Tasks / Challenge

1000x an idea

Can the model expand an idea's ambition while keeping it tethered to a real mechanism?

Measures the systemTask v1.0 · 2 casesDifficulty

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

From 14 graded outputs by 7 models. 36% were usable with at most a quick edit.

Reliably right

  1. Extreme, then back to buildable96% pass
    It pushes the worker dimension to an extreme portable earnings network, then works back to a concrete first step that tests the same mechanism.
    GPT-6 Astra · ChatGPT · From tip calculator to worker network
  2. Finds the mechanism the data hides93% pass
    It centers the mechanism on workers with second jobs pulling new restaurants onto Tally, using the 41% second-job share and the 57 faster, cheaper worker-led signups.
    GPT-6 Astra · ChatGPT · From tip calculator to worker network
  3. Addresses the actual decision91% pass
    It commits early to the worker-led distribution thesis over the AI operating system and states the result that would stop the bet.
    GPT-6 Astra · ChatGPT · From tip calculator to worker network

Where it slips

  1. Proposes tests that could fail39% pass
    Several gates lack explicit measurement windows or clear actions for every outcome, such as the 25% lift gate and the 10% monthly adoption gate.
    GPT-6.1 Sol · API · From tip calculator to worker network
  2. Avoids unsupported claims63% pass
    It presents several causal, competitive, and data-structure claims as established facts without support in the pack.
    Sonnet 5.5 · API · From tip calculator to worker network
  3. Respects explicit constraints66% pass
    It violates the length constraint and proposes a stop threshold whose arithmetic is internally inconsistent.
    Sonnet 5.5 · API · From tip calculator to worker network

Case viewer

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

The brief

You're a PM at Pantry. Aisha, one of our PMs, has proposed letting people share a shopping list with one partner. Before we commit next quarter to it, our CEO, Lena Brandt, wants to see the 1000x version: how big could this idea get? Write a memo for Lena and Aisha of no more than 700 words that takes the idea to its most ambitious version and then back to what we should build first, and how we'll know it's working. What we know is below.

What the model was given5 items: About Pantry, Aisha's proposal, What we already see, What hasn't worked, Constraint
About PantryA shopping-list app. 2.4 million monthly active users, free, with a $4.99-a-month Plus plan that 3% of users pay for. The team for this work: four engineers and a designer, for one quarter.
Aisha's proposal“Let a user invite one partner to a list. Both can add and tick off items.”
What we already seeSome users already share their login with a partner to get the same effect. Lists edited from two or more devices on different phone numbers: 11% of lists. Users whose list is edited from two or more devices keep using Pantry at month 3 at 71%, against 22% for everyone else. On those lists, 38% of items are added by the second device. The most common support request is 'we both bought the milk': 1,900 tickets last quarter about duplicate purchases.
What hasn't workedRecipes and meal plans launched a year ago; 4% of users have opened them in the last month. Lena's note: “Whatever we do, it probably needs AI in it.”
ConstraintLegal: we can't track anyone's location without an explicit opt-in from that person, and we can't show one person's location to another without the second person's opt-in too.
What a strong answer doesThe answer key the graders mark against

Finds the mechanism in the data: the value is several people coordinating one household's shopping (71% against 22% month-3 retention when a list is edited from two devices, 38% of items added by the second person, 1,900 duplicate-purchase tickets), not the list itself and not recipes (4% use them). Pushes that one dimension to the extreme, such as the list as the household's shared memory where everyone's needs flow in, nobody double-buys and each errand goes to whoever is best placed (with location only on opt-in from both people), rather than adding AI features for their own sake. Works back to a first step four engineers can ship in a quarter that tests the coordination mechanism: shared lists for the whole household (not just one partner) with a live 'I'm at the shop' claim so items aren't bought twice. Sets a threshold, such as the share of lists with a second active editor and duplicate-purchase tickets per shared list, and says what result would stop the bigger bet. Uses AI only where it serves coordination, if at all.

Critical failures (cap the score at 40)
  • Builds the ambitious version on recipes, meal plans or AI for one person, dropping household coordination
  • Proposes tracking or sharing a person's location without their explicit opt-in
Case

Mid-level · v1.1 · synthetic · B2C, consumer app, retention, ambition

The outputs

7 setups answered this brief. Pick up to three to read side by side.

Usable with a quick edit9 of 10 checks passed

A PM could use this with light edits, mainly tightening the measurement windows and decision rules. The evidence, mechanism, first step, and legal constraint are already sound.

