--- title: "AI for Asset Managers and Allocators | By The People Tech" description: "AI for New York asset managers, allocators and family offices: DDQs, manager research, investor relations and LP reporting, on your own documents." url: https://bythepeople.tech/industries/asset-managers --- [Industries](/industries) Asset managers and allocators “Most of these questions were answered last quarter.” # The answer was written last quarter. Someone just has to _find_ it. By The People Technologies works with asset managers, allocators, funds of funds and family offices in New York on the paper that moves between managers and investors: due diligence questionnaires, manager research, investor relations correspondence and LP reporting. Teams learn to draft from the firm’s own approved answers in a [workshop](/workshops), a [retainer](/retainer) keeps IR, operations and research using the method, and a [forward deployed engineer](/forward-deployed-engineer) can rebuild one process with the people who run it. By [Tim Miner](/about-us/tim-miner), founder and CEO · Published 21 October 2026 · Updated 22 October 2026 What moves between managers and investors 1. As they arrive Due diligence questionnaire The answer the firm already approved Where we usually start 2. Each quarter Manager quarterly letter What changed since the last one 3. Each quarter Capital account statement Figures that tie to the last statement 4. At commitment Side letter What each manager owes you 5. All day Investor email Who needs an answer, and by when The DDQ ## How can AI help answer a _due diligence questionnaire_? A DDQ is assembly. Most of its questions were answered last quarter, in a document somebody could locate, and the work goes on locating it. Answered from a blank page, the same question gets a slightly different answer each quarter, and an allocator’s due diligence team reads for exactly those differences. Used well, the model drafts from the firm’s library of approved answers, says which past answer and which date each draft came from, and marks the questions with no approved answer for a person to write. Investor relations, compliance and operations confirm their own sections. The same method works from the allocator’s side: the manager’s answers checked against its own last DDQ, its letters and its side letter, with every difference cited. Manager research ## How do allocators use AI for _manager research_? An allocator reads statements from every manager in a different format, and a file of side letters recording what each of them owes it. Most of that is reading to find what changed: a new name on the team, a strategy described in slightly different words, a fee term that differs from the side letter. ### Letter against letter Each manager’s quarterly letter set against the last four, with every change in language or team cited to its paragraph. ### Statement against statement The same fields pulled from every capital account statement and tied to the prior quarter’s, with breaks flagged. ### Side letter against practice What each manager agreed to in writing, held in one place and checked when the terms come up. The allocator still decides whether a change matters. What changes is that the reading reaches every manager every quarter rather than the few someone had time for. Investor relations ## What are the best AI use cases for _investor relations_? The two places IR loses the most time are the inbox and the CRM. LP emails sit in inboxes, and the CRM is days behind every meeting, because it is only ever as current as the last person who remembered to update it. In one engagement the correspondence is now read and the contacts, firms and follow-ups become CRM entries; the owner confirms rather than retypes. At one private equity firm, [investor-email triage](/case-studies/investor-email-triage-private-equity) went from days to under five minutes: a flagged investor email becomes reviewable CRM updates, and a person still decides what goes back to the investor. Both are processes a [forward deployed engineer](/forward-deployed-engineer) builds with the IR team that runs them. LP reporting ## How can a team use AI for LP reporting _without a wrong number_? By keeping the model away from the judgment and close to the tie-out. Quarterly LP reporting is reconciliation: the same schedules, the new quarter’s figures, and a tie-out to the prior quarter that has to be right before anything leaves the building. A model reads and reconciles well when every figure it moves is tied to the cell it came from and a person reviews every break. On one [quarterly roll forward](/case-studies/quarterly-roll-forward-private-equity) we documented how the file was really produced, then built the roll and the tie-out with the team that owns it. They review and explain what moved in hours, not days. A model does not mark positions; a valuation is a signed judgment, and it stays with the person who signs it. The first step ## Where should an allocator or family office _start_ with AI? “We have no CTO, and we don’t want a connector that exposes