Room Wisdom by Ethos LabPilot
Hear the whole room.
Recorded group conversations, turned into verified qualitative evidence and quantitative data. In all 24 official EU languages.
The demo is a synthetic session in English.
The problem
Qualitative evidence is rich, but slow, hard to compare and easy to distort.
Time
Coding one 90-minute group by hand takes days. A multi-group study takes weeks.
Inconsistency
Two analysts code the same talk differently, and nobody measures by how much.
The loudest voice wins
Summaries over-weight whoever spoke most. Quiet participants and silence disappear.
Counting the wrong thing
“Mentioned 40 times” can mean one person repeating themselves.
Untrustworthy AI
Generic AI summarisers paraphrase quotes, invent consensus and leak personal data.
Language
Non-English material is often translated before analysis, and nuance is lost.
The product
What a group thinks, person by person.
Who took which position
Consensus · 5 for, 1 silent
The divide over a car-free centre is about who has an alternative, not about age.
Contested · 3 for, 2 against, 1 mixed
A pilot closure is the best first step before any permanent pedestrianisation.
Insufficient coverage · 1 for, 1 mixed, 4 silent
A car-free centre would allow people to give up owning a car.
- For
- Against
- Mixed
- Silent, not agreement
How it works
A transcript in. Evidence out, with a researcher in charge.
01
You provide the transcript
From a focus group or an interview, in any of the 24 official EU languages. Names are removed before analysis.
02
Room Wisdom analyses it
It maps who said what and who took which position. A researcher approves the key judgement calls.
Researcher sign-off
03
You receive the evidence
An interactive dashboard and a written report, with every quote checked against the transcript.
Processing in a matter of hours, not months.
The dashboard
Every theme, every voice, every quote
Trust
Trust is the product.
Six design choices set it apart from AI summarisers.
Verbatim guarantee
Every quote is checked word for word against the transcript. Nothing unverified is published.
Silence is not agreement
Each person is recorded as for, against, mixed or silent. Silence is never counted as consent.
People, not mentions
Every finding counts participants as well as mentions, so one talkative person cannot fake a trend.
Human in the loop
A researcher approves the key judgement calls before the analysis continues.
Measured reliability
Consistency is measured on every analysis and reported alongside the results.
Private by design
Names are removed before analysis, and the client keeps control of the raw data.
Pilot results · one session
One 30-minute focus group, six participants, 188 contributions.
- quotes verified word for word against the transcript
- 188 / 188quotes verified word for word against the transcript
- coding consistency, and κ 0.93 on stance judgements
- κ 0.95coding consistency, and κ 0.93 on stance judgements
- of processing for the session
- ~120 minof processing for the session
- in numbers or quotes, found by an independent audit of the report
- 0 errorsin numbers or quotes, found by an independent audit of the report
Synthetic demo session, in English: 122 of 122 quotes verified word for word, κ 0.96 coding consistency and κ 0.94 on stances, about 14 minutes of processing.
The pilot session was held in Greek. These are results from one pilot session, not general accuracy rates. Consistency scores show the analysis is repeatable, not that it is correct. A human-coding benchmark is still running. Results describe the people in the room.
Use cases
For people who run structured conversations and must defend their conclusions.
Academic and policy researchers
Audit trail, reliability scores, verbatim quotes
Public bodies and consultation teams
Who supported what, minority views kept, fair airtime
Deliberation and dialogue organisers
Stance maps, consensus and contested points
Market and UX research agencies
Same-day reports, cross-group comparison, dashboards
NGOs and think tanks
Reports in the session’s language, quotes that hold up
Roadmap
From what one group said to how any conversation moves.
Horizon 1 · Product
What groups say
Focus groups and interviews in all 24 official EU languages, with a report and a dashboard.
Working pilot today
Gate 1
Horizon 2 · Vision
How a conversation moves
Following positions as they shift, across sessions and in negotiations.
Horizon 3 · Vision
Any conversation where people take positions
Larger and more varied settings, added as each stage is validated.
From conversation to evidence.
Three ways to start
Pilot it on a live study
Bring one project of 4 to 20 groups or interviews. You get a report and a dashboard for every session.
Write us into your next bid
Room Wisdom as the qualitative evidence partner in your EU, national or client proposals.
Become a launch partner
Early access, and a say in which settings and features we build next.
Use only with participants’ informed consent. Names are removed before analysis.