Output

To: Lena Brandt, Aisha Subject: From a shared list to the system that runs household replenishment

Recommendation

Commit next quarter to testing household coordination, not merely adding a second editor. Aisha’s proposal is the smallest entry point into a much larger opportunity: Pantry becomes the trusted system that ensures a household has what it needs, without anyone having to manage every purchase.

The 1000x version

Today, Pantry remembers what someone typed. At its most ambitious, Pantry coordinates the entire replenishment cycle: noticing needs, resolving preferences, assigning responsibility, finding the right purchase, and confirming arrival.

Imagine: “Keep our household stocked within $150 a week.” Pantry learns recurring needs from explicitly connected purchase history and household corrections, drafts a replenishment plan, reconciles conflicting requests, and coordinates whoever—or whichever service—will fulfill it. Eventually, users could authorize bounded automatic purchases, with spending limits, explanations, and easy overrides.

The expansion is from one person’s checklist to a household’s purchasing agent, then a demand network connecting households to retailers and delivery providers. Core sharing could remain free; trusted automation could support Plus, with optional transaction revenue later. Recommendations must serve household preferences, not whichever retailer pays most.

AI belongs in interpreting messy requests, recognizing equivalent items, and proposing replenishment—not in deciding whether two phones agree. Synchronization must be reliable software. This is an ambition, not a forecast or next-quarter roadmap.

Why this is the right starting point

Users are already working around our product: 11% of lists are edited from multiple devices with different phone numbers. Second devices contribute 38% of items on those lists. This looks like genuine joint work, not occasional viewing.

Those users retain at month three at 71%, versus 22% for others. That is compelling but not causal: committed households may naturally retain better, and devices are not verified partners.

The clearest unmet need is operational: 1,900 duplicate-purchase tickets last quarter. Meanwhile, only 4% opened recipes or meal plans last month. We should earn the right to expand by preventing “we both bought the milk,” not by launching another planning destination.

What we build first

With four engineers and one designer, ship a narrow, free, two-person shared-list pilot:

  • Separate accounts; a simple invitation, acceptance, and removal flow.
  • Fast shared additions and check-offs, with clear sync status and safe recovery after disconnection.
  • An explicit, visible “I’m buying this” claim, followed by “bought,” so coordination happens before checkout. Surface conflicting claims; never imply an offline claim is confirmed.
  • Clear attribution and controls to leave sharing without exposing other private lists.

Test with existing login-sharing households first. Do not require location: voluntary item claims solve the immediate problem with less risk. Any future location feature requires explicit opt-in from both the person being tracked and the intended recipient, plus revocation controls.

For Lena’s AI question: time-box a prototype that flags likely duplicate entries such as “milk” and “whole milk.” Test precision with users; never silently merge. It ships only if useful and does not delay the coordination foundation. No ordering, retailer integrations, inventory prediction, or meal-planning expansion this quarter.

How we know it works

Instrument the funnel from invitation through acceptance to both people contributing and completing a shopping cycle. Weekly active collaborating households is our adoption metric—not invitations sent.

Randomize eligible initiating accounts to invitation access or control, keeping linked partners in the same experiment. Compare outcomes by assignment, not just among successful sharers.

Before launch, power the test around these proposed decision thresholds:

  • At least a 20% relative reduction in self-reported duplicate purchases per shopping trip.
  • A five-percentage-point improvement in day-28 retention, followed by month-three measurement.

Track sync failures, lost edits, accidental sharing, notification complaints, and Plus cancellations as guardrails. Support tickets are corroborating evidence, not the sole outcome.

Expand only if households repeatedly coordinate and randomized outcomes improve without material trust failures. If invitations succeed but duplicate buying persists, fix coordination before broadening the vision.

Claims the judge couldn’t find in the brief

  • The clearest unmet need is operational.
  • Support tickets are corroborating evidence, not the sole outcome.

Grades and run details

Decision model 95 · LLM judge 10 of 11 checks
Decision model checks
  • passUses the supplied evidence correctly45%
  • passAddresses the actual decision98%
  • passRespects explicit constraints31%
  • passIdentifies material uncertainty99%
  • passAvoids unsupported claims62%
  • passProduces the required deliverable44%
  • passFinds the mechanism in the data91%
  • passA first step that tests the big idea91%
  • passExtreme, then back to buildable97%
  • partialProposes tests that could fail69%
Run
Run
#1
Time to output
37 s
Submitted
1 Oct 2026

Usable with a quick edit9 of 10 checks passed

The memo is well-structured and evidence-based; adding numeric thresholds (e.g., a minimum retention lift and a target reduction in duplicate tickets) is a quick edit that does not require re-checking evidence or redoing reasoning.