everything.” Which document costs your team the most each quarter? With the tool the firm already has and one document that costs the most, often the DDQ or the quarterly letters. A [workshop](/workshops) can run on mock files written for your functions, so no real investor data has to leave, and nothing is connected to anything until the firm decides it should be. From there, the [retainer](/retainer) carries the method to IR, operations and research, and keeps the firm’s one-page usage policy current as the tools change. **An allocator or manager with tools bought and left, and one whose people use them well** | | Tools bought and left | People who use them well | | ---------------------------- | -------------------------------------------- | ---------------------------------------------------------------------------- | | The DDQ | Answered from a blank page each time | Drafted from approved answers, each tagged with its source and date | | A manager’s quarterly letter | Read when someone has time | Set against the last four, every change cited to its paragraph | | Capital account statements | Rekeyed from each manager’s format | The same fields pulled and tied to last quarter, breaks flagged | | The investor inbox | Emails wait days for the right person | Triaged into reviewable CRM updates; a person decides the reply | | The CRM | As current as the last person who remembered | Entries made from the correspondence; the owner confirms rather than retypes | | The quarterly LP report | The same file rebuilt by hand | A roll and tie-out the team reviews and explains in hours, not days | We have delivered over 150 workshops at more than 100 financial services firms. We are in the Claude Partner Network, Select tier. ## Questions asset managers and allocators ask us ### We’re a family office in New York. Who can help us choose AI tools and train our team? By The People Technologies works with family offices, allocators and asset managers in New York. We start from the tools you already have, teach your team on mock files or your own documents in a workshop, and help write the one-page usage policy. Nothing is connected to your systems until you decide it should be. ### How can allocators use AI for operational due diligence on a manager? Give the model the manager’s DDQ, its previous DDQ, its quarterly letters and your side letter, and ask it to list every difference between them, cited to the page. The ODD team reads that list against the sources and decides which differences need a call to the manager. ### What are the best AI use cases for investor relations at a private equity firm? The investor inbox and the CRM. At one private equity firm, investor-email triage went from days to under five minutes, and in another engagement correspondence becomes CRM entries the owner confirms rather than retypes. DDQ drafting from approved answers is the third. In each, a person still decides what goes to the investor. ### How should portfolio managers at an asset manager use Claude without trusting a wrong number? Give it the source documents rather than asking from memory, make it cite the page or cell for every figure, and let it say “not in the document”. Then check the three numbers that matter against the source before anything moves. A citation makes that check take seconds. ### Our CFO wants to use Claude in Excel for LP reporting and reconciliations. Where do we start? With one file the team rebuilds every quarter, and the tie-out to the prior quarter. Document how the file is really produced first, then have the model do the roll and flag every break for a person to explain. Valuations stay a signed judgment. A forward deployed engineer can build this with the team that owns the file. ## Sources * [Anthropic, “Prompting best practices”, Claude documentation.](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practices) * [Anthropic, “Reduce hallucinations”, Claude documentation.](https://platform.claude.com/docs/en/test-and-evaluate/strengthen-guardrails/reduce-hallucinations) * [Anthropic, Claude Partner Network directory: By The People Technologies, Select tier.](https://partnerhub.claude.com/directory/partner/by-the-people-technologies-partner) ## Read next * [The AI usage policy a desk reads](/guides/ai-usage-policy-investment-firms). What may go in, what may not, and what LPs now ask about it. * [Making AI accurate on investment documents](/guides/ai-accuracy-investment-firms). Why a model misreads an LP letter, and the five things that cut the errors. * [Forward deployed engineers](/forward-deployed-engineer). The engagement behind the investor-email triage. * [AI for hedge funds](/industries/hedge-funds). Research, earnings and compliance from the manager’s side of the table. ## Tell us which document comes round every quarter. Who answers it, and where last quarter’s answer lives. We will say what we would look at first, and if we are the wrong firm for it, we will say so. [Book 30 minutes with Tim](https://calendly.com/timminer-1/chat-with-tim) [hello@bythepeople.tech](mailto:hello@bythepeople.tech)