How an AI agent can produce a sourced private-markets research brief with open questions
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 Chronograph 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 private-markets research brief with open questions?
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.
Fix the entities and as-of date
Provide the funds, companies, or investments, the research question, reporting period, currency, and relevant measures. Clarify whether the agent should compare reported values, calculate a derived measure, or only summarize existing records.
Entity names can be ambiguous across vehicles and share classes. Ask the agent to include the Chronograph identifier and context for every item. It should state the period and source behind each number.
Example starter prompt
Use Chronograph to research [funds/investments] as of [date] for this question: [question]. Report in [currency/units] and use these measures: [measures].
For every figure, cite the Chronograph record, period, and whether it is reported or calculated. Show the calculation for derived values. Separate confirmed findings, stale data, missing fields, and questions for the deal or portfolio team. Do not infer values from peers.Review comparability
Check that periods, currencies, valuation dates, and definitions align before comparing investments. A value reported at quarter-end should not be placed beside a current estimate without a label. The agent should explain exclusions and avoid ranking an investment when key measures are missing.
Questions this workflow answers
What can we reliably say about these private investments as of a specific date?
An agent can answer a narrow portfolio question when the entities and measurement rules are fixed first. Provide the fund, company, or investment identifiers, as-of date, reporting period, currency, units, and measures. State whether the job is to summarize reported records, calculate a named measure, or compare selected entities. The agent keeps the Chronograph identifier and source period beside every number.
Private-market records often resist casual comparison. Two vehicles may share a name but represent different holdings or share classes. Valuations may come from different quarter ends. One measure may be gross while another is net, or one may use committed capital while another uses invested capital. The agent checks those definitions and excludes or labels values that cannot be aligned. A missing value remains missing; it is never estimated from peers.
Derived measures include the formula, source inputs, rounding, and date. If the inputs do not share a period, the agent should decline the calculation or show the limitation clearly. Rankings require comparable coverage. An investment with incomplete data should not fall to the bottom merely because the agent lacks a field.
The research brief contains the answer supported by the records, an entity table, calculations, comparability warnings, stale data, and questions for the portfolio or deal team. A human analyst checks source links and decides how the facts affect valuation, risk, or action. The agent does not make an investment recommendation. It creates a repeatable, auditable starting point and preserves what was known at the stated date.
Comparisons should retain the original reporting period, currency, valuation date, and calculation method. If two funds use different definitions for the same-looking measure, the agent places those definitions beside the figures instead of ranking them as though they were equivalent. That caveat is part of the answer, not a footnote to remove for presentation.
Expected handoff
Return the research answer, entity table, source register, calculations, comparability warnings, and open questions. Keep factual findings separate from analyst interpretation and decisions.