The decision layer for product teams
Evidence behind every
product decision.
Humfrid connects your product data, customer conversations and team goals with research agents, to make the right decision obvious in 10 days.
Delivered by product leaders with
AI made building fast. It didn’t make the right thing obvious.
We bring structure to your research, identify and find the data you need, and deliver results that both humans and agents can apply, so your team builds the right thing as fast as it can now build anything.
We deliver an evidence-backed answer in days, not weeks. While others are still debating, your team is already building the right thing.
01Launch Research
Turn open questions into evidence.
Humfrid breaks each question into the hypotheses that would settle it, finds what evidence is missing, and sends agents to close the gaps. Every finding maps to the hypothesis it answers.
-
Product Analytics
Connect PostHog, Mixpanel, Google Analytics or your warehouse. Humfrid pulls the numbers that settle a hypothesis and shows its working.
-
AI + Human-Led Interviews
Combine methods on one question. Every interview is built around the hypotheses, transcribed, coded, and quoted back to the question.
-
Interview Library
Bring in the calls and transcripts you already have. Every call is analyzed and mapped to your hypotheses.
-
Synthetic Testing
Pressure-test an idea against synthetic customers built from your real research, before it reaches a real one.
-
Internal Expertise
The most important knowledge lives in your colleagues’ heads. Humfrid asks the people who know, in Slack, and brings the answer back with who said it.
The method
Rigor before research.
Every question gets the same scientific treatment before anyone goes looking for answers: understand the goal, map the opportunities, write hypotheses that could be proved wrong, and list exactly which evidence is needed and which is missing.
- 01 · Opportunity Mapping
Map every bet.
A product goal becomes a tree: the bets underneath it, ranked by impact, each with the hypotheses that would prove or kill it.
- 02 · Hypothesis Design
Write it so it can fail.
Each hypothesis says what “yes” looks like, which evidence settles it, and what result would change your mind, before any data comes in.
- 03 · Evidence Gaps
Know what’s missing.
Humfrid lists the data each hypothesis needs, checks what you already have, and names exactly what is missing, so nobody researches what you already know.
02Track Your Opportunities
Every product bet, weighed and kept current.
The questions teams stall on, should we build this, should we launch, answered with evidence and kept current as things change.
- Answers with Evidence
Ask any product question in plain words. Humfrid reads your metrics, customer conversations, Slack and docs, and answers with the reasons for and against, each linked to its source.
- Alignment Check
Every answer is checked against your goals and the decisions your team already made, so you know straight away when a new idea pulls the other way.
- Continuous Updates
Monitor the questions you care about. Answers update as new data comes in, and the right person hears the same day something important changes.
03Embedded Intelligence
Evidence right where you work.
The evidence shows up where the decision gets made: with you before the meeting, and with the agents writing your code.
-
Ask in Slack
Add Humfrid to Slack and ask any question where you already work. Before a meeting it briefs you on what needs asking, what changed since last time, and the evidence for your case. Answers and alerts arrive the same way, with the sources attached.
-
MCP for Coding Agents
Claude Code, Cursor and other agents get the same goals, evidence and past decisions as your team, and write their own decisions back.
-
Ready-made Product Briefs
When the evidence settles a bet, Humfrid writes the product brief: the problem, who it is for, the evidence behind it and what success looks like. For what is still uncertain, it prepares an experiment brief, with the hypothesis, the test and the metric that decides it, ready for a coding agent to pick up and run.
Works with
AI-native
This is what an AI-native product team looks like.
Not a tool bolted on to the old way of working. Research agents and the agents writing your code, all working from the same evidence as your team.
- 01
Evidence, not opinions
Before↓The loudest voice wins.
With HumfridThe evidence decides.
- 02
Research that runs itself
Before↓Questions stay open for months.
With HumfridAnswered in days, by agents.
- 03
Agents with judgment
Before↓Builds exactly what it’s told.
With HumfridChecks before it builds.
7× faster decisions 10 weeks → 10 days
Make the right decision
obvious.
Humfrid finds the bets worth making, puts the evidence behind each one, and keeps the whole team building toward them.