01 / Field Notes
Practical notes on AI leverage.
Observations from the places where expertise, workflow and automation start to compound.
- Notes recorded
- 08 notes
- Kinds
- Signal / Method / Build Note
- Last updated
- 2 October 2026
- Written by
- Rich, Walking With Robots
Filter by kindShowing 08 / 08
Who owns the relationship when your agent does the work?As work moves through a preferred AI agent, software vendors may keep the data while losing the direct relationship with their customers.Read note 02Build the System, Not Another AppAI makes it easier to build quick proofs, but disconnected tools can recreate the fragmentation they were meant to fix.02Text only03The Bar Is Hours Back, Every WeekAI confidence theatre makes useful, practical wins look too small. The better test is simple: did the work change, and did someone get meaningful time back?03Text only04Organisations MUST own the learning loopModel access is rented and isn’t durable. The real advantage is the learning loop your business owns.
05The harness matters more than the modelThe model matters, but the system around it matters more. This is how Codex started feeling less like chat and more like delegated work.
06The value is in the overlapAI adoption keeps stalling in the gap between tool expertise and business reality. This is why WWR is choosing to work in the middle.
07The pro-worker principleYour best people carry more business knowledge than your systems do. Pro-worker AI starts by capturing that expertise, not replacing it.
08The ceiling set by the leader’s first experienceOne underwhelming AI experiment can set the ceiling for the whole business. The ceiling lifts when AI can work with a leader’s own judgement.08Text only
05The harness matters more than the modelThe model matters, but the system around it matters more. This is how Codex started feeling less like chat and more like delegated work.06The value is in the overlapAI adoption keeps stalling in the gap between tool expertise and business reality. This is why WWR is choosing to work in the middle.07The pro-worker principleYour best people carry more business knowledge than your systems do. Pro-worker AI starts by capturing that expertise, not replacing it.08The ceiling set by the leader’s first experienceOne underwhelming AI experiment can set the ceiling for the whole business. The ceiling lifts when AI can work with a leader’s own judgement.08Text onlyEnd of index / 08 notes
Next / Working conversation
Bring one real workflow.
Use a 30-minute working conversation to look at where the work slows down, what knowledge it depends on and whether there is a useful first step. The conversation is not part of the paid engagement.