed: Should we pause the planned price change for existing customers scheduled for Week 10 until the current support backlog is cleared and SLAs are back above 90%?

Proceeding with the existing customer rollout while our SLA is at 72% risks a wave of churn and avoidable escalation tickets. I recommend pausing, but I need a final call from the exec team so Product and Support can communicate the adjustment internally and externally.

Claims the judge couldn’t find in the brief

  • We’ve brought in $1.73M in new ARR across 8 weeks

Grades and run details

Decision model 75 · LLM judge 46
Decision model checks
  • partialUses the supplied evidence correctly21%
  • partialAddresses the actual decision49%
  • passRespects explicit constraints24%
  • partialIdentifies material uncertainty68%
  • partialAvoids unsupported claims52%
  • passProduces the required deliverable51%
  • partialFlags the redefined pipeline97%
  • passReports the support breach plainly100%
  • partialOwns the problem52%
  • passReady for the reader as delivered37%
  • passAsks leaders only for the big calls54%
  • passReports outcomes and learning, not just output92%
Run
Run
#1
Time to output
3 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 6

Uses the supplied evidence correctlyRightRightWrong
Sonnet 5.5 · API

All factual claims about the current situation are directly supported by the supplied metrics or team notes, or follow from simple arithmetic.

GPT-6.1 Sol · API

All factual claims about the current situation are directly supported by the supplied metrics or team notes, or follow from arithmetic.

Gemini 3.5 Flash-Lite · Gemini

The total new ARR is stated as $1.73M, but the sum of weeks 1-8 is $1.673M; this factual error means the output does not use the supplied evidence correctly.

Addresses the actual decisionRightRightWrong
Sonnet 5.5 · API

Commits unambiguously to pausing the week 10 price change, frames it for the exec team, and implies the call could change if Finance confirms the revenue impact is unaffordable.

GPT-6.1 Sol · API

Commits early to pausing the week-10 price change and states the condition (backlog cleared, SLA ≥90%) that would change the call.

Gemini 3.5 Flash-Lite · Gemini

The output recommends pausing the price change but does not state what result or condition would change that recommendation, as required.

Identifies material uncertaintyRightRightWrong
Sonnet 5.5 · API

Identifies the pipeline restatement, active teams backfill, revenue impact of delay, and sufficiency of two agents as unknowns, and says how each will be resolved.

GPT-6.1 Sol · API

Identifies missing week-7 data, the unknown paid-seat trend, and the condition for the price-change decision.

Gemini 3.5 Flash-Lite · Gemini

It does not name the specific unknowns that could change the decision about the price change, nor say how they would be resolved or what would change the call.

Avoids unsupported claimsRightRightWrong
Sonnet 5.5 · API

Causes and forecasts are clearly labelled as such (e.g., 'risks churn', 'I think the delay is affordable'), and confident claims are backed by the evidence.

GPT-6.1 Sol · API

Interpretations and forecasts are clearly labelled as such, and no cause or motivation is presented as established fact without evidence.

Gemini 3.5 Flash-Lite · Gemini

The $1.73M ARR figure is presented as fact but is not supported by the data; it is a confident claim that the evidence does not establish.

Produces the required deliverableRightRightMixed
Sonnet 5.5 · API

The update is complete, in the right form for the CEO and exec team, within the word limit, and the reader could act on the asks without filling major gaps.

GPT-6.1 Sol · API

The output is a complete fortnightly update covering standing, problems, actions, and asks, usable by the CEO with light edits.

Gemini 3.5 Flash-Lite · Gemini

The update contains a material factual error in the ARR total, so the reader cannot act on it without correcting that number.

Flags the redefined pipelineRightWrongWrong
Sonnet 5.5 · API

Explicitly names the redefinition and gives the like-for-like trend of five straight weekly declines from $2.41M to $2.23M.

GPT-6.1 Sol · API

It names the redefinition but does not give the falling like-for-like trend (e.g., pipeline fell from $2.41M to $2.23M on the old definition).

Gemini 3.5 Flash-Lite · Gemini

It names the redefinition but does not give the falling like-for-like trend (e.g., pipeline on the old definition fell from $2.41M to $2.23M).

All got right 6

Respects explicit constraintsRightRightRight
Sonnet 5.5 · API

The update is under 450 words, written as a fortnightly update for the CEO and exec team, and respects the requested form.

GPT-6.1 Sol · API

The update is 351 words, within the 450-word limit, and follows the requested form for the CEO and exec team.

Gemini 3.5 Flash-Lite · Gemini

The update is under 450 words and addressed to the CEO and exec team as requested.

Reports the support breach plainlyRightRightRight
Sonnet 5.5 · API

Reports the backlog (140 to 455) and SLA (96% to 72% against 90% target), links them to the pricing change, and asks for the pause decision.

GPT-6.1 Sol · API

Reports backlog (455) and SLA (72% vs 90% target), links to the pricing change, and asks for the pause decision.

Gemini 3.5 Flash-Lite · Gemini

It reports the backlog (455) and SLA (72% vs 90% target), links them to the pricing change, and asks for the pause decision.

