Use cases
every opportunity, from first mention to value delivered, scored against the same criteria.
What is a data and AI programme record?
Assettia is the data and AI programme record: one place where an organisation keeps what its programme owns, what it has decided, and what it has proved.
A data and AI programme record is the single, evidence-linked account of an organisation's data and AI programme. It holds the programme's assets, the decisions taken about them, and the proof of what those decisions delivered, as linked records rather than as documents. It is kept by the people who run the programme and read by the executives and directors who are accountable for it.
Why the record exists
Most organisations run their data and AI programme across several places at once: a strategy in slides, a list of use cases in a spreadsheet, a risk register in a governance tool, a delivery plan in a project tool, and the reasons for decisions in the memory of the people who took them. Each is right on its own. None of them agrees with the others for long, and none of them can answer a director's question without someone assembling an answer by hand.
A programme record replaces the assembling. One set of records, kept as the work is done, from which any question a board asks can be answered directly: what do we own, what have we decided, what have we proved.
What the record holds
Six kinds of record, linked to one another.
Data assets
what data exists, who owns it, its quality, its rights, and which use cases depend on it.
Decisions
the gating decisions of the programme, with the rationale and the person who took each one.
Risks
the AI risk register, with controls, regulatory reference points and attestation sign-off.
Value evidence
benchmarks, tracked metrics and measurements, linked to the use cases they support.
Board reports
documents produced from the record, in which every figure names the population it counts.
The three states of a record
Every record in the programme is in one of three states, and the state is derived from the record, never assigned by opinion.
Owned
The thing exists and is on the record: a data asset, a use case, a risk, a stakeholder. Nothing more is claimed about it.
Decided
A named person has taken a decision about it, and the reason is written down. Prioritised, approved, controlled, deferred.
Proved
Evidence is linked to it: a measurement, an attestation, a document. This is the only state a board should be shown as fact.
The sequence matters. A programme that has decided a great deal but proved little looks very different, on the record, from one that has proved what it set out to do. Both are visible at a glance, because the states are drawn the same way everywhere.
Who keeps the record, and who reads it
The record is kept by the chief data officer and the team who run the programme. They capture evidence, prioritise use cases, take decisions and track value in the record itself, so keeping it is not a separate job.
It is read by two further groups. The executive team reads the state of the programme in the words the business uses, with the next decision always on screen. The board reads a document produced from the record, in which every figure traces to a population, empty registers are stated as empty, and nothing has been composed by a model without a person approving it.
What makes a record trustworthy
Four rules, held by the software rather than by good intentions.
- 1
Deterministic first. Where a fixed rule over records can answer a question, it does. Composed narrative is the exception, and it is labelled as such.
- 2
AI drafts, a person approves. Nothing generated by a model writes to a live record. It lands in a reviewable state and becomes a record only when a named person accepts it.
- 3
Every figure names its population. A count says what it counted. The reader can open it and see the list.
- 4
Honest gaps. A register with nothing in it says so. An assessment not yet run is shown as not yet run, never as zero.
What a programme record is not
Three neighbouring categories hold parts of the picture. The programme record sits beside them and links to what they hold.
Not an AI governance platform.
Governance platforms hold models, agents and controls. The programme record holds the whole programme, and links governance to the use cases and decisions it governs.
Not a portfolio or project tracker.
Those hold projects, dates and tasks. The programme record holds evidence, decisions and value, and produces the paper the board reads.
Not decision intelligence.
Decision intelligence tools automate operational decisions. The programme record documents the strategic ones, with the person who took each and the reason.
Some people describe a programme record as the system of record for a data and AI programme. That is fair as far as it goes. The difference is what it records: not the technical estate alone, but the decisions and the proof, in a form a board can read.
How it differs from the documents it replaces
A slide deck describes the programme as it was on the day the deck was finished. A spreadsheet holds a list without the reasons behind it. A programme record is kept as the work happens, so it is current by construction; it holds the reason beside each decision; and it produces documents from itself, so the board paper and the working record can never disagree.
Questions people ask
What is a data and AI programme record in one sentence?
One place where an organisation keeps what its data and AI programme owns, what it has decided, and what it has proved, as linked records that every figure traces back to.
Who is a programme record for?
The chief data officer and team who run the programme keep it. The executive team and the board read it. It is designed so that the same records serve all three.
Is Assettia an AI governance tool?
No. Governance tools hold models, agents and controls. Assettia holds the whole programme: use cases, data assets, decisions, risks, value evidence and board reports, and links governance to the decisions it governs.
What does "every figure names its population" mean?
Every number shown to a reader states what it counted, and the reader can open it to see the list. A count of prioritised use cases is a count of a named set of records, not a figure typed into a slide.
Does the record use AI?
Yes, to draft. A model may propose a summary, a risk or a classification. Nothing it produces becomes a record until a named person reviews and accepts it. Executive surfaces show deterministic facts only.
How does a board use a programme record?
It reads a board report produced from the record, opens any figure to its population, reads the decision behind any priority, and sees which registers are still empty. For boards
How is a programme record different from a spreadsheet of use cases?
The spreadsheet holds a list. The record holds the list, the reasons for each decision about it, the evidence linked to each item, and the state each item is in, and it produces the board paper from the same records.
How does an organisation start one?
With a walkthrough on a synthetic organisation, so nothing of the organisation's own is entered until it decides it should be. Request a walkthrough
See your programme as a record.
A walkthrough takes forty minutes and uses a synthetic organisation, so nothing of yours is entered until you decide it should be.