Output

To: Lena Brandt, Aisha Re: Shared lists: the 1000x version, and what to build first

The short version

Aisha's proposal is right, and it's the first step toward something much bigger. Our data suggests Pantry's real unit isn't the person but the household. Next quarter we should build partner sharing around the moment that actually breaks today: two people at two stores, both buying the milk. Then we should use the result to test whether sharing causes the retention gap we see.

What the data tells us

  • Demand already exists. 11% of lists are edited from two or more devices on different phone numbers, and some of that is people sharing a login to get around a missing feature.
  • The second person contributes, not just watches. They add 38% of items on those lists.
  • The pain is at purchase time, not list-building time. 1,900 tickets last quarter were about duplicate purchases. That's our top support request.
  • Retention is striking, but we can't yet say sharing causes it. Shared-list users are at 71% at month 3, against 22% for everyone else. Committed, organised households may simply be more likely to share and to stick. We should design the launch to find out.

The 1000x version

Pantry becomes the operating system for a household's food. Four ideas stack up:

  1. Every member, not one partner. Families, roommates, and adult children shopping for elderly parents all share one live list, with roles.
  2. The list knows the household's rhythm. Pantry learns that you buy milk every five days and nobody has added it. It notices that eggs appear on both partners' mental lists and merges them. This is where AI earns its place: predicting and reconciling what the household needs, not generating content.
  3. Coordination in the moment. "Sam is shopping now, add anything." Items are claimed as they go in the cart. Later, with explicit opt-in from both people, the app can say "Sam is near the grocery store."
  4. The household becomes the business model and the growth engine. Every invite is an acquisition channel. Plus becomes a household plan, which is an easier purchase than a personal one. At scale, Pantry holds the best dataset anywhere on what households actually consume, which retailers and delivery services would pay to plug into.

A warning from recipes and meal plans, which only 4% of users opened last month: ambition should deepen the core job, which is the list, not bolt on adjacent ones. Everything above makes the list itself smarter and more shared.

What we build first (one quarter, four engineers and a designer)

  1. Invite one partner to a list, as Aisha proposed. Build the data model for more members so we don't rebuild later. Offer an easy migration for people currently sharing a login.
  2. Real-time sync with attribution, so you can see who added and who ticked off each item.
  3. "I'm shopping" mode. This is a manual toggle, so it needs no location data and raises no legal issue. It notifies the partner, and items ticked off in-store update live so the other person can't buy them too. This targets the 1,900 duplicate-purchase tickets directly.
  4. One small AI feature: duplicate detection. It flags "milk" and "2% milk" added by different people and offers to merge them. It's cheap, directly useful, and meets Lena's AI bar without building a speculative product.

Deliberately out of scope for now: location features, predictive restocking, more than two members, and household pricing. Each depends on proving the basics first.

How we'll know it's working

  • Adoption: the share of active lists with a second member, measured against today's 11% baseline (which includes login-sharing). Also track the invite acceptance rate.
  • The pain goes away: duplicate-purchase tickets per 1,000 shared lists should fall sharply from the current 1,900 a quarter.
  • Causation, the metric that matters most: randomly hold back the invite prompt from a slice of eligible users. Then compare month-3 retention for those offered sharing against the holdout. If invited users retain meaningfully better, sharing is our biggest retention lever, and the 1000x roadmap is justified. If not, the 71% was mostly selection. We'll still have fixed our top support issue, and we'll invest elsewhere.
  • Growth: new sign-ups arriving through invites.

Decision needed

Approve the one-quarter scope above, including the holdout test. We'll return with retention results at the end of the quarter plus 90 days, and a recommendation on whether to pursue the household roadmap.

Grades and run details

Decision model 95 · LLM judge 10 of 11 checks
Decision model checks
  • passUses the supplied evidence correctly60%
  • passAddresses the actual decision98%
  • passRespects explicit constraints37%
  • passIdentifies material uncertainty100%
  • passAvoids unsupported claims43%
  • passProduces the required deliverable31%
  • passFinds the mechanism in the data100%
  • passA first step that tests the big idea97%
  • passExtreme, then back to buildable100%
  • partialProposes tests that could fail67%
Run
Run
#1
Time to output
35 s
Submitted
30 Sept 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.

All got wrong 1

Proposes tests that could failWrongWrong
GPT-6 Astra · ChatGPT

The proposed thresholds lack a clear measurement window and explicit stop/continue actions for each outcome, so the tests are not fully killable as specified.

Opus 5.5 · Claude

The retention test lacks a numeric threshold for 'meaningfully better', and the duplicate-purchase metric only says 'fall sharply' without a specific number, so the tests do not have the required numeric thresholds.

