Assess cofounder values: four conversations before a working trial

To assess a potential cofounder’s values, choose decisions you might actually face, write your own response, then compare the reasons and boundaries behind each answer. The immediate outcome is a clearer decision about whether to explore working together—not a verdict about someone’s character.

This guide is for the conversation stage. If you already have enough agreement to try a small project, use the separate cofounder working-trial guide. Keep introductions, a working trial and formal company commitments as different decisions.

What the source supports—and what this guide adds

Y Combinator’s questionnaire, published 27 April 2023, recommends answering independently before comparing responses. It covers goals, roles, location, commitment, finances and disagreement, and cautions that a questionnaire cannot replace time working together. This is practitioner guidance, not a validated predictor of compatibility.

The four conversations, examples and record below are MLAI editorial exercises. They are not YC’s questionnaire reproduced or a research-backed ranking of people. Use them to make a particular working decision clearer.

Separate a value, a preference and a constraint

Do not treat a caring responsibility, a need for paid work or a different communication style as evidence of weak values. An arrangement may be impractical without either person being untrustworthy. Ask whether a difference can be accommodated, needs more information or prevents this particular partnership.

Fictional worked example: Alex and Morgan, their answers, permission checks, work, time log and decisions are invented. These are not interviews, observed founder outcomes or a validated compatibility assessment. Their four conversations below lead to one limited learning exercise, not a decision to form a company.

1. Ambition and the first commitment

Discuss what you each want from this project and the smallest commitment you are considering now. A weekend experiment, a part-time business and a venture seeking rapid growth are not interchangeable. Neither a shared product idea nor a shared label such as “startup founder” settles the question.

• What useful outcome would make the next experiment worth doing, even if we do not start a company?
• Which kinds of customer problem do we want to work on, and which would we decline?
• What evidence would make either of us change direction or stop?

Write down the differences. If one person wants a learning project and the other expects immediate full-time commitment, resolve that before assigning work.

2. Capacity, cost and expectations

Exchange the constraints that affect the proposed work: available hours, response windows, spending limits and whether paid work is required. Each person chooses how much personal context to share. Do not use this worksheet to demand financial records, private family details or evidence of willingness to sacrifice wellbeing.

A useful output is specific: “We can each spend up to six hours on the next exercise; neither can promise daytime availability.” That is an example, not a recommended workload. If the task needs more capacity, reduce its scope, find another arrangement or pause. Do not silently assume the other person will absorb the gap.

3. Evidence, customer data and AI boundaries

For an AI startup, discuss how you will handle a confident but unsupported output, a request to upload private data, or pressure to describe a prototype as production-ready. Ask what each person would do, who may authorise an action and what evidence would be needed first.

This records a specific decision, not proof of either person’s overall values or the legality of a future data use. Whether the agreed boundary is followed in practice remains an observation to make, not an assumption to score as passed.

4. Disagreement, feedback and stopping

Use a low-stakes scenario: the demo has a known error and the deadline is close. Would you reduce scope, delay or show the limitation clearly? Compare the reasons, the people affected and who would decide. Do not manufacture a crisis, conceal information or provoke someone to “reveal their true self.”

Agree how someone can raise a concern, request a pause and correct the written record. Ask for a past example only if they can share it without exposing another person’s private information. A single anecdote is context, not verification. Any later reference conversations should be transparent and consent-based.

Values discussion record

Complete your answers separately, then let each person check how their position is recorded. Use one record per important decision; there is no aggregate compatibility score.

Download the editable values discussion record (.txt). Includes all four examples, the completed record and blank fields; nothing is submitted to MLAI.

Decision or working situation:
My preference and why:
Their preference in their own words:
Concrete example each person offered:
What we agree on:
What remains different or unknown:
Boundary each person needs respected:
Next conversation or observation:
Decision date, including permission to stop:

Keep the record private to the agreed participants. Do not upload identifiable notes to an AI tool without appropriate permission. The record is a conversation aid, not a legal agreement, psychological assessment or confirmation that a partnership will work.

Choose the next step without forcing a yes

• Explore a bounded trial: both people understand its purpose and agree to its boundaries. That authorises only the agreed exercise.
• Pause: a consequential question is unresolved, or the proposed commitment does not fit someone’s capacity.
• Stop: either person does not want to continue. Document the practical loose ends without pressuring them to justify private circumstances.

Equity, pay, ownership and company responsibilities need their own appropriate advice and agreements. Do not infer them from a positive conversation or from completing this worksheet.

Take one process question to a relevant event

You might ask a peer, “How did you record a disagreement about demo quality before starting a project?” Ask about their process, not for a verdict on a named potential cofounder. Check the event’s topic and online or in-person format, and keep the other person’s circumstances confidential.

MLAI publishes this guide and runs events; this is an affiliated invitation, not an assessment or matching service.

Scope and update

YC source rechecked 11 September 2026. This revision completes the four conversations with supplied answers, follow-up questions and a filled record. The retained byline is not evidence that the fictional conversation occurred or that an independent professional reviewed this worksheet. No MLAI matching service, independent assessment or successful partnership is implied.

Frequently Asked Questions

About the Author

Dr Sam Donegan

Medical Doctor, AI Startup Founder & Lead Editor

Sam leads the MLAI editorial team, combining deep research in machine learning with practical guidance for Australian teams adopting AI responsibly.