> ## Documentation Index
> Fetch the complete documentation index at: https://docs.salestools.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Commission AI Deep Research on Companies and People

> Set the research recipe Sally runs before writing — depth tier, language, custom prompt — and browse every completed brief in a searchable library.

Before Sally drafts a single message, she reads. Deep research is where you tell her what to read — the public sources, news beats, hiring signals, and career history that go into a one-page brief she cites in every outbound. The result is outreach that sounds like you did the homework, because she did.

Deep Research lives in **AI Setup** under the Research section of the left catalog. You'll find two entries:

* **Deep Research** — research about a *company* (the account).
* **Deep Research · People** — research about a *person* (the individual in the seat).

Both use the same engine; they differ only in what unit they read about.

## What you can do

* **Commission a research run** against any company or person using a recipe built from your ICP, persona, and company profile.
* **Find prospects from a profile** — instead of pasting IDs manually, pick your ICP or persona and Sally pulls matching accounts and contacts for you to select.
* **Choose a depth tier** from Quick through Ultra Max, trading thoroughness against credit cost.
* **Pick the report language** — over 30 languages supported for company research runs.
* **Write a custom prompt** with merge variables so Sally researches exactly the angles you care about.
* **Browse every completed brief** in a searchable library, filterable by ICP, persona, company profile, or agent.
* **Ask questions across all your research** in plain English and receive an AI-synthesized answer drawn from matching briefs.
* **Read the full report** for any account or person, including the source links Sally visited.

## How to run account research

<Steps>
  <Step title="Open Deep Research">
    In **AI Setup**, click **Deep Research** under the Research section. You land on the **Research** sub-tab, which shows your current recipe, run settings, and any runs from the last 24 hours.
  </Step>

  <Step title="Open the run drawer">
    Click **Run research** in the top-right corner. A side drawer slides in where you build the run.
  </Step>

  <Step title="Name the run and set the recipe">
    Give the run a descriptive **Name** — it appears in History and on the run card. Then set the recipe:

    * **ICP profile** — defines the kind of account Sally reads about.
    * **Persona profile** — the role she is writing for, which shapes which signals matter most.
    * **My Company** — the lens she researches and writes from.
  </Step>

  <Step title="Choose which accounts to research">
    Click **Find prospects** to search your selected ICP and persona for matching accounts, then tick the ones you want to include. Each company run takes up to 10 accounts. You can also paste prospect IDs directly if you already have them.

    <Note>If **Find prospects** returns no results, your ICP or persona may have no targeting criteria set yet. Add attributes on the ICP or Persona tab and try again.</Note>
  </Step>

  <Step title="Set depth, language, and prompt">
    Pick a **Depth** tier (see the table below), select the **Language** for the report, and optionally edit the **Custom prompt**. The prompt supports `{{company_name}}` and `{{company_domain}}` merge variables.
  </Step>

  <Step title="Run it">
    Click **Run research**. Sally queues the run and it appears under **Recent runs** with a Pending status. Reload the page once she has had time to work — a completed run shows a green status dot.
  </Step>
</Steps>

### Depth tiers and credit cost

Depth controls how thorough each brief is, and that determines the credit cost per account researched.

| Tier          | Credits per account | What it does                                                                   |
| ------------- | ------------------- | ------------------------------------------------------------------------------ |
| **Quick**     | 10 cr               | Surface-level scan — one fast pass across the main sources. Lowest cost.       |
| **Standard**  | 10 cr               | A focused pass across the primary sources with better signal extraction.       |
| **Pro**       | 17 cr               | Multi-source pass — broader coverage with cross-checks across sources.         |
| **Max**       | 25 cr               | Deep dive — many sources including news, careers pages, and filings.           |
| **Ultra Max** | 25 cr               | Exhaustive — the most thorough pass Sally runs, across every available source. |

<Tip>The credit cost shown in **Run settings** updates live as you select a depth tier, so you always see exactly what a run will cost before you commission it.</Tip>

<Note>Sally deduplicates credits automatically. If a completed brief already exists for the same company account, she returns that brief instead of running a new one and spending credits again. When you enroll multiple people from the same company in an agent, only one company brief is run and shared across all of them.</Note>

## How to research a person

<Steps>
  <Step title="Open Deep Research · People">
    In **AI Setup**, click **Deep Research · People**. This is a single scrolling screen — the recipe, the feed of people Sally has researched, and the prompt are all on one page.
  </Step>

  <Step title="Open the run drawer and build the run">
    Click **Run research**, then build the run the same way as an account run: name it, pick a **persona**, choose people with **Find prospects** or paste IDs directly, and set the **Depth** tier.