All got right 9

Uses the supplied evidence correctlyRightRight
GPT-6 Astra · ChatGPT

The memo uses the supplied figures correctly and does not invent current facts; its few interpretive statements are either labelled as interpretations or follow from the supplied data.

Opus 5.5 · Claude

All claims about the current situation are taken directly from the brief and supplied context, with no invented numbers or facts.

Addresses the actual decisionRightRight
GPT-6 Astra · ChatGPT

It commits early to testing household coordination rather than merely adding a second editor and states the randomized results that would justify or stop expansion.

Opus 5.5 · Claude

The memo commits to a clear call (approve the one-quarter scope with holdout test) and says what would change it (if the retention test shows no lift, invest elsewhere).

Respects explicit constraintsRightRight
GPT-6 Astra · ChatGPT

It is addressed to Lena and Aisha, stays under 700 words, respects the team/quarter constraint, and enforces the legal location constraint by requiring explicit opt-in from both people and revocation controls.

Opus 5.5 · Claude

The memo is under 700 words, addresses Lena and Aisha, and respects the location opt-in constraint by using a manual toggle and requiring explicit opt-in from both people for any future location feature.

Identifies material uncertaintyRightRight
GPT-6 Astra · ChatGPT

It names material unknowns, including causality, device identity, duplicate-purchase reduction, retention, sync failures, and trust issues, and proposes randomized measurement to resolve them.

Opus 5.5 · Claude

It names the key uncertainty (whether sharing causes the retention gap or is selection) and proposes a randomized holdout test to resolve it, with clear actions for each outcome.

Avoids unsupported claimsRightRight
GPT-6 Astra · ChatGPT

It labels the 1000x version as ambition, not forecast, and treats causal explanations and interpretations as hypotheses rather than established facts.

Opus 5.5 · Claude

Interpretations like 'the pain is at purchase time' are grounded in the 1,900 tickets, and the retention gap is explicitly labelled as not yet proven causal.

Produces the required deliverableRightRight
GPT-6 Astra · ChatGPT

It is a usable memo for the named readers that moves from the extreme version to a first build and measurement plan within the requested length.

Opus 5.5 · Claude

The memo is a complete, actionable document for the CEO and PM, within the word limit, covering the 1000x version, first step, and success metrics.

Finds the mechanism in the dataRightRight
GPT-6 Astra · ChatGPT

It centres household coordination and backs it with the 71% versus 22% retention gap, 38% second-device item contribution, and 1,900 duplicate-purchase tickets.

Opus 5.5 · Claude

The memo identifies household coordination as the core mechanism, backs it with the retention gap, second-device contribution, and duplicate-purchase tickets, and keeps it central to both the extreme version and the first step.

A first step that tests the big ideaRightRight
GPT-6 Astra · ChatGPT

The first step is a scoped two-person shared-list pilot with claims and sync controls that four engineers and a designer could ship in a quarter and that directly tests coordination.

Opus 5.5 · Claude

The first step (partner invite, real-time sync, 'I'm shopping' mode, duplicate detection) fits the team and quarter, directly tests coordination, and includes a holdout test whose result would stop or continue the bigger bet.

Extreme, then back to buildableRightRight
GPT-6 Astra · ChatGPT

It pushes the idea along the household-coordination dimension to a purchasing-agent/demand-network extreme, then returns to a concrete first step preserving that mechanism.

Opus 5.5 · Claude

It pushes the dimension of household coordination to an extreme (household OS, every member, predictive rhythm, location coordination) and works back to a concrete, buildable first step that tests the same mechanism.

Results

Every setup we’ve tested on this task, across all cases and repeats, graded on the current checklist. Calibrated: the graders match our PM on 89% of checks.

#Model · HarnessTask scoreDecision modelLLM judgeRunsCritical failures
1GPT-6 AstrawithChatGPT95.295.52None
2GPT-6.1 SolwithAPI92.786.72None
3GPT-6 LunawithAPI88.278.82None
4Opus 5.5withClaude90.774.62None
5Sonnet 5.5withAPI90.265.92None
6Gemini 3.8 FlashwithAPI71.134.521 capped
7Gemini 3.5 Flash-LitewithGemini61.417.02None

About the task

The PM job

Finding the bigger version of a good idea.

Why it matters

Ambition without mechanism is fan fiction. The useful version pushes to the extreme, then works back to something buildable.

What good looks like

  • Names the mechanism that scales
  • Keeps the core insight
  • Works back to a first step you could build

Deliberately not measured

    Capability tested

    Ambitious expansion

    The failure we’re looking for

    Bigger adjectives, same idea

    Grading

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

    This task measures the whole setup. Tools, instructions and skills in the harness do real work here, so read the harness as carefully as the model name.