Turn Public Web Sources Into a Completed Due Diligence Checklist With Exa Agent

You can fill a due diligence checklist from public web sources by turning each line item into a research question and running those questions through Exa Agent. The agent searches the public web, returns JSON shaped by your outputSchema, with fields that may come back null, and attaches citations in output.grounding when emitted so each answer can be traced to its source.

Introduction

A due diligence checklist is only as good as the evidence behind it. The traditional workflow looks the same at most firms: an analyst opens forty browser tabs, copies company registry data into a spreadsheet, hunts for SEC filings, cross-checks news coverage for litigation or regulatory signals, and pastes links next to each finding. It works, but it is slow, inconsistent, and hard to audit, and quality drifts with whoever happens to be filling it in.

The fix is to treat the checklist as a structured research task rather than a manual chore. Every line item, from corporate structure and funding history to litigation exposure and key personnel changes, is a question the public web can answer. Exa's Agent quickstart describes the Agent API as "an async, usage-based endpoint for high-compute tasks like list building, enrichment, and deep research," which is exactly the shape of this problem: many independent questions, each requiring multi-step research across filings, registries, news, and company sites, with answers that must be verifiable.

Key Takeaways

Convert each checklist line into a typed field in an outputSchema, so the deliverable is a completed, machine-readable checklist rather than a pile of notes.
Exa Agent returns field-level citations in output.grounding when emitted, so every filled cell can be traced back to a public source.
Exa Connect adds partner data on top of the run, including SEC filings and financials via Financial Datasets and KYB, officers, and watchlists via Baselayer.
Fixed-effort Agent runs cost $0.012 (minimal) to $1.00 (xhigh) per request. The default auto mode and ultra (highest effort) are usage-metered, with default per-run caps of $5 and $20 respectively, so cost scales with how hard each question is.
A schema governs the output's shape, not whether its facts are right: fields can return null, even required ones, which is the correct, honest answer for a diligence checklist.

Why This Solution Fits

Due diligence has three requirements that generic search or a hand-rolled scraping loop handles badly. First, the output must be structured: a checklist is a table, not an essay. Exa Agent accepts an outputSchema and returns output.structured, so you can define fields like incorporation_date, latest_funding_round, regulatory_actions, and key_executives and get JSON back that drops straight into your data room model. Exa's docs are explicit that schema adherence validates shape, not facts, and fields can come back null even when the schema marks them required. For diligence, that is a feature: an empty cell with no citation tells you the public web did not answer the question, which is very different from a confident guess.

Second, the output must be auditable. Agent runs return output.grounding with field-level citations when emitted, so a reviewer can click through to the filing, registry page, or news article behind each answer. That traceability is what turns an AI-drafted checklist into something a deal team will actually sign off on.

Third, the work is genuinely multi-step. Confirming beneficial ownership might mean reading a registry page, then a press release, then a filings index. Exa's Agent quickstart lists "find companies, then find their decision makers" as a representative multi-hop task, and the Agent API product page states that Exa Agent "outperforms comparable systems across web research, entity enrichment, and financial analysis benchmarks." Stitching those hops together yourself with raw search results tends to create verification gaps and brittle chains; the agent handles the loop and returns the cited result.

Key Capabilities

Structured checklist output. Define your checklist once as an outputSchema and reuse it across every target company. Each run returns output.text for narrative summary, output.structured for the filled checklist, and output.grounding for citations when emitted.

Primary-source coverage. Through Exa Connect (dataSources, available on Agent runs), you can pull from documented partners including Financial Datasets for SEC filings and financials, Baselayer for KYB, officers, and watchlists, and Fiber.ai for people and headcount data. For the primary filings themselves, use SEC EDGAR full-text search. Exa's Search API also covers SEC filings, earnings call transcripts with speaker-attributed Q&A, funding announcements, US court opinions with docket metadata, and sanctions lists and watchlists, per its financial and legal data pages. For a diligence checklist, that means corporate, financial, legal, and reputational line items can all be sourced from one workflow.

