01 · Problem
Problem validation
Customer interviews, usage data and lost deals show which problem is big enough. Assumptions without evidence stay visible as open questions.
AI product management · for founders, CPOs and product teams
The problem first. Then the AI feature.
We turn customer evidence into prioritised AI features, testable requirements and a message sales can explain. Your engineering team builds on a clear basis.
The bottleneck
AI features rarely fail because of the technology.
They fail because nobody checked whether the problem is big enough, how good an answer has to be and who decides when a result is wrong. The team then measures usage instead of quality, and sales explains a feature customers were not looking for.
In The Science of GTM method this work belongs to the SHAPE phase: problem, solution and message come from the same customer evidence. AI product management is therefore not an add-on to GTM but its foundation.
What we work out together
From the customer problem to a testable requirement
Six steps that build on each other. Which of them you need, we clarify in the first conversation.
02 · Strategy
AI product strategy
Where a model creates value, where a simple rule is enough and where you deliberately build nothing. With assumptions, limits and dependencies.
03 · Prioritisation
Weighing value and risk
AI features ordered by customer value and by the risk of a wrong result. High value with high risk means more evaluation and human approval.
04 · Requirements
Testable requirements
Interview statements become requirements with situation, expected result, edge cases and acceptance criteria. Engineering knows what done means.
05 · Evaluation
Evaluation criteria before the rollout
Test cases, thresholds and failure types are set before anything is built. After the rollout the team measures answer quality, not only clicks.
06 · Message and handover
Messaging and handover
The message for sales and marketing from the same evidence, and a handover to engineering with rationale, dependencies and open questions.
Results
What you have in hand at the end
Versioned documents your team carries on. The exact scope is in the offer after the scoping call.
Evidence
Problem evidence
Analysed interviews and data with patterns, counter-examples and open questions.
Portfolio
AI feature portfolio
Prioritised by value and risk, including deliberate stops and waits.
Requirements
Requirements document
Testable requirements with acceptance criteria and edge cases.
Evaluation
Evaluation plan
Test cases, thresholds, failure types and who decides edge cases.
GTM
Messaging foundation
Positioning, core message and evidence, aligned with sales and marketing.
Handover
Handover package
For engineering and GTM: rationale, dependencies, next decisions.
How we work
Analyse. Build. Hand over.
As with every Pedalix service, the work starts with real material and ends with the handover to your team. A reasoned wait or stop is a valid result.
01 · Decision Lab
Sharpen the problem
One to two days on interviews, usage data and the roadmap. A joint decision on which AI features to pursue.
02 · Execution Track
Build requirements and evaluation
In small steps with reviews: requirements, evaluation criteria and message, tested with real customers.
03 · Handover
Your team carries on
Engineering builds on the requirements, GTM on the message. Documented and versioned, no retainer.
Fixed price, agreed in the 30-minute scoping call and confirmed in writing.
Read on
From the blog
SHAPE
AI product management and positioning
All articles on AI product management, requirements and positioning.
ReadArticle
AI Product Management: The PM as Signal Curator
Why AI features need their own measure of answer quality.
ReadArticle
Requirements Analysis for SaaS: Build Less, Learn More
How requirements become testable before anyone writes code.
ReadFAQ
Frequently asked questions
Who is AI product management for?
For founders, CPOs and product teams at B2B technology companies who plan AI features or have already built them and are not sure whether they solve the right problem and how good they need to be.
Do you replace our product management?
No. We work with your product team on real material and hand over requirements, evaluation plan and message. Your team then carries on.
How does this relate to GTM Engineering?
Both use the same customer evidence. AI product management decides what gets built and how it is explained. GTM Engineering puts the message to work in CRM, pipeline and outreach. You can start with either path.
Do you also build the AI features?
Your engineering team builds them. If you want support, Product & Engineering with coding agents joins in, technically led by Ivan Adutskevich. That is a separate service with its own scope.
Who works on it?
Valentin Binnendijk, founder and CPO of Pedalix, and Marc Gasser, who has worked at the intersection of software product management and go-to-market for twenty years. Ivan Adutskevich answers technical questions as lead engineer.
What does it cost?
We clarify the scope in the 30-minute scoping call and confirm a fixed price in writing. We do not publish prices on the website.
Bring an AI feature you are unsure about.
30 minutes with a founder. Afterwards you know whether a Decision Lab fits and what the next step looks like.