05 / Keep going
The material.
The next conversation.
The session ends. Your opportunity to explore does not.
The presentation and guides.
Session slides / PDF
The full presentation.
The 33-page final deck from 12 October. The companion explains the frameworks and demonstrations that the slides alone cannot show.
PracticeThe complete Painkiller Protocol.
Three conversations, copyable prompts and a reusable working kit.
Learning planTwelve weeks to a better way of working.
The on-screen example, a sequence to adapt and a prompt for your own plan. Targets remain targets until tested.
FeedbackA more useful conversation with AI.
The feedback shown in the session, explained and turned into a practical exercise.
From the recorded Q&A
How does an AI team
work together?
Is this a multi-agent setup?
Yes. I described using different agents and services for different strengths, with a coordinating agent managing tasks. It is a designed way of organising my work, not a claim that every item in my catalogue is an independently running agent.
Can the agents interact on one task?
In my setup, they can hand work over through a shared working space. One task might produce a source summary, another a research note and another a draft. The next contributor can read what already exists and continue from it.
The useful idea is a clear handover: what was asked, what was found, what remains uncertain and what the next task needs. You can practise that with one AI conversation before building a more elaborate system.
These two questions were spoken in the closing Q&A and are paraphrased without attendee names. The separate meeting-chat export and complete poll results have not yet been incorporated.
Places I pointed to.
My working setup
Meet my AI team ↗
The public catalogue shown during the session. Use it to explore roles and tools, not as a list you need to purchase.
Public personal contextGet to know me. Let your AI read along. ↗
My own account of my life, values and working preferences, with sources and room for your judgement.
Microsoft 365Start a conversation with Copilot ↗
Sign in with your own account. Capabilities and access depend on your organisation and subscription.
Another tool shownChatGPT ↗
I used a context-aware feedback conversation. The activity access shown is optional and specific to my setup.
Pocket-agent example / AppleMuse from Meta ↗
The App Store listing for the example discussed in the session. Check current availability and the permissions you would be granting.
Pocket-agent example / AndroidMuse on Google Play ↗
The corresponding Android listing. Ratings and features are snapshots, not guarantees of how a tool will work for you.
Research shown on screenExplore questions with sources ↗
Perplexity was used in the pocket-agent discussion. This is the service home, not a link to the particular research conversation.
Garry KasparovThe Chess Master and the Computer ↗
His account of human-machine chess and why the process matters. The three-law framework in the talk is my teaching interpretation.
Ethan MollickOne Useful Thing ↗
Further reading from the author of Co-Intelligence, credited in the Four Principles.
Stephen CoveyConcern, influence and proactive choice ↗
The source framing behind the three-circle adaptation I used.
Don't be a stranger.
If something here helps you make a useful change, I would love to hear about it. If it raises a better question, that is a good reason to connect too.
Read with your AI · Markdown →
About this companion and its sources
Prepared from the final 33-page presentation, the Circleback session transcript and selected frames from its screen-share recording. The pages follow the delivered talk and expand its methods into practical exercises. They are an edited companion, not a verbatim transcript.
Karin, Marvin and my working setup are my accounts of experience. The Kasparov framework is my teaching interpretation. The fear curve is an illustration. The 16-hour learning plan is a target-based example, not measured savings. Product snapshots and broad statistical claims have not been republished as current facts.
The recorded closing questions are paraphrased without attendee names. Meeting-chat contributions and full poll results will be added after they have been supplied and reviewed. The raw recording, private context and source transcript are not published here.
AI + JJ · 12 October 2026. No added analytics, sign-in or data collection. The appearance preference is stored only in your browser. External links have their own services and policies.