# Lookups

Look up a **domain, keyword, company, person** (email, GitHub, social handle) or **app**, the way you would in an SEO or traffic tool, but across search, AI answers, company records, reviews, social, e-commerce and your own data, on one page. Every block has a price before it runs.

You can use lookups on their own. An account that only wants to look things up needs no product, no connected data and no setup. Connect your own data later and the "you and them" block appears, along with the follow-up actions (add to people, attach as evidence, draft outreach).

## Two modes

Both come with every plan and are billed the same way.

**Lite: lay it all out.** One page of blocks. Click a block to run it. Good when you know what you want to see.

**Pro: let the AI find out.** A full-screen investigation with the coworker. You ask a question, it runs lookups and shows what it finds on a board next to the conversation. For a big question it splits the work into parallel subagents; you see progress and results. It asks before spending more than 50 credits on data.

## How a lookup runs

1. `lookup_start` with the kind and input. It cleans up the input, lines a company up with its official record, and returns a price per block. Nothing is fetched or charged, except that a few companies cost a little to line up: you're told the price and asked first. Starting the same input again within 10 minutes, before anything has run, returns the same lookup (`reused` is true). Notes and reasons come back in English unless you pass `locale`.
2. `lookup_run_blocks` with the blocks you want, or with none to run the default set. `estimateOnly` shows the price without running. Credits are reserved up front and returned if a block fails.
3. `get_lookup` until the status is `done`. Blocks arrive one at a time.
4. `list_lookups` and `get_query_home` show your history; `rerun_lookup` fetches again and keeps the old snapshot.

The same object looked up again within a block's cache time costs nothing. Clicking twice doesn't charge twice.

## Blocks

Identity, company registry, website, what AI answers say about it, traffic, search, rivals, ads, community and social voice, social accounts, influencers, software reviews, apps, e-commerce listings, jobs and funding, risk, government open data, prediction markets, your own traffic, you and them, your custom data sources, and industry data where your industry has a deep pack.

A block with no data says so in one line: what's missing and why. If a source is temporarily unavailable it says that, and you aren't charged. You never get a row of zeros. A few blocks are charged even when nothing is found, because the source charges us either way; the quote says so.

## Price

Each block is charged at actual usage, with a minimum of 1 credit, and the quote is an upper limit. Your own data (you and them, your traffic, custom data sources) uses no data credits. See [Credits and billing](credits.md).

## Five-step deep dive

Every object can go deeper: a product, a lookup, a theme, a person or an AI claim. `get_deep_dive` walks five steps and tells you where you are:

1. **Focus**: what you want to win.
2. **Status**: how often it's mentioned and how positively.
3. **Trace**: what drags it down and which sources feed the AI answers.
4. **Changes**: what to change, in priority order. Each one can become a draft.
5. **Simulate**: estimate the effect before publishing. Simulations are estimates, not forecasts.

Each step comes with the next action to call.
