We turn assumptions into evidence, and evidence into product decisions.
How we decide what to do, how we deliver it, and the ways you can work with us, from a consultation to a dedicated team.
The proof loop
Every engagement moves through the same five steps, whatever the function involved.
Define the decision
What choice is on the table?
Map the assumptions
What must be true?
Choose the proof
The cheapest credible evidence.
Build or run the test
Only what creates the evidence.
Decide what changes
Build, narrow, reposition or stop.
Discovery Before Larger Work
Before larger work, we often start with discovery.
Discovery is a short paid phase before a larger build, program, retainer, or partnership. It helps us understand the decision behind the request, the risks that matter most, the constraints around the work, and the kind of proof needed before execution begins.
It is not an open-ended strategy deck. It is a way to turn a broad request like "build an MVP", "help with GTM", "launch an AI workflow", or "enter a new market" into a clear scope, timeline, team shape, and recommended path.
Discovery answers how to approach the work. Validation answers whether there is enough market, workflow, buyer, or product proof to justify the next commitment.
Discovery usually takes a few working days to a few weeks depending on the complexity of the product, GTM motion, technical system, stakeholders, and decision at stake. Not every engagement needs a separate discovery phase. Validation Sprints and Demand Validation Sprints usually include enough discovery inside the sprint itself. For larger builds, programs, retainers, and partnerships, discovery protects both sides from starting with the wrong scope. Discovery can also be useful as a standalone diagnostic when the next step is unclear.
Not all signal is proof.
Different stages need different standards of evidence. A conversation can be useful, but it is not the same as a paid commitment. A prototype demo can reveal usability, but it is not the same as workflow adoption. A pilot can create learning, but only if it has success criteria and a conversion path.
- Weak signal: compliments, vague interest, broad market size, passive survey responses.
- Useful signal: specific problem stories, active pull, repeated use cases, buyer urgency.
- Strong proof: paid commitments, signed pilots, repeated usage, conversion, retention, operational adoption, or budget allocation.
For how this plays out in real engagements, see our case studies.
Build the smallest thing that can create the right evidence.
Sometimes that is a landing page, offer narrative, customer interview sequence, or concierge workflow. Sometimes it is a real MVP, internal tool, AI workflow, dashboard, or integration. The point is not to avoid building. The point is to build when the artifact can create better evidence than conversation alone.
- Do not build complexity to hide uncertainty.
- Do not over-validate when users need something tangible to react to.
- Do not run GTM tests that the product cannot support.
- Do not build products without knowing what evidence they are supposed to generate.
GTM is part of the product system.
We do not treat marketing as a detached campaign layer. GTM work changes what should be built, how the product should be packaged, which buyer matters, and what proof the team needs. Product work changes what can be sold, onboarded, measured, and scaled.
- If development is requested, we ask who it is for and what validates the need.
- If GTM is requested, we may build or change product surfaces that block conversion.
- If AI is requested, we map workflow adoption before building the tool.
- If market entry is requested, we test the buyer path before recommending scale.
This is what the full range of capabilities is for.
Stage-specific proof
The methodology adapts to maturity. The proof question changes with the stage, and so do the engagements that answer it.
| Stage | Main proof question | Common engagements |
|---|---|---|
| Operator-led idea | Is this worth leaving comfort, budget, or career capital for? | Validation Sprint, Demand Validation Sprint |
| Pre-seed | Is there real demand and a wedge worth building? | Validate, Validation MVP Build |
| Seed | Can we turn product into revenue, pilots, or repeatable traction? | MVP Diagnostics, V1 Product Build, Grow |
| Growth-stage | Which motion, product bet, or market should scale next? | GTM Motion Buildout, First Traction System, Scale |
| Mature company | Should this initiative receive serious budget and roadmap commitment? | Market Entry Proof, AI / Data / Cloud Modernization Engagement |
| Fund or partner | How do we create repeatable proof across multiple opportunities? | Portfolio Proof Program, Co-Build / Joint GTM Partnership |
Build-stage engagements live in the Build family.
Good proof creates a decision, not just a deliverable.
A useful engagement may lead to a build, launch, market entry, GTM investment, or modernization plan. It may also show that the team should narrow the segment, change the offer, revisit the workflow, or stop. That is not failure. That is the point of proof.
- Build
- Launch
- Sell
- Pilot
- Scale
- Sequence
- Change
- Stop
Working with us should feel like one connected product conversation.
You should not have to brief one team on strategy, another on development, another on marketing, and another on analytics. Proof Engine keeps the core question visible while moving across the work needed to answer it.
- Clear assumptions before execution.
- Tangible artifacts when evidence needs something real.
- Fast loops between customer signal, product work, and GTM.
- Decision summaries that make next steps explicit.
- Scope discipline tied to the proof needed.
The way we run delivery, or fit into yours.
We can run any of these ourselves, join the one you already have, or help you introduce it and make it routine.
Scrum and sprints
- When it fits
- Product work where scope changes as you learn.
- What we do
- Run the sprints as your scrum team, or join yours with engineers and a delivery lead.
Kanban
- When it fits
- Support, maintenance and operations, where work arrives unpredictably.
- What we do
- Set up the board and the policies, then run the flow, or take over the one you have.
Scaled Agile (SAFe)
- When it fits
- Large organizations with many teams on one product.
- What we do
- Join your release train, or introduce and run one.
Waterfall and stage gates
- When it fits
- Fixed scope, regulated work, or software that depends on hardware.
- What we do
- Plan the stages and gates, write the lifecycle down, and keep change controlled and visible.
However you run the work, we build with an agentic harness.
AI coding agents write the code inside a controlled loop. Specs go in, automated tests and evaluations decide what passes, and a person reviews before anything ships. All of our delivery now runs this way: the same quality gates, reached faster.
- Specs before code. Agents work from what was agreed, not from guesses.
- Tests and evaluations as the gate. Failures go back to the agents.
- A person approves what ships. Nothing reaches production unreviewed.
Pick the format that fits.
From a single consultation to a dedicated team, the format follows what you need.
Consulting
Advisory work: second opinions, architecture and process guidance, answers to the hard questions, written down.
Work by the hour
Engineers and leads billed by the hour, with no minimum, for work that does not fit a fixed scope. Fixed-scope engagements start at $4,500.
Dedicated team
A multi-role team, with engineers, QA and a delivery lead, that works only on your product, as an extension of yours. Three months minimum, then month to month with 30 days’ notice.
Questions about how we work.
Can you work inside our process, or do we have to adopt yours?
What is the agentic harness?
Does the harness replace Scrum or SAFe?
Can you provide a dedicated team?
Do you work by the hour?
Have you run SAFe?
What is the difference between discovery and validation?
What is the Proof Loop?
What is the difference between signal and proof?
Why is discovery paid, and what do you get from it?
How do you decide what to build for a test?
What happens if the evidence says stop?
Tell us how you like to work.
Describe the team, the process and the problem. We reply with a clear next step — usually within 24 hours.