A candid word about authorship. The discussion paper came from the HR practitioner, not the scientist — and it said so on page one. What the people function brought was the pattern: delegations fluency, the instrument archaeology, the design discipline borrowed from another industry. What it could not bring was the thing that makes any of it true — the lived decision moments of a principal scientist, the technical judgement about which risks are real, and the standing to make the frame legitimate with peers.
This is augmented partnering in practice: the partner brings a working idea at 70%, the technical leader contextualises and corrects it, the floor tests it, and the loop runs again. The idea travels; the ownership transfers.
And it began, literally, with the technical leader’s question — how could decision-making be made clearer for her leaders? The practitioner, knowing the direction, workshopped the concept with AI in days: practitioner judgement steering, AI drafting and redrafting cheaply, every verdict priced to be corrected. That pairing is the augmented practitioner — and the paper was its output. The fast part. What follows it needs the people who live the decisions.
The direction came from the technical leadership — the paper answers a question, it doesn’t invent one. Fast means nothing here is precious: every verdict is priced to be corrected.
The lived decision moments, named in the leaders’ own words; the judgement about which risks are real and which checks are load-bearing; the peer legitimacy that turns a frame into how we work here — only they can grant it.
The frame, the instrument analysis, the card architecture — a concrete starting artefact, not a blank page.
Which decision moments matter, what the cards must say to be credible, where the frame would mislead.
Cards used in anger for a full cycle. Every miss is logged — a wrong light is data, never blame.
Corrected cards become the standard, owned by technical leadership, maintained by the people function as instruments change.
The paper closed with a pathway, not a plan — deliberately small: one lane, one directorate, one card set, proven before it’s scaled.
A 90-minute working session with one technical directorate: surface and rank the top decision moments across all four lanes. No pre-work, no slides — just the map.
Output → a ranked Decision Moments Map, in the leaders’ own language.The people function drafts 12–15 Decision Cards (People lane first — the map already exists) plus a Bench Card per leadership role. The technical leader red-pens every card.
Output → a v1 card set with clause references, endorsed where the instrument requires it.One directorate uses the cards for a full month. Every decision that doesn’t fit a card gets logged in two lines. Misses are the most valuable output of the whole pilot.
Output → a tested deck, a miss log, and honest verdicts on the frame.Review with technical leadership and the workforce director: fix, extend to a second lane (research is the obvious next), or stop. Stopping cheaply is a legitimate success.
Output → a recommendation to the executive, with evidence attached.Total ask: one working session, one red-pen cycle, and one month of honest use. The design labour sits with the people function, where it belongs. The judgement sits with technical leadership, where it belongs.
Practising what it preached, the paper closed with one decider and three calls — each with a suggested default the technical leader was free to overrule.
One directorate, chosen for readiness rather than need — a leader who’ll red-pen generously and use honestly.
The people who live the decisions — not a committee. A bench: two or three technical voices plus the pen.
The frame fails if it’s filed. The cards need a home at the point of decision — physical, digital, or both.
Every instrument in this case exists for a good reason. Delegations protect people, public money and procedural fairness — nothing here proposes weakening them. What the case proposes is a translation layer: the hard thinking done once, by the people paid to do it, so the person at the bench gets clarity at the moment it counts.
For practitioners, four things to take away:
Never ask the operator to navigate the instrument. Sort the instrument for the role — clause references intact — and the defensibility travels with the simplicity.
The best decision-design on earth lives in places experts don’t look. Complexity of the domain is the argument for better-engineered supports, not against them.
A finished artefact gets admired; a 70% artefact gets owned. Ownership transfers through correction, not presentation.
The tool succeeds when the technical leader hands it to her people in her own voice — and the practitioner’s name isn’t on it anywhere.
Augmented partnering, on your organisation’s hardest governance problem. When you’re ready, it starts with a conversation.
Start the conversation →