Where does the enriched data actually come from?

From the live pages the agent opens in your own browser at run time: the company’s official site and about page, its LinkedIn company page, the Crunchbase profile, recent news, and for people the current LinkedIn page. Every field it fills carries the exact source URL it was read from, so each value is auditable rather than a black-box guess.

How is this different from buying a static enrichment database?

A purchased database is a periodic snapshot: the headcount and funding are whatever they were at the last refresh, there’s no source attached to a field, and you pay per credit to look anything up. This agent reads the live page the moment it runs, attaches a source to every cell, uses access you already hold, and lets you re-verify a column as often as you want without a per-match charge.

How is this different from your LinkedIn email finder page?

The LinkedIn email finder is purpose-built for one thing: source-backed work emails. This page is the general enrichment workflow, filling any field on a company (headcount, funding, HQ, stack) or a person (current title, company, location) from live pages with a source per cell. If verified work email is what you need, start on the email finder page; for everything else, start here.

Is writing back to my CRM safe?

Nothing is written without your sign-off. The agent prepares a field-level diff with the old value, the new value, and the source for each, then queues it for approval. You see exactly what would change before anything is saved, and writes run through your own logged-in session and permissions, so they look like any other edit you’d make by hand.

Which AI agents can run enrichment through Actionbook?

Actionbook works with ChatGPT, Claude, Gemini, Perplexity, OpenClaw, Hermes, and any MCP-compatible client. Connect the one you already use and it can read your list, visit live sources in your own browser, fill the missing fields with a source per cell, and queue the writeback for your review.

Data enrichment AI agent

Every cell filled from the live web

Paste a list. Your AI fills the missing fields from live sources in your browser, a source per cell, and you approve the writeback.

Start free See example prompts

docs.google.com/spreadsheets/accounts

Reading your listChatGPT, Claude or Gemini

“Fill headcount, funding, and CEO for these accounts, one source per cell, and flag anything that conflicts.”

CompanyHeadcountFundingCEO

Northwind

Meridian Labs

Cobalt Health

Waiting to read the live sources…Writeback: you

[ 01 ]How it works

From a pasted list to a reviewed writeback

Connect the AI you already use, whether ChatGPT, Claude, or Gemini, to your browser. From a single prompt it runs four steps, and stops before anything is written.

1
Hand it the list you already have
Point your AI at a Google Sheet, a CSV, or a saved CRM view of company names, domains, or people. It splits the file into one row per record and reads the columns that are still blank.
2
Say which fields to fill
Name the gaps and the rules: headcount, funding, HQ for companies; current title, company, location for people. Skip anything already filled, flag rows that disagree.
3
It reads the live pages a person would open
In your own logged-in browser it opens the official site, the LinkedIn company and about pages, the Crunchbase profile, and recent news, then writes each field back with the exact page it came from.
A static database hands you a value with no source attached.
4
It queues the writeback for your review
Filled fields, flagged conflicts, and a source per cell land as a field-level diff. You approve before anything is written to the sheet or the CRM.

[ 02 ]You approve every writeback

salesforce.com/contact

Awaiting approval

Your AI wants to update 2 fields on Jane Doe’s Salesforce contact.

salesforce.com · contact

Jane Doe

Title: Manager VP, Operations

Location: — Austin, TX

Source: linkedin.com/in/janedoe

2 fields

Enrich people, then write back on your approval

For a list of people the agent reads the live profile for current title, company, and location, attaches a source to each, and prepares the CRM update as a field-level diff. The writeback is a real action in your system of record, so, like every consequential Actionbook step, it waits for your approval first.

Current, not cached: Title and company come from the live profile the agent opens at run time.
Field-level diff: You see old value, new value, and the source for each change before it saves.
Your session, your rules: Writes run through your own logged-in CRM, so they look like any edit you’d make.

Start free

[ 03 ]Who it's for

Built for the teams that own CRM hygiene

One enrichment workflow, whether you’re cleaning accounts or building a list from scratch.

RevOps managers

Keep the CRM honest: re-verify headcount and funding on your target accounts and queue the writeback as a diff you approve before it touches a field.

Source per cell

CompanyHeadcountFunding

Northwind210Series B

Meridian Labs48Seed

Cobalt Health1.2kSeries D

Founders with a scraped list

Turn a raw export of company names into a clean sheet of HQ, headcount, and funding stage, each cell carrying the page it was read from.

Analysts building market maps

Fill a grid of companies and the people who run them from live sources, flag the ones that were acquired or renamed, and cite every value.

Live pages, not a quarterly snapshot

It opens the official site, LinkedIn, and Crunchbase in your own session and reads headcount and funding off the page today, reaching firmographics a static database refreshes only once a quarter.

Agencies prepping client data

Enrich a client’s contact list with current title, company, and location before handoff, every value sourced from the live profile rather than a year-old snapshot.

[ 04 ]Copy a prompt

Prompts to start enriching

Copy a prompt, paste it into your AI, and Actionbook runs it in your own browser session.

Fill company fields with a source per cell

Open my Google Sheet “Target accounts” and for each of these 40 companies fill the empty Headcount and Funding columns. Read the official site, the LinkedIn company page, and the Crunchbase profile, and put the source URL next to each value.

Enrich a people list

For each person in this list, find their current title and company and their city. Read the live LinkedIn page, not a cached one, and return name, title, company, location, and the source link for every field.

Flag acquired or renamed companies

Go through my HubSpot “Accounts to review” view and flag any row where the company was acquired or renamed. Check the official site and recent news, and add the new name and a source link for each flagged row.

Re-verify before a writeback

For each account in my Salesforce “Top 50” view, confirm today’s headcount from the LinkedIn company page. Where it differs from the CRM by more than 15%, prepare the update and show me the field-level diff before saving anything.

Build a market map grid

I’m mapping 25 companies in this category. For each, read the official site and Crunchbase and fill HQ city, headcount band, funding stage, and the founder’s name, with one source per cell in a new sheet.

Update CRM fields after my review

Take my approved enrichment for these contacts and update the matching Salesforce records for title, company, and location, then stamp Last Enriched to today. Pause and show me each change before you write it.