Owns the problemRightRightRight
Sonnet 5.5 · API

For each problem, states actions already taken (working with Support, asking Sales Ops and Data to restate) and recommends a way forward with trade-offs.

GPT-6.1 Sol · API

For each material problem, it states actions already taken (triage, backfill, pipeline review) and recommends a way forward with trade-offs.

Gemini 3.5 Flash-Lite · Gemini

For support it notes the approved hires and recommends pausing the price change; for data and pipeline it states the actions being taken.

Ready for the reader as deliveredRightRightRight
Sonnet 5.5 · API

Written entirely for the exec team, uses only owners and terms from the context, and contains no placeholders or notes to the requester.

GPT-6.1 Sol · API

Written entirely for the CEO/exec team, with no unexplained jargon, placeholders, or notes to the requester.

Gemini 3.5 Flash-Lite · Gemini

The update is written for the exec team, uses no unexplained jargon from the brief, and all owners/dates are from the material or clearly proposed.

Asks leaders only for the big callsRightRightRight
Sonnet 5.5 · API

Separates the big calls for leaders (pause price change, set start dates, agree reporting) from what the team is handling itself (invoice fixes, pipeline restatement, backfill).

GPT-6.1 Sol · API

Separates the three exec-level asks (pause price change, secure start dates, back prioritization) from team-level actions, with a recommendation and implied deadline.

Gemini 3.5 Flash-Lite · Gemini

It asks for the pause decision with a recommendation and a date, and separates it from the operational actions the team is taking.

Reports outcomes and learning, not just outputRightRightRight
Sonnet 5.5 · API

Leads with outcome metrics (ARR, active teams, SLA, pipeline) and what the team learned (pipeline redefinition, invoice ticket surge), not just activities.

GPT-6.1 Sol · API

Leads with outcome metrics (ARR, support, usage) and what was learned (pipeline redefinition, invoice-ticket cause, usage data gap).

Gemini 3.5 Flash-Lite · Gemini

It leads with outcome metrics (ARR, pipeline, support) and explains what changed (support broke after pricing change), showing learning.

Results

Every setup we’ve tested on this task type, across all its tasks and repeats, graded on the current checklist. Calibrated. Where our PM checked the graders by hand, they agreed on 76% of 89 checks.

#Model · HarnessTask scoreDecision modelLLM judgeRunsCritical failures
1GPT-6 AstraChatGPT94.292.92None
2Sonnet 5.5API84.696.42None
3Haiku 5.5API89.692.32None
4GPT-6.1 SolAPI80.896.22None
5Opus 5.5Claude86.377.52None
6GPT-6 LunaAPI76.766.82None
7Gemini 3.5 Flash-LiteGemini70.851.62None
8Gemini 3.8 FlashAPI66.351.92None

What you get for the cost

Haiku 5.5 · API scores within 5 points of GPT-6 Astra · ChatGPT at a tenth of the cost.

Task score against relative cost for each setup. The table below has the same figures.50607080901001×3×10×30×Cost per output, relative to the cheapest setup (log scale) →↑ Task scoreGPT-6 Astra · ChatGPT: task score 94, 26× the cheapest (1 of 2 outputs estimated)Sonnet 5.5 · API: task score 90, 40× the cheapestHaiku 5.5 · API: task score 89, 2.6× the cheapestGPT-6.1 Sol · API: task score 88, 15× the cheapestOpus 5.5 · Claude: task score 81, 55× the cheapest (1 of 2 outputs estimated)GPT-6 Luna · API: task score 71, the cheapestGemini 3.5 Flash-Lite · Gemini: task score 60, the cheapest (1 of 2 outputs estimated)Gemini 3.8 Flash · API: task score 58, 13× the cheapestGPT-6 Astra · ChatGPTSonnet 5.5 · APIHaiku 5.5 · APIGPT-6.1 Sol · APIOpus 5.5 · ClaudeGPT-6 Luna · APIGemini 3.5 Flash-Lite · GeminiGemini 3.8 Flash · API
Best valueNobody beats these on both score and costCost partly estimated (app runs with no usage meter)

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

Writing the weekly update, or the reply to an executive's question, that the team's credibility rests on.

Why it matters

Updates are where bad news gets softened. A fluent update that buries a slipped date or presents a guess as the cause does more harm than no update: leadership decides on it, and trust goes when the truth comes out.

What good looks like

  • Leads with the news and any decision needed
  • States dates, numbers and causes plainly, including bad news
  • Separates what is known from what is suspected
  • Brings a recommendation and what's been tried, not just the problem
  • Fits the reader and the length asked for

Deliberately not measured

  • Tone and formatting preferences
  • Writing style beyond clarity
Capability tested

Honest, decision-first status communication

The failure we’re looking for

Spin: burying or softening the bad news, or presenting a guess as fact

Grading

Decision model, LLM judge and a browser check (for the deck), calibrated against a blind PM review

Variants

A short written update · Staff level: a QBR deck in brand colours