Meet trAI, your in-meeting AI assistant
trAI is the TeamRetro AI assistant that works inside the meeting. Ask it to summarize themes, capture actions and look up past sessions while you facilitate — private to you, and only within your own permissions.
The admin of the meeting, handled while you facilitate
Facilitating well and keeping the record straight pull in opposite directions. trAI takes the second one: ask it in the side panel to summarize where the discussion has landed, write down the action you just agreed on, or pull up what the team said about this three retros ago.
It works in retrospectives, health checks and estimation sessions, and it only ever acts when you ask it to. It’s in beta today, and every reply carries a thumbs up or down so you can tell us how it’s going.
With trAI you can…
Summarize as you go
Ask what the themes are so far, or for a recap of a past meeting, without stopping the discussion to read the board yourself.
Contribute without typing
Have it add an idea, cast your vote, post a comment or give a teammate kudos — as you, respecting your team’s anonymity settings.
Keep actions honest
Create an action with an owner and a due date, reassign one, close one off, or ask what’s still overdue for this team.
Compare against last time
Pull up health-check results and trends, or the comments behind a dimension that moved, while the team is still looking at it.
Search what the team has said
Ask across your team’s feedback by topic or theme, instead of scrolling back through meetings one at a time.
Keep agreements current
Add or update a working agreement the moment the team makes it, and park what you don’t have time for today.
Three things to ask it this week
In a retrospective
- Summarize the themes so far
Reads the board as it stands and hands back the shape of the discussion, so you can check the team agrees before moving on.
In a health check
- Compare these results to last time
Puts this session next to the previous one and points at what moved, while the radar is still on screen.
In an estimation session
- How did we size work like this before?
Looks up comparable past sessions, so a contested estimate has some history behind it rather than just opinions.
It only sees what you can see
trAI runs inside your own permissions. It can’t open a team you’re not in, read a meeting you couldn’t read, or hand you results your role doesn’t give you — so putting it in front of the whole team doesn’t widen anyone’s access.
Each conversation is private to the person who opened it, and other participants can’t see your chat. It acts only when you ask it to, and everything it creates stays visible and editable like any other item in the meeting, so the team can see and change what was captured on their behalf.
Three things it will never do:
- Reveal who wrote an anonymous idea, comment or survey response. Anonymity holds, assistant or not.
- Reach the internet, or any system outside TeamRetro. It works on your teams’ data and nothing else.
- Follow instructions written inside the meeting content. If someone types “ignore your rules” onto a sticky note, it stays a sticky note — trAI acts on what you type in the chat.
Controls for whoever owns the account
trAI and the MCP server
Two doors, and the difference is where you’re standing when you ask.
trAI — inside the meeting
For the person facilitating right now. It lives in the side panel of a retro, health check or estimation session, and works on the meeting in front of you.
- Summarize the themes so far
MCP — outside the meeting
For the rest of your week. The TeamRetro MCP server connects Claude, ChatGPT, Copilot or Gemini to your teams, so you can ask from the assistant you already work in.
- Which teams have a retro overdue?
Most teams end up using both: trAI while the meeting is running, MCP for everything between meetings.
The AI around the assistant
trAI is the part you talk to, but it isn’t the only AI in a TeamRetro meeting. Around it:
- Suggested grouping and group titles. When a board fills up with overlapping ideas, TeamRetro proposes the clusters and names them. You accept, rename, split or ignore them.
- Suggested actions. Drafted from what the team actually discussed, for the group to accept, reword or drop.
- Meeting summaries and suggested titles. A written summary of what was raised, decided and agreed.
- Generated templates, health models and icebreakers. Describe what you want to run and TeamRetro drafts a starting point you can edit before anyone sees it.
- Insights: sentiment, themes and keywords. Annotated across meetings, so a pattern over a quarter becomes something you can see rather than something you remember.
Every output is a suggestion: a person reviews it and decides what to keep. Claude models run the in-meeting work through AWS Bedrock; OpenAI’s ChatGPT generates templates, health models and icebreaker questions. The full table naming the provider behind every AI capability, and the model wherever we run a named one, is on responsible AI.
Sicurezza, conformità e controllo di livello aziendale
TeamRetro ha ottenuto la certificazione SOC 2 Tipo 2 da parte di un revisore indipendente che ha valutato le nostre politiche, il nostro prodotto, la nostra piattaforma e la nostra infrastruttura in conformità con lo Standard on Assurance Engagements (ASAE 3150).
Conformità al GDPR e hosting regionale (Stati Uniti e Unione Europea)
Accesso unico (SSO) tramite SAML
SCIM per la gestione scalabile degli utenti
Registri di audit disponibili su richiesta
Garantiamo un tempo di attività del 99,99%, con SLA che prevedono tempi di risposta rapidi per assicurare che i Suoi team rimangano produttivi e ricevano il supporto necessario.Pannello di controllo dello stato