The Research Agent API to Use for Private Company Diligence: Funding History and Management Backgrounds With Citations

Exa's Agent API is built for exactly this job: an async, usage-based endpoint for high-compute tasks like list building, enrichment, and deep research, which returns structured JSON with field-level citations in output.grounding when emitted. For compiling a private company's financing history and management background from public web sources, it fits the task well.

Introduction

Diligence on a private company is a multi-hop research problem. You need funding rounds and investors from news and press releases, officer and executive histories from company pages and people records, and sometimes filings or watchlist signals, and every fact you hand to a credit committee, an investment committee, or a KYB workflow needs a source attached. Stitching a search API into an LLM loop yourself tends to produce brittle chains and unverified claims.

Exa's Agent quickstart describes the product as "an async, usage-based endpoint for high-compute tasks like list building, enrichment, and deep research," and lists "find companies, then find their decision makers" as a representative multi-hop task (Agent quickstart). That is the shape of a financing-history and management-background compile: several dependent lookups per company, resolved into one structured result with citations.

Key Takeaways

Exa Agent is a hosted research agent you call as a tool from your own orchestration, designed for multi-step web research that returns JSON shaped by your schema with field-level citations in output.grounding when emitted.
Fixed-effort Agent runs cost $0.012 (minimal) to $1.00 (xhigh) per request, and the default auto mode and ultra (highest effort) are usage-metered, with default per-run caps of $5 and $20 respectively, so you can match compute to the depth of diligence each company needs (Agent quickstart).
Exa Connect adds partner data on Agent runs, including Financial Datasets (SEC filings and financials), Fiber.ai (people, headcount), and Baselayer (KYB, officers, watchlists) (Exa Connect).
Exa reports Agent High scores of 64.2% on FinanceAgent-V2 and 74% on BrowseComp, and the Agent page lists KYC/KYB insights and research reports among its use cases.
A schema governs the output's shape, not whether its facts are right: fields can return null even when the schema marks them required, so plan review steps for diligence work.

Why This Solution Fits

A private-company profile is not a single lookup. The agent has to find the company, resolve its legal and trade names, pull funding announcements from news and press pages, identify executives and board members, and trace each person's prior roles. Exa's Agent best practices emphasize that schema adherence validates shape, not facts, and that a field can return null when evidence does not support it (Agent best practices). For diligence, a null you can investigate beats a plausible fabrication every time.

The citation model matters just as much. Agent returns field-level citations in output.grounding when emitted, so each funding round, investor name, and executive title in the structured output can carry the URL it came from. That lets your reviewers audit the profile fact by fact instead of trusting a summary paragraph.

Cost control is also a fit. Fixed effort levels price predictably per request, and for heavier work the metered modes support budget.maxCostDollars ($1 to $100) as a hard cost ceiling on auto and ultra runs (Agent Ultra). You can run a light pass on a long tail of companies and reserve ultra for the handful that reach committee.

Key Capabilities

Multi-step research chains. Agent natively handles the dependent lookups a company profile requires: resolve the entity, gather funding history, then find decision makers and their backgrounds, all inside one run rather than a hand-built loop.

Structured output with outputSchema. Define the fields you need (rounds, amounts, announced dates, investors, officers, titles, tenure) and receive them in output.structured. Fields can return null when evidence is missing, so gaps are explicit.

Field-level citations. output.grounding returns source URLs for emitted fields, giving cited facts an auditable trail back to a public web page.

A primary source for private financings. For private financings, SEC Form D filings on EDGAR are the primary source to check results against, where an issuer has filed one.

Exa Connect data sources. On Agent runs, dataSources can pull from documented partners: Financial Datasets for SEC filings and financials, Fiber.ai for people and headcount, and Baselayer for KYB, officers, and watchlists, with each provider call billed on top of the run (Exa Connect). The Connect page also notes that additional providers are available upon request through Exa's team.