    <Note>Person runs focus on the individual, not their company — covering LinkedIn history, content they have published, mutual connections, and recent career moves. Person runs do not take a language setting.</Note>
  </Step>

  <Step title="Watch the feed update">
    Newly commissioned people appear in the feed with a live status that flips from Pending to Completed automatically — no reload needed.
  </Step>
</Steps>

## How to browse and search results

<Steps>
  <Step title="Open the Results tab">
    On the **Deep Research** screen, click the **Results** sub-tab. It lists every account you have researched, grouped by company, with a status dot and timestamp for each run.
  </Step>

  <Step title="Filter the list">
    Narrow the results by **name**, **ICP**, **Persona**, **Company** profile, or **Agent**, and sort by **Date** or **Company** to find the brief you need quickly.
  </Step>

  <Step title="Ask across your research">
    Use the natural-language search box at the top to ask a question — for example, *"companies in fintech with recent Series B funding"* or *"SaaS tools with AI features."* Sally runs a semantic search across every brief and writes an **AI Synthesis** answer, with matching briefs listed below as cards showing a relevance score.
  </Step>

  <Step title="Open a full report">
    Click any account card or search result to read the full report, including the list of source URLs Sally visited during the run. A run that is still in progress or has failed will tell you its status instead of showing a report.
  </Step>
</Steps>

## How to customize the research prompt

<Steps>
  <Step title="Open the Prompt sub-tab">
    On the **Deep Research** screen, click **Prompt**. This is the standing instruction Sally follows for every research run in this scope.
  </Step>

  <Step title="Edit, save, or reset">
    Edit the template in plain English. Use merge variables to personalize each run:

    * **Company runs**: `{{company_name}}`, `{{company_domain}}`
    * **Person runs**: `{{person_name}}`, `{{person_title}}`, `{{person_email}}`, `{{person_linkedin}}`, `{{company}}`

    Click **Save** when you are happy with the prompt. Click **Reset** to discard unsaved edits, or **Restore default** to load Salestools' recommended template back into the editor — then save to keep it. Use **Preview** to see the prompt rendered with sample data before committing.
  </Step>
</Steps>

<Warning>The prompt you set on the Prompt sub-tab and the one inside the Run drawer share the same recipe. Editing in one place will not silently overwrite an edit you made in the other — but if you have both open, save intentionally so the version you want is the one that sticks.</Warning>

<Tip>Use `{{company_name}}` and `{{company_domain}}` to make the research prompt feel bespoke for each account. For example: *"Research {{company_name}} ({{company_domain}}) with a focus on their current RevOps toolchain and any signals of a CRM migration."* Sally substitutes the real values at run time.</Tip>

## FAQ

<AccordionGroup>
  <Accordion title="How long does a research run take?">
    Runs are queued and processed in the background, so they are not instant. Deeper tiers read more sources and take longer. On **Deep Research · People**, the feed refreshes automatically as each person's run completes. For company runs, reload the page to pull in the latest statuses.
  </Accordion>

  <Accordion title="Will I be charged twice for the same account?">
    No. Sally reuses recent research automatically. If a completed brief already exists for an account, she returns that brief rather than re-running. This deduplication also applies when multiple contacts from the same company are enrolled in the same agent.
  </Accordion>

  <Accordion title="What's the difference between company and person research?">
    Company research reads about the account — news, hiring activity, tech stack, and filings. Person research reads about the individual — their LinkedIn history, content they have published, and recent career moves. Use **Deep Research** for companies and **Deep Research · People** for people.
  </Accordion>

  <Accordion title="A run failed — what should I do?">
    Failed runs show a failed status badge and have no report. The most common cause is missing prospect data such as an account with no domain attached. Re-run using **Find prospects** to pick the account cleanly, which carries all the data the engine needs.
  </Accordion>

  <Accordion title="I can't run research — why?">
    Running and clearing research requires the deep-research permission on your account. If you see a "You do not have permission to run deep research" message, ask your workspace admin to grant it.
  </Accordion>
</AccordionGroup>

## Related

<CardGroup cols={2}>
  <Card title="ICP Profiles" icon="bullseye" href="/ai-library/icp-profiles">
    Define the accounts research targets — industry, size, region, and signals.
  </Card>

  <Card title="Persona Profiles" icon="user" href="/ai-library/persona-profiles">
    Bucket job titles by role so research frames the right buyer angle.
  </Card>

  <Card title="Company Profiles" icon="building" href="/ai-library/company-profiles">
    The lens Sally researches and writes from on every outbound.
  </Card>

  <Card title="AI SDR Builder" icon="robot" href="/ai-agents/ai-sdr-builder">
    Put research to work inside an automated outbound agent.
  </Card>
</CardGroup>