A public affairs firm sells judgement about people and positions. Most of that judgement lives in the heads of a few senior people and in a decade of documents nobody has time to reread.
Who matters on this issue, what they have said before, what the client has already committed to in writing, and what changed last week. That works until the environment moves faster than the people can. A reshuffle changes every stakeholder map at once. A new consultation lands with a three-week deadline and the firm’s best answer to a similar question sits in a response written four years ago by someone who has since left. A client asks what a particular committee chair has said about their sector, and the answer takes an afternoon.
The firms that come out ahead will not be the ones with the best AI tool. They will be the ones whose institutional knowledge is finally in a form that can be used.
Connected, current, and answerable by anyone in the firm, not just the partner who was there.
What stays expensive today
Stakeholder maps are rebuilt, not maintained
Maps are made for a project, go into a slide, and drift out of date the day they are delivered. When the political landscape shifts, every map for every client needs doing again, by hand, under time pressure.
Monitoring produces lists, not understanding
A daily feed says what was published, debated, or tabled. It does not say which of the firm’s client issues it touches, whether a position has shifted, or which stakeholder has entered a conversation they were not in before.
The archive cannot be asked a question
Consultation responses, briefings, position papers, and meeting notes from ten years of work hold the firm’s accumulated view on almost everything. They are stored as files, which means they can be searched for words but not for positions, arguments, or who said what.
What changes
AIP is a company memory: a knowledge graph of the people, organisations, positions, documents, and conversations a firm works with, assembled as a byproduct of the work rather than as a filing task. On top of it sit agents that do defined pieces of the job. For a public affairs firm, the pieces look like this.
The stakeholder map that keeps itself current
Every mention of a person in a speech, a committee transcript, a news article, or an internal meeting note attaches to that person in the graph, with the date and the source. When a minister moves, the map does not need redrawing; the relationships are already recorded, and the question becomes “who now holds the levers this client depends on”, which the memory can answer. We built this shape for an advisory firm working on public affairs in a regulated sector, where the central question was who influences whom on an issue and through whom a client can reach them.
A legislature’s day in the inbox by morning
Bills, amendments, hearings, and statements are natively graph-shaped: members, committees, parties, and sectors, all cross-referenced. Structured that way, both chambers’ daily output becomes a focused briefing for each client’s issues before the morning meeting, with positions tracked over time rather than reported as isolated quotes. The parliamentary intelligence work we have published shows the shape; the difference is doing it on the firm’s own issues, every day, with the results feeding the stakeholder map above.
An archive that answers
A decade of consultation responses becomes a set of positions the firm has already taken, connected by topic, client, and date. When a new consultation opens, the first draft starts from what the firm has argued before and flags where its own past positions disagree with each other. That is a drafting advantage, and it is also a risk control.
A firm that can see its own history does not contradict itself in public.
Reasoning kept alongside decisions
A briefing records what the firm advised. The graph also has somewhere to put why: the alternatives considered, the concern that shaped the recommendation, the line a client asked to remove. Six months later, when the same question comes back, the reasoning is there. Why this matters more than the documents themselves: knowledge graphs for communications teams.
What it refuses to do
Public affairs runs on confidentiality, so the rules matter as much as the capability. Each client’s material lives in its own space; nothing from one client’s work is visible from another’s. Every claim in the memory carries its source, so an answer can be checked rather than trusted. A figure or a quote that is not in the record is shown as a gap, not invented. And the memory makes the firm’s judgement faster to apply; it does not replace it.
The firm this produces
A junior consultant on their second week can ask the firm what it knows about an issue and get the answer the senior partner would have given, with the sources attached. A reshuffle becomes a morning’s work rather than a fortnight’s. New business conversations start from what the firm has already done in that sector, not from a blank page. And the firm’s knowledge stops walking out of the door when people do, which is the quiet reason this is worth doing: the thing being built is an asset the firm owns, and it is worth more every year it runs.
Where to start
Not with a platform decision. With one recurring task that a senior person currently does by hand: the monitoring digest, the stakeholder remap, the consultation first draft. We investigate that one task over ten working days, test it against the firm’s real material, and say plainly whether it is worth building. If it is not, that is also a result.
Introducing a client or a colleague is welcome; they can join the first conversation.
Related: What is a company memory? · AIP · AIP for other kinds of organisation