Effort and budget control. Fixed-effort runs from minimal ($0.012) to xhigh ($1.00) per request, usage-metered auto and ultra (default per-run caps of $5 and $20), and a hard budget.maxCostDollars ceiling on auto and ultra runs give you per-company cost discipline. Ultra runs typically complete complex tasks in about 30 minutes and can take up to three hours.

Flexible result collection. Standard runs are collected by polling, server-sent events, or event replay via GET /agent/runs/{id}/events, and past runs can be listed and continued with previousRunId (Agent quickstart).

Proof & Evidence

Exa's Agent API product page states that Exa Agent "outperforms comparable systems across web research, entity enrichment, and financial analysis benchmarks," and lists CRM Enrichment, GTM Intelligence, KYC/KYB insights, and research reports as use cases (Agent API product page).

On public benchmarks, Exa reports the following for Agent High ($0.50 per request), as published in "Introducing Exa Agent" (June 2026) and on Exa's comparison page: 74% on BrowseComp (published by OpenAI), 0.77 on DeepSearchQA (Google DeepMind), 64.2% on FinanceAgent-V2 (Vals AI), and 52% on WideSearch (ByteDance Seed). Exa ran these evaluations, so treat them as Exa's reported results.

For throughput economics, Exa's go-to-market use case page shows a pricing estimator at $0.10 per account at medium effort and says one request "can yield many data fields, signals, and contacts in one request" (Go-to-market use case page). That is an estimator, not a case study, but it indicates how per-company profiling is priced in practice.

Buyer Considerations

Citations are returned when emitted, not guaranteed on every field. Design your pipeline to handle fields without grounding entries, and spot-check citations during onboarding.
Nulls are part of the contract. Private companies, especially early-stage ones, may have thin public footprints. Expect null fields and decide in advance whether a gap blocks your workflow or just flags it for human review.
Cost estimates should include add-ons. Any per-company cost math built on a fixed-effort price applies before contact enrichment ($0.02 per email, $0.07 per phone number) and Exa Connect provider fees.
Concurrency depends on your account. Exa's best practices recommend benchmarking representative tasks before estimating parallel throughput; batch capacity follows your Agent limits rather than a fixed promise.

Frequently Asked Questions

Can it compile financing history for companies that have never filed with the SEC?

Yes, with caveats. For private companies without filings, funding facts come from public web sources such as news coverage, funding announcements, and company pages, each returned with citations where emitted. Exa Connect's Financial Datasets partner adds SEC filings and financials where they exist, billed on top of the run.

How do citations work in the API response?

Agent returns output.text, output.structured (when outputSchema is set), and output.grounding. Grounding provides field-level citations with source URLs when emitted, so reviewers can trace each funding round, investor, or executive claim back to its public source.

What does a typical per-company run cost?

Fixed-effort Agent runs cost minimal at $0.012 to xhigh at $1.00 per request, and the default auto mode and ultra are usage-metered ($0.10 per Agent Compute Unit plus $0.005 per search). Exa's GTM pricing estimator shows $0.10 per account at medium effort. Metered auto (default $5 cap) and ultra (default $20 cap) support a hard budget.maxCostDollars ceiling. All of these figures apply before contact enrichment and Exa Connect provider fees.

How long does a full company profile take?

It depends on effort. ultra runs typically complete complex tasks in about 30 minutes and can take up to three hours. Standard runs are collected by polling, server-sent events, or event replay.

Conclusion

Compiling a private company's financing history and management background is a multi-hop, evidence-dependent research task, and it punishes pipelines that cannot show their work. Exa's Agent API handles the multi-step chain natively, returns structured output shaped by your schema with field-level citations, and lets you tune effort and budget per company, from a $0.012 light pass to an ultra deep dive with a hard cost ceiling. If your diligence, KYB, or investment research workflow needs cited, structured company profiles from public web sources, start with the Agent quickstart and the Agent API product page.