AI Adoption

Something very big is happening. This is the biggest transformation in my lifetime — bigger than the rise of the internet. And what matters most right now is what it means for how we work as a team.

The Core Thesis

The leading AI-native companies have already crossed a threshold: their engineers are not writing code — they are architecting and coordinating agents that write code. This is where all organisations are heading — and Datopian needs to get there fast.

For the Innovations Team and Lead Link, this shift needs to happen now. Ultimately, the whole organisation needs to move this way — engineering, marketing, GTM, services — but it starts with the core group leading by example.

The destination is clear: every person operates as an agent coordinator, not a manual creator. The cost of creation has collapsed. What matters now is imagination, architecture, and the ability to orchestrate AI to execute at speed.

What This Means in Practice

Go all-in. This isn't something you dabble in. If you're only using AI a little bit, you're falling behind. We want a team where people are spending serious time and tokens pushing the frontier of what they can build with agents.

Stop touching code. The direction of travel is clear: from AI-as-autocomplete, to AI-as-pair-programmer, to managing fleets of autonomous agents. We should be pushing hard towards that latter end — architecting systems, coordinating multi-agent workflows, and letting AI do the execution.

Rethink your tools and habits. The old workflows are dead. Dictate instead of type. Markdown in repos instead of Google Docs — because AI can work with that efficiently. Use skills, multi-agent workflows, long-running systems. The friction of creation is collapsing and your habits need to keep up.

Work in the open. Work publicly, even within the team. Public repos, open markdown, visible progress. This is the easiest way for AI (and humans) to read and build on what you're doing.

The Bar is Higher, Not Lower

This shift doesn't mean the work gets easier — it means the nature of the work changes. Operating this way requires a higher level of capacity: you need to think architecturally, manage complexity across multiple agents, maintain quality when output velocity is 10x what it was, and constantly adapt your mental models. It's a fundamental reorientation of mind — from "how do I build this?" to "how do I describe and orchestrate this so agents build it well?"

Not everyone will get there at the same pace, and that's fine. But the willingness to make this mental shift is non-negotiable.

The Innovations Team

We're forming a core group — the Innovations Team — to lead this transformation and prove out these ways of working. The Innovations Team and Lead Link go first and go hard. As patterns and workflows mature, they spread to the rest of Datopian.

The Opportunity

The constraint is no longer skill or time — it's imagination and compute. What we can produce as a small team operating this way is extraordinary. We should be spending money on tokens and massively leveraging what our existing people can do. A one-pizza team coordinating agents can outproduce traditional teams many times its size.

---

Rollout Plan

The People

• Innovations Team (go now): Anu, Joao, Ola, Rufus
• Lead Links (onboard next): Osahon, Daniela (+ Anu who bridges both groups)
• Wider Datopian (~25 people): follows as patterns are proven

Step 1 — Baseline Survey

Before we can move people forward, we need to know where they are. Send a quick survey to the whole team. The key question: where are you on the 8 Stages of AI-Assisted Development?

1. Zero/Near-Zero AI — maybe code completions, sometimes ask Chat questions
2. Coding agent in IDE, permissions on — agent in a sidebar, asks permission to run tools
3. Agent in IDE, YOLO mode — permissions off, trust is up, agent gets wider
4. In IDE, wide agent — agent fills the screen, code is just for diffs
5. CLI, single agent, YOLO — diffs scroll by, you may or may not look at them
6. CLI, multi-agent, YOLO — 3-5 parallel instances, you are very fast
7. 10+ agents, hand-managed — pushing the limits of hand-management
8. Building your own orchestrator — on the frontier, automating your workflow

Reference: https://ailearnedtoday.com/ref/evolution-of-programmer-yegge

Step 2 — 1:1 Conversations

Follow up the survey with short conversations, starting with the Innovations Team and Lead Links:

• Where did you place yourself? Do you want to move up? What's blocking you?
• What tools are you using today? What's your daily workflow?
• What would help you move to the next stage?

Step 3 — Lead by Example

The Innovations Team works visibly in the new way. Share workflows, share wins, share failures. Make it concrete — not a memo, but a lived demonstration of what Stage 6+ looks like in practice.

Step 4 — Spread

As workflows and patterns stabilise, bring in Lead Links, then the wider team. Pair people up. Run working sessions, not training sessions — real work, done together, with agents.

---

Reference: The 8 Stages of Dev Evolution to AI-First

From Steve Yegge's Welcome to Gas Town:

Recorded work and operating context — September 2026 scope review

• AI Frontier Fridays: A recurring sharing practice. The first G-Stack session ran May 22; ownership rotation was still unresolved in the last record. Execution/evidence. Historical source.

• Data Portal SRE Agent: User reports on September 28 that it is done, working and deployed. Treat it as an operating capability; deployment details and acceptance evidence are not recorded here. The previous checklist is historical, not a current development backlog. Execution/evidence. Historical source.