Northwind Studio
Media / Creative · United Kingdom · 6–15 people · Mixed OS · Restricted, IT-managed
Existing stack: Microsoft 365, Slack, Notion, Figma, GitHub
- PriyaStudio lead
Cautious optimist. Uses ChatGPT for first drafts; worried about the team's wellbeing.
- SamSenior designer
First time properly trying AI. Open about feeling behind.
- JordanEngineer
Daily Cursor + Claude user. Wants the team to catch up but knows it's complicated.
- MayaProducer / Ops
Has tried to start this conversation before and it stalled. Re-entering quietly.
- TheoJunior designer
Uses ChatGPT for research and writing. Worried their role is the one being automated.
- AnikaClient services
Hasn't used any AI tool yet. Sceptical but willing.
In place
3We have an acceptable use policy from last year — covers data and confidentiality.
In place— PriyaEngineering already uses Copilot and Cursor with paid licences.
In place— JordanStanding weekly studio meeting where we could pick this up regularly.
In place— Maya
Decided
3Client deliverables are always reviewed by a human before going out — non-AI-specific, but it stands.
Decided— PriyaCurious about Figma's AI features for first-pass layouts.
Tool of interest— SamNotebook LM — to make our messy research browsable for the whole team.
Tool of interest— Jordan
Open questions
3Can we use ChatGPT with anything that mentions a client by name? We've never actually agreed.
Open question— MayaWho decides when we adopt a new AI tool — and how do we tell the rest of the team?
Open question— SamWhat does 'good' look like for someone who isn't using any AI yet? Is that fine?
Open question— Anika
Not yet
3No training plan for non-technical staff.
Not yet— PriyaNo clear budget owner for paid AI tools — we're expensing piecemeal.
Not yet— MayaNo shared place to share what's working and what isn't — knowledge stays in DMs.
Not yet— Theo
Worries & non-negotiables
5I'm worried my role is the one being automated, and nobody's saying that out loud.
Worry— TheoI don't want to look stupid in front of people who already use this stuff every day.
Worry— AnikaIf we move too fast we'll lose the craft we're known for.
Worry— SamClient data does not go into any tool we haven't approved. Full stop.
Non-negotiable— PriyaNo AI used to evaluate or rank team members.
Non-negotiable— Priya
Alignments
Quality and craft come first
Across roles, people agreed any AI use has to protect — not erode — the studio's craft and review process.
We need a shared place to learn together
Engineers, designers, and ops all named the lack of a shared channel for what's working and what isn't.
Tensions
Daily users vs first-timers
Some people are well ahead and want to move; others are just opening the door. Both are valid — the pace question is unresolved.
'Try things' vs 'don't break trust with clients'
Curiosity about new tools sits alongside a strong non-negotiable about client data.
Open questions worth answering
Where exactly is the line on client data and AI tools?
It's the single most-cited worry and currently unwritten.
How does someone who isn't using AI stay a full member of the team?
Anika asked it; nobody else answered. It deserves a clear answer.
Possible next steps
Write the client-data line in plain English, together
30 minutes at next studio meeting. Priya facilitates.
Start a #ai-learning channel — 'what I tried this week'
Low-stakes, no expectation to post. Jordan kicks it off.
Pair Anika with Theo for one hour on a real task
Anika opts in. Theo guides — not teaches.
In place
- Acceptable use policy (last reviewed 12 months ago)
- Human review on all client deliverables
- Engineering tool licences via IT
Decided
- Figma AI features to be trialled by Sam on internal work first
- Notebook LM trial led by Jordan, 4 weeks
Open questions
- Client-data line is not yet written down
- No clear sign-off path for new AI tools
Missing / Not yet
- No training for non-technical staff
- No shared learning channel
- No named budget owner for paid AI tools
Worries & non-negotiables
- Client data stays out of any unapproved tool
- No AI used to evaluate or rank team members
- Human review remains mandatory on client work
We will do
- Agree the client-data line in writingOwner: Priya
Drafted at next studio meeting; signed off within two weeks.
- Run a 4-week Notebook LM trial for shared researchOwner: Jordan
Two test projects, weekly check-in, clear stop-or-keep call at the end.
- Start a low-pressure #ai-learning channelOwner: Maya
No expectations to post; Maya seeds one observation a week to start.
We will not do (and why)
- Roll AI tools out to the whole team this quarter
Half the team isn't ready, and forcing it would undo trust we've spent years building.
- Use AI for performance review or workload allocation
Named as a non-negotiable by the lead and supported by the room.
Words for the moments you'll face
Hosts get a copy-and-read prompt bank for opening the room, surfacing worries, breaking deadlock, and closing with named owners. No improvising the hard bits.
Use one of these in the first two minutes. Read it out, slowly.
- Before we start: there's no required level of AI experience here, including none. Skepticism is welcome. Curiosity is welcome. Nothing you write will be used to evaluate you.
- We're not deciding everything today. We're trying to see what we already think — together — so we can choose the next small move.
- If something we say bothers you, that's data. Please put it in worries or non-negotiables rather than holding it.
Ready to do this with your team?
You can run a session solo to try it, then invite the team when you're ready. Nobody else needs an account.