AI Impact HubProjects
A project takes your outputs from zero to 80%.
For anyone still opening a fresh chat and typing from zero every time.
Try it right here ↓Free, like everything in the Hub. No signup to read it.
A project is what Kyle calls a contextually aware ecosystem: a folder inside Claude or ChatGPT holding your organisation's context — strategy, reports, guidelines — that the AI draws on in every conversation. Same prompt in a bare chat: a template anyone could have written. Same prompt inside a project: something 80% there.
Kyle's standing frame for all of this is the intern. Imagine an intern on their first day: intelligent, motivated, has read every book and every website — and knows nothing about your organisation beyond what is public. Send them off to work without onboarding and the work will be poor, and it will not be their fault. A project is the onboarding.
His crispest distinction: the prompt is the how — the instructions. The context is the what — the material to work with. Most people polish the how and starve the what, and the what matters at least as much.
One mechanical fact changes how you build them: the instructions field is fed to the AI on every single message — that is where guardrails live — while uploaded files are retrieved only when relevant. Instructions always on, files on demand.
I open a temporary chat, so it has no memory, and I say: draft a fundraising email. What you get is a template. Then you run the exact same prompt in a project that has the information about the organisation — and all of a sudden you're getting something that's 80% there.Kyle, running the same prompt twice in front of a room, March 2026.
- Create the containerOne project per ecosystem of work — leadership, a program, a campaign. Switch memory to project-only where offered. Kyle's line: what happens in this container stays in this container. It's like Vegas.
- Write the instructions for the environmentKeep them generic to the space, not task prompts: who you are, how to speak, what to never invent. Task prompts live in a prompt library or a skill, called when needed.
- Add the onboarding documentsWhatever you would hand a new staff member: strategic plan, impact report, communication guidelines. Curated — the essence, not the whole shared drive.
- Or get interviewed insteadKyle's favourite way to fill the context is not writing documents — it is having the AI interview you, one question at a time, and write them for you.
- Watch for clutterHis coinage: context clutter, context slop. Five years of history dumped in is worse than nothing — the model drowns. Give it enough to do the job well, and no more.
Three prompts from the way Kyle actually teaches projects. Start with the test — it is the whole argument in two minutes.
Draft a fundraising email for our current appeal.
Run this twice: once in a bare temporary chat, once inside a project that holds your organisation's documents. Put the two outputs side by side. That difference is what context is worth.
Next: Connectors.
AI that can see your work is a different tool entirely.
Read Connectors →