The design-partner program

Five teams · four weeks · measured honestly · Last updated 2026-07-10 · Ousios LLC

Composa's limitations page says it plainly: we have no published outcome evidence yet. This program is how that changes — with real teams, real measurements, and case studies that include what didn't work. If you've been waiting for proof before trying a tool like this, you can be the proof instead, on unusually good terms.

The deal, whole and honest: you get everything Composa offers, free, plus hands-on setup help directly from the founder. We get your measured before/after numbers and permission to publish a case study — anonymized if you prefer, named if you're willing. The case study gets published whatever it shows. If Composa doesn't work for your team, that finding is part of the bargain too.

Who it's for

Teams of roughly 5–15 doing knowledge work — mixed roles beat uniform ones (the tool maps working-style differences; a team of eleven identical specialists gives it less to map).
Already AI-curious — you don't need experience, but you need appetite. This isn't a program for convincing skeptical leadership; it's for teams whose leadership already wants to find out.
Willing to measure. The whole value of the program — to both of us — is the numbers. If self-timing a few recurring tasks for two baseline weeks sounds like too much, this isn't the right fit (and that's fine).

What the four weeks look like

Week 0	A 30-minute setup call. You pick 2–3 success metrics from the pilot menu (or bring your own), we write down pass/fail thresholds together, and your team starts baseline measurement.
Week 1	Everyone takes the 90-second assessment; you run the team debrief with the Manager's Guide (founder on the call if you want backup). Team picks 2–3 workflows with volunteer owners.
Weeks 2–3	Run the workflows in real work. Mid-point check at day 10 — anything not sticking gets swapped, not forced. Second measurement window at the end of week 3.
Week 4	A 45-minute results call: your numbers vs your thresholds, what survived, what didn't, and why. We draft the case study together; you approve every word before anything is published.

What we ask, precisely

Two calls (30 + 45 minutes) and the day-10 pulse-check email.
Your baseline and week-3 measurements — the same simple self-timed numbers the pilot uses.
Permission to publish the case study, with your choice of attribution: named, industry-only ("a 9-person product team at a healthcare SaaS company"), or fully anonymized. You approve the final text either way, and the privacy policy covers everything else.

What you should honestly expect: two weeks measures quick wins, not compounding effects — some of the value shows up at day 60, after the program ends. And roughly one workflow in three gets swapped at day 10; that's the system working, not failing. The case studies will say this too.

Why only five teams

Because the founder personally runs every setup call, debrief backup, and results session — and because five honest case studies beat fifty shallow ones. When the five slots are taken, this page will say so.

Ask about a slot →

One email, three lines: team size, what you do, and the recurring task that eats the most time. You'll get a straight answer about fit within two business days.