AI mini-briefing for the team

For the team meeting, 8 October 2026 (10 minutes). Drawn from my raw notes on AI as the new interface.

TL;DR

The AI is becoming the interface for knowledge work. It is not just automating tasks. Talk to it, and it builds you the view you need (a table, a graph, a plan). Apps like GitHub, Asana and Google Docs become filing cabinets.
Worked example: last night we published the Wisdom paper (web page, full essay, PDF, cover, figures, social images) in a few hours. Before, that needed a separate team and weeks, and we probably would not have done it at all. It is now a written procedure we can repeat.
Building this capability is the work, not a side project: onboarding, planning practice, and a shared plan of work (Beads) that people and agents both read.
Way Into AI (wayintoai.com) is how we share this outward.
Ask: this week, run one real piece of your work through your AI by talking to it, starting from a bead or your plan, and bring back what worked and what got in the way.

1. The shift

Old model: we spent decades, and billions, building applications so humans could do knowledge work: word processors, spreadsheets, Asana, GitHub.
New model: the AI chat (Claude, or ChatGPT, it works the same) is your control plane. You work in it, and the apps behind it are where things get stored.
It is a two-way interface. In: talk, don't type. Dictation is now fast and good, and when you talk you give more context. Don't go back and tidy what you said; just say more. Out: the AI can build whatever view you need in real time, a table, a graph, a page, rather than you hunting through an app.
Task management becomes a conversation: "I finished this", "what's next for me", "show me what this depends on", "give me a plan for the next five hours". That used to need a senior executive with two full-time assistants.
So we choose tools by whether they work well for the AI: fast, reliable, well structured. That is why I use Beads: it holds a proper dependency graph of tasks, runs locally and is quick and cheap for the AI. GitHub issues are slower and clumsier for this.
Caveats: you still need to understand your tools a little, for when things break. We still need Google Docs for real-time co-writing and close editing.

2. Worked example: the Wisdom paper

Wisdom and Wanting What's Good (the wisdom gap paper, Seeds of Renaissance No. 6) went out last night: landing page, full essay on the web, typeset PDF, cover, figures in house style and social images in three crops.

Time: a few hours, me working with AI sessions.
Before: a separate team, over weeks. Realistically we would not have done the PDF and web edition at all.
Now repeatable: one source text makes every output, two commands publish it, and there is a checklist anyone can follow. See How we publish a paper now.

3. What it means for us

Training and building this capability is the work. Fae's and Asdra's onboarding, and AI training and integration, sit at the centre, not on the side.
Planning practice is the skill underneath. Situation, complication, question, then a plan of work (see the September briefing). The AI does far more for you when you can say clearly what you are trying to do.
Beads is our shared plan of work. People and agents read the same list of tasks and dependencies. Start an AI session with a bead ID ("work on lip-xxx") and it picks up the context. The planning repo is the index that tells your AI where everything lives.
How to see Beads today (in a terminal in the planning repo, or ask Claude to run them for you):
- bd ready: what is unblocked and ready to work on.
- bd list: all open tasks. bd show <id>: one task with its details and dependencies.
- A friendlier view is coming (to confirm: what and when).

4. Way into AI

Way Into AI is where we share this outward: a guided way in, and Way Into AI Weekly, now in a new design with a past-issues archive. Read it, forward it, and tell us what is missing. What we learn doing this internally feeds it.

5. The ask

This week, take one real task and do it by talking to your AI rather than typing into an app: start from your bead or plan, ask it what's next, and let it show you the result. Bring one thing that worked and one that got in the way to next week's meeting (to confirm: format and who collects these).