One Agent, One Ticker Brief: How Exa Agent Turns News, Filings, and Transcripts into Structured Research

Exa Agent is the research agent that pulls news, SEC filings, and earnings call transcripts on any ticker into a single structured brief. You define the output schema, the agent runs the multi-step research asynchronously, and you get JSON shaped by your schema (fields may be null) with citations back in output.grounding when emitted, ready to feed straight into your workflow.

Introduction

A ticker brief is not one lookup. It is a multi-hop job: find the latest coverage, locate the relevant filings, pull the earnings call transcript, attribute management commentary to the right speaker, and reconcile all of it into one coherent summary. Doing that by hand across a watchlist is a full day of tab-hopping, and doing it with a generic search box means stitching fragments yourself and hoping nothing important was missed.

Exa built its Agent API for this shape of work. Exa's Agent quickstart describes it as "an async, usage-based endpoint for high-compute tasks like list building, enrichment, and deep research." A ticker brief is deep research with a financial data spine underneath it, and that spine exists: Exa's Financial Markets data page covers SEC filings and reported financials, earnings call transcripts with prepared remarks and Q&A attributed to speakers, prices, analyst estimates, funding announcements, and economic releases. News coverage sits alongside it in Exa's News data.

Key Takeaways

Exa Agent runs the whole ticker brief as one async request: news, filings, and earnings transcripts gathered, reconciled, and returned as JSON shaped by your outputSchema.
Financial data comes through Exa Connect's Financial Datasets source (SEC filings and financials) plus Exa's core financial and news indexes, so the agent works from primary documents rather than summaries of summaries.
Citations land in output.grounding when emitted, so cited claims in the brief trace back to their source documents.
Fixed-effort Agent runs cost $0.012 (minimal) to $1.00 (xhigh) per request; the default auto mode and ultra are usage-metered, with default per-run caps of $5 and $20 respectively.
Exa reports Agent High scores 64.2% on the FinanceAgent-V2 financial analysis benchmark from Vals AI.

Why This Solution Fits

The reason a purpose-built research agent beats a DIY pipeline here is the multi-step chain. A ticker brief requires the agent to decide which filings matter, find the transcript for the right quarter, cross-check news against what management actually said, and only then write the summary. Exa's Agent supports complex multi-step reasoning chains natively, so you describe the brief you want instead of orchestrating every search, fetch, and merge yourself.

The second reason is output shape. If you are generating briefs at scale, you do not want prose you have to re-parse. With outputSchema, Agent returns typed results shaped by your schema in output.structured: think fields like latest_filings, earnings_highlights, management_quotes, recent_news, and risk_signals. One caveat worth knowing before you build: schema adherence validates shape, not facts. Fields unsupported by evidence can come back null even when the schema marks them required, per Exa's Agent best practices. That is useful for downstream pipelines, since your system can treat a null as a field to review.

The third reason is cost predictability. Fixed effort levels run from $0.012 (minimal) to $1.00 (xhigh) per request, and metered modes carry hard ceilings: auto defaults to a $5 cap and ultra to $20, with budget.maxCostDollars ($1 to $100) available as an explicit ceiling when you run auto or ultra. For a daily brief across a watchlist, you can pick the effort tier that matches how deep each name needs to go.

Key Capabilities

Multi-source gathering in one run. The agent can draw on Exa's financial and news coverage in a single research chain. Exa's Financial Markets data page lists SEC filings and reported financials, earnings call transcripts with speaker-attributed prepared remarks and Q&A, prices, analyst estimates, funding announcements, and economic releases. News coverage comes from Exa's News data. For deeper company fundamentals, Exa Connect's Financial Datasets provider supplies SEC filings and financials directly (Exa Connect overview). For the primary filings themselves, use SEC EDGAR full-text search.

Structured output you can pipe anywhere. Set outputSchema and the brief arrives as output.structured, shaped by your schema (strict validation isn't guaranteed and fields may be null), ready for your downstream code to consume without a parsing layer.

