The limiting factor isn't the technology. It's organisations — governance built for a different shape of risk, leadership miscalibrated about its own workforce, and what the shift feels like twenty years into a career.
Seven talks and three panels, nearly all from the Leadership track, plus the interviews. The track disagreed with itself in the most useful ways.
¶1Andy Kelk opened with a bait-and-switch: “I'll level with you — this is not 2026, this is 1976. And the technology we're talking about is the personal calculator. And those professionals were maths teachers.” Then his hypothesis about today's resisters: “they're not actually afraid of AI. What they're afraid of is becoming junior again.” Kelk ▸ ≈03:44 When he dug into the resistance, it wasn't dislike of change: “what it really boiled down to was — I don't actually know what I'm supposed to be good at anymore.” Kelk ▸ ≈04:51
¶2The numbers say the problem is organisational, not technical. Christian Dandre brought the survey data — 63% of AI implementation challenges stem from human factors slide — and a sharper finding from running the same workshop with ten CEOs: leadership overestimated its governance readiness by a full point and underestimated its own workforce by the same margin. “The readiness gap doesn't begin at the frontline. It begins in the room where strategy is made.” slide And the human stakes underneath, from Inga Pflaumer's time as a head of engineering — a senior engineer telling her: “I spent twenty years learning this craft. And now AI can do it in an hour. I feel like I wasted my life.” Pflaumer ▸ ≈02:49
¶3Governance got argued out live. Hamish Songsmith's position: adoption isn't optional, traditional risk assessment doesn't fit, and most review processes are theatre — he told one firm: “you don't have a review process, you have a liability.” Songsmith ▸ ≈04:33 His alternative is telemetry over surveys: “session-level telemetry is where your risk sits… if you don't understand how people are using these tools, you cannot speak to whether they're using them safely.” Songsmith ▸ ≈18:32 Aubrey Blanche, from the same stage, refused the binary — “there are more than two speeds; there's not nothing and sprinting” — and argued for deciding your kill-switch thresholds before the incident, because “in the moment, the economic pressure to not shut something down will be overwhelming.” Blanche ▸ ≈21:37
¶4Mandate or grassroots? The track supplied its own experiment. Kelk, day two: “the most effective thing to do was not a mandate, it was experimentation.” One session later, Culture Amp reported the other arm: a deliberately culture-first, volunteer-led year — followed by a reversal nobody planned. “We didn't have any intention of mandating… we're now at a point where we're asking all of our engineers to embrace agentic engineering. That's not a decision I was expecting we were going to make.” Hughes ▸ ≈18:41 The trigger was structural, not ideological: one pioneer doing thirty-five PRs a day next to nine colleagues doing one or two — “the whole team's collaboration system just completely falls apart.” Grigson ▸ ≈20:14 Eric Grigson's summary phrase for all of it: AI is an intensifier.
¶5Two findings to hold onto. The productivity gains are real and partly borrowed: PRs up 27% — and out-of-hours commits up almost 20%. “Yes, there are productivity gains, but they're coming from your engineers' evenings and weekends.” Kelk ▸ ≈11:59 And the thing that actually worked in Dandre's telling wasn't the enterprise rollout but a deliberately unglamorous shipped tool people chose to use — “that's not a product success story, it's a trust story. And trust is the infrastructure that everything else runs on.” Dandre ▸ ≈24:42 Kelk's practical reframe for leaders ties it off: “your most cautious engineers are your early warning system.” Kelk ▸ ≈20:05
“My hypothesis: they're not actually afraid of AI. What they're afraid of is becoming junior again.”
“What it really boiled down to was: I don't actually know what I'm supposed to be good at anymore. I'm not really sure what this means for me. And that's a very different kind of resistance. That's not ‘I don't like change.’”
“He said: Inga, I spent twenty years learning this craft. And now AI can do it in an hour. I feel like I wasted my life.”
