How an AI agent can produce a sourced company brief with gaps clearly marked
This workflow gives an AI agent a defined job, a bounded set of records, and a result a person can review. The agent reads the relevant PitchBook context, applies the rules in the prompt, and keeps the source behind every recommendation. It returns a proposed handoff rather than taking consequential actions on its own.
Can an AI agent produce a sourced company brief with gaps clearly marked?
Yes. Start with the scope, date range, decision rules, and fields that identify the right records. The agent can collect the evidence, compare states or sources, mark conflicts and missing data, and organize the result around the outcome above. A reviewer then checks the matches and judgment calls before approving messages, record updates, bookings, purchases, publishing, or other write actions. The guide below shows the records, boundaries, prompt, and handoff needed for this specific workflow.
Set an as-of date
A PitchBook company research agent answers a defined question about named companies. Provide the company identifiers when possible, geography, sector, research date, and the fields that matter, such as funding history, investors, ownership events, leadership, or comparable transactions. The agent should confirm entity identity before comparing anything.
Every figure needs its date, currency, unit, and source record. Missing data should remain blank or marked unavailable. A financing event does not prove current valuation, and an investor listed in an old round does not prove current ownership.
Example starter prompt
Research [company list] in PitchBook as of [date] to answer [question].
Confirm each entity using [identifiers]. Collect only these fields: [field list]. For every figure or event, include date, currency, unit, and the PitchBook record that supports it. Mark missing, stale, or conflicting data.
Return a company-by-company brief and comparison table. Do not estimate missing values or present an investment recommendation.Review entity and time boundaries
Check subsidiaries, former names, and similarly named companies. If an event falls outside the requested period, keep it in background rather than using it as current evidence.
Questions this workflow answers
What do we know about this private company, and which parts of the story are still unverified?
An agent can prepare a research brief that is useful precisely because it does not smooth over missing private-market data. It begins by resolving the entity through legal name, headquarters, website, industry, and known aliases. Subsidiaries, similarly named businesses, and companies that changed names remain separate until an analyst confirms the relationship. That identity check prevents another company’s financing or leadership record from entering the brief.
The research plan follows the decision at hand. A sourcing screen may need recent financing, investors, location, employee signals, and sector fit. A transaction review may need prior ownership events, deal type, announced date, and comparable companies. The agent collects only the fields requested and attaches dates, currencies, units, and record links. If a figure is estimated or calculated, the label and formula sit next to it.
The narrative then distinguishes events from current state. A funding round shows what was announced at that time; it does not establish today’s valuation, cash balance, ownership, or operating performance. An executive shown in an older record may need verification before a meeting. The agent can place stale or conflicting fields in a verification table with the source, as-of date, and question an analyst should answer elsewhere.
Comparisons use the same basis across companies. Revenue ranges, financing amounts, dates, and sectors should not be ranked together if definitions differ. The result includes a company profile, event timeline, comparison table where valid, and source register. An analyst reviews identity and calculations before the brief supports outreach, diligence, or an investment discussion.
The final brief should separate reported data, agent calculations, and unanswered questions. An analyst reviews any calculation before it enters a decision document.
Entity resolution comes first for private companies. The agent should record legal and operating names, former names, headquarters, website, parent or subsidiary relationships, and the identifiers behind every event. A financing attributed to a similarly named company belongs in the rejected-match list. When ownership, valuation, or revenue fields come from different dates, the brief keeps those dates visible rather than presenting them as one current snapshot.