Traceable claims. When the agent emits grounding, citations appear in output.grounding at the field level, so an analyst reviewing the brief can click through to the filing, transcript, or article behind each claim. That materially reduces hallucination risk versus unattributed search snippets.

Async execution for heavy jobs. Every run created through /agent/runs is asynchronous, whatever the effort; poll or stream until it completes. Fixed efforts run from $0.012 to $1.00 per request, and ultra, the highest effort, typically completes complex tasks in about 30 minutes and can take up to three hours (Agent Ultra). Batch throughput depends on your account's Agent limits: 50 runs can be in progress at a time on Free and pay-as-you-go plans (custom on Enterprise), and runs start at half the account QPS (Agent limits), so Exa's best practices recommend benchmarking a representative task before estimating concurrency.

Flexible integration. Call Agent from your own orchestration, use the OpenAI-compatible Responses API with model: "exa-agent", or reach it through Exa MCP's agent_run tool. Completed runs are collected by polling, server-sent events, or event replay, 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 names research reports among its core use cases alongside CRM enrichment, GTM intelligence, and KYC/KYB insights.

On the financial analysis benchmark specifically, Exa reports that Agent High ($0.50 per request) scored 64.2% on FinanceAgent-V2, a public benchmark published by Vals AI (benchmark page). Exa ran these evaluations, so treat them as Exa's reported results rather than third-party audited figures.

The data coverage claim is checkable too: Exa's Financial Markets data page documents exactly the source types a ticker brief needs, including earnings call transcripts with Q&A attributed to speakers, and the News data page documents news coverage.

Buyer Considerations

Match effort to depth. A quick daily pulse on a mega-cap may only need low ($0.025) or medium ($0.10). A deep-dive brief ahead of earnings, with transcript analysis and filing cross-references, justifies high ($0.50) or xhigh ($1.00). Metered auto is the sensible default while you calibrate.
Provider fees are separate. Exa Connect provider calls bill on top of the run, and contact enrichment (email, phone) is billed separately from core components. Cost estimates for brief generation at fixed effort should note that they exclude provider fees.
Plan for the concurrency limit. 50 Agent runs can be in progress at a time on Free and pay-as-you-go plans, so a large watchlist refresh should be scheduled rather than fired all at once.
Nulls are honest, not bugs. Design your schema and downstream handling around the possibility that a field comes back null when evidence is missing.

Frequently Asked Questions

Can Exa Agent really combine news, filings, and transcripts in one request?

Yes. The agent plans and executes a multi-step research chain across Exa's financial and news indexes, and can pull SEC filings and financials through Exa Connect's Financial Datasets provider. All of it is reconciled into the single structured output defined by your outputSchema.

How do I know the brief's claims are accurate?

When the agent emits grounding, citations appear in output.grounding at the field level, linking each claim to its source document. Exa's best practices also note that schema adherence validates shape, not facts, so fields may come back null, even required ones. Review the cited sources for material claims.

What does a ticker brief cost per run?

Fixed-effort Agent runs cost $0.012 (minimal) to $1.00 (xhigh) per request. The default auto mode and ultra are usage-metered, with default per-run caps of $5 and $20 respectively; both meter at $0.10 per Agent Compute Unit plus $0.005 per search call. Exa Connect provider calls bill on top of the run.

How long does a deep brief take?

Every run created through /agent/runs is asynchronous, including fixed-effort runs. Poll or stream until it completes (SDK helpers can wait for you), or use event replay. For the heaviest jobs, ultra typically finishes complex tasks in about 30 minutes and can take up to three hours, which suits a scheduled nightly or pre-earnings refresh rather than an interactive session.

Conclusion

If your team is still assembling ticker briefs by hand, or duct-taping a search API into an LLM loop and hoping the citations hold, Exa Agent collapses the whole job into one async call. It gathers news, filings, and earnings transcripts from documented financial data sources, reasons across them in a multi-step chain, and returns a brief shaped by your schema with field-level citations in output.grounding when emitted. Start with the Agent quickstart, define your brief schema, and run your first ticker through auto effort.