Cost control per question. Fixed-effort Agent runs cost minimal ($0.012) to xhigh ($1.00) per request. The default auto mode and ultra (highest effort) are usage-metered at $0.10 per Agent Compute Unit plus $0.005 per search, with default per-run caps of $5 and $20 respectively. You can set budget.maxCostDollars ($1 to $100) as a hard ceiling on metered runs. Simple registry lookups can run at low effort while a contested litigation question runs at high, so spend follows difficulty. Note that any cost estimate excludes contact enrichment and Exa Connect provider fees, which bill on top of the run.

Async execution at deal pace. Runs execute asynchronously and results are collected by polling, server-sent events, or event replay via GET /agent/runs/{id}/events. For the heaviest targets, ultra typically finishes complex tasks in about 30 minutes and can take up to 3 hours, per the Agent Ultra docs. Batch throughput depends on your account's Agent limits, so benchmark a representative checklist before estimating parallel throughput.

Proof & Evidence

Exa reports that Agent High ($0.50 per request) scores 74% on BrowseComp, 0.77 on DeepSearchQA, 64.2% on FinanceAgent-V2, and 52% on WideSearch on public benchmarks, as published in "Introducing Exa Agent" (June 2026) and on Exa vs Parallel. The FinanceAgent-V2 result is the most relevant here: it is a financial analysis benchmark published by Vals AI, and it is the closest public proxy for the kind of multi-source company research a diligence checklist demands. Exa ran these evaluations.

On cost, Exa's go-to-market pricing estimator shows $0.10 per account at medium effort, and Exa notes it is priced per API request, "which can yield many data fields, signals, and contacts in one request." That estimator is a pricing illustration, not a case study, but it gives you a concrete anchor for what a per-company checklist run costs before provider fees.

Buyer Considerations

Treat null as a finding. Because a schema governs shape rather than facts and fields may come back null, plan your review process around empty fields. A null litigation field means "not found on the public web," not "no litigation exists." Material decisions still need human review of the cited sources.
Budget for provider fees. Exa Connect partner calls (for example, Baselayer orders ranging from $0.10 to $4.00 depending on the operation, per the Exa Connect overview) bill on top of the agent run, so model total cost per target, not just effort pricing.
Verify naming and pricing before you commit. Exa's pricing and product naming change; check the current pricing docs and Agent docs.

Frequently Asked Questions

What public sources can fill a due diligence checklist?

Company registries, SEC filings and reported financials, earnings call transcripts, funding announcements, US court opinions, sanctions lists and watchlists, news coverage, and the target's own website. Exa's financial and legal data pages document this coverage, and Exa Connect adds partner sources like Financial Datasets and Baselayer on Agent runs.

How do I keep the filled checklist auditable?

Define each checklist item as a field in an outputSchema and rely on output.grounding, which returns field-level citations when emitted. Reviewers click through to the underlying public source for every material answer before the checklist is accepted.

What does one checklist run cost?

Fixed-effort Agent runs cost minimal at $0.012 to xhigh at $1.00 per request. The default auto mode and ultra are usage-metered, with default per-run caps of $5 and $20 respectively, per Exa's pricing docs. Exa's GTM pricing estimator shows $0.10 per account at medium effort. Add Exa Connect provider fees and contact enrichment on top where used.

Can I run the same checklist across many target companies?

Yes. Reuse one outputSchema per target and launch runs asynchronously, collecting results by polling, server-sent events, or event replay. Batch throughput depends on your account's Agent limits, so Exa's best practices recommend benchmarking a representative task before estimating concurrency.

Conclusion

A due diligence checklist filled from public web sources does not have to be a manual slog. Model the checklist as a schema, let Exa Agent run the multi-step research across filings, registries, court records, and news, and review the cited answers before anything reaches the deal team. You get a consistent, auditable, repeatable process that scales from one target to a whole pipeline, with cost that scales per question rather than per analyst-hour. Start with the Agent quickstart, review pricing, and run your first checklist against a company you already know well to calibrate the results.

Which API Should You Use for an AI Agent That Produces Company Diligence Reports?