“‘Every time we want to do something with AI, we've got the review process.’ — ‘Have you given everyone [ChatGPT]?’ — ‘Yeah, yeah.’ — ‘Then you don't have a review process. You have a liability.’”
“The session-level telemetry is where your risk sits, and that's what you have to understand. If you don't understand how people are using these tools, you cannot speak to whether they are using them safely.”
“I promise you, in the moment, the economic pressure to not shut something down will be overwhelming towards any wonderful individual's ethical commitments.” And on the governance panel: “There are more than two speeds. There's not nothing and sprinting.”
“When we started this journey, we didn't have any intention of mandating the use of agentic or AI coding tools… and we're now at a point where we are actually asking all of our engineers to embrace agentic engineering techniques. That's not a decision I was expecting we were going to make.”
“Maybe there's one pioneer in a team of ten people… literally outputting 35 PRs a day, and the other nine folks are doing one or two PRs a day. And the whole team's collaboration system just completely falls apart.”
“For teams using AI, the number of PRs went up by 27%. Fantastic. They also found that out-of-hours commits went up by almost 20%. So yes, there are productivity gains — but they're coming from your engineers' evenings and weekends.”
“That message is not a product success story, it's a trust story. And trust is the infrastructure that everything else runs on.”
“Your most cautious engineers who are telling you this — they're your early warning system.” And from his closing slide: “Some of your engineers should be a little scared. The question is whether you're listening to what that fear is telling you.”
“Everyone who sets their objective for AI adoption as productivity is not qualified to be deploying AI.”
“The most successful organisations choose something like mission impact or customer value creation as the objective — and then increases in productivity or efficiency are two of a set of tactics.”
“I'm tired. I'm exhausted working at the moment. The excitement — but the level of cognitive load is starting to come at me.”
“They're being empowered to do more across their domain areas… and you just see the speed the organisation is now starting to work at as a result. It's incredible.”
“The flip side is the other half of the businesses I've spoken to, which are too timid to even start taking the first steps… those are the ones that actually need to go and experiment with this, and you do need to accept the risk of the unknown.”
Three arguments ran through the track. Adopt-now versus governance-first (Songsmith and Blanche — converging on sandboxed experimentation, splitting on the default posture). Mandate versus grassroots (Kelk's “not a mandate” against Culture Amp's reluctant reversal, one session apart). Regulation versus transparency (Blanche's pre-committed kill switches against Dandre's “we can innovate without regulation… through transparency”). Under all of it sits Kelk's finding, which both sides need: the resistance isn't fear of a tool — it's not knowing what you're supposed to be good at anymore.
| Engineers | Don't hide your AI use — hidden experimentation means the learning never transfers (younger workers are 38% more likely to hide it — Kelk). Run Songsmith's GRASP on your own agent sessions: governance, reach, agency, safeguards, potential damage. |
|---|---|
| Team leaders | Treat cautious seniors as signal, not blockers — they're the early-warning system (Kelk). Watch the pioneer-versus-everyone-else gap before it breaks collaboration (Grigson's 35-PRs-a-day story). Try a skeptics session: only the unconvinced in the room — “sometimes the most skeptical person has the insight that makes the thing better for everyone” (Hughes). |
| Org leaders | Ask employees before executives; find the product owner, not the most senior person; ship something (Dandre). Set mission or customer value as the objective — productivity is a tactic, not a goal (Blanche). Decide kill-switch thresholds before the incident, and swap surveys for session-level telemetry (Songsmith). |
| Key takeaway | Stop rolling AI out as a tool rollout. Kelk's slide: “Stop treating this like a tool rollout. It is a fundamental change to what expertise means in your organisation.” |
“Stop blocking, start building” versus “there are more than two speeds” — where is your organisation on that line right now? And is that position a decision someone made, or an accident nobody did?
For now these are placeholders. Once the individual sessions are up on Conffab, each title here — and every quote above — will link straight to the talk: video, transcript and all.