How to Use ChatGPT for Lead Generation with Live Business Data

TL;DR:

• ChatGPT can only generate real leads when it is connected to live business data. The Firecrawl plugin connects it to Google Maps, website scrapes, Apollo, and FullEnrich.
• The demo ran three prompts to find accountants and financial advisers near Brighton, UK, as prospects for an AI chatbot business.
• A Maps search returned 40 listings in 43 seconds. Firecrawl's Maps tool has no radius option, so ChatGPT checked distance from each listing's coordinates.
• Scraping and screenshotting 10 firm websites took 5m 15s and found no chatbot on any of them. Late-loading widgets can hide from a scrape, so Interact is the stronger check.
• A final prompt pulled each firm's most senior decision maker and work email through Firecrawl Alexandria in 15m 15s. The three prompts took about 21 minutes in total.
• Apollo people search costs 0 credits and a FullEnrich work email costs 20. Accept each provider's terms in the Firecrawl dashboard before ChatGPT can use them.

Most guides on how to use ChatGPT for lead generation stop at prompts for writing outreach. Those guides skip the hard part: finding the real firms near you that are worth emailing, and who to contact at each.

The Firecrawl plugin fixes that by giving ChatGPT live business data. The example business sells AI chatbots to accountants and financial advisers in Brighton, UK. Three prompts take ChatGPT from an empty chat to a qualified list with decision makers' work emails.

ChatGPT on its own can score leads you already have and draft emails, but it cannot reliably build the list. Lead generation means finding real businesses that fit your offer, with someone to contact at each. That needs current directory listings, website checks, and work emails, and ChatGPT has no built-in contact database.

The fix is to connect ChatGPT to a data source through a plugin, an add-on that lets ChatGPT call outside tools during a chat. The Firecrawl plugin gives it web scraping plus Firecrawl Alexandria's structured data providers.

Firecrawl Alexandria is a catalogue of data providers that an AI agent can call through Firecrawl. Providers include Google Maps business search and people-data companies such as Apollo, FullEnrich, and Data Legion. Discovering tools is free, and each call costs the credits listed on the tool.

The providers used in this demo each answer one question about a lead:

You do not have to name a provider in your prompt. Ask ChatGPT to "use the relevant Alexandria providers" and it picks the tools itself, which the Firecrawl Alexandria page lists by category.

Setting up Firecrawl in ChatGPT takes three things: a Firecrawl account, accepted terms for each people-data provider, and the Firecrawl plugin installed in ChatGPT. The free plan works for a first run. Once connected, you call Firecrawl in any chat by typing @Firecrawl before your prompt.

1. Sign up for Firecrawl. The free plan includes 1,000 credits a month and needs no card.
2. Open Alexandria in the Firecrawl dashboard and search for Apollo.
3. Review and accept the provider terms for Apollo, FullEnrich, and any other people-data provider you want ChatGPT to use.
4. Open Plugins in ChatGPT, search for Firecrawl, and select Install.
5. Sign in to Firecrawl and approve the connection when ChatGPT prompts you.
6. Start a new chat and type @Firecrawl followed by your request.

You now have a ChatGPT chat that can call Firecrawl's scraping tools and every Alexandria provider your team has accepted.

The provider terms screen lists each provider with its agreement and an on/off toggle:

Provider terms matter because people data carries usage rules. FullEnrich's terms, for example, bar marketing to personal email addresses, which is why this walkthrough only asks for work emails. The People enrichment in Firecrawl Alexandria launch post covers what each provider adds.

Firecrawl also has a Claude plugin, so the same prompts work in Claude. The official Firecrawl ChatGPT plugin post covers install in more detail.

You do not paste a Firecrawl API key into ChatGPT. The Firecrawl plugin signs in to your Firecrawl account and approves access, and the calls are billed to that account's credits. A free account covers a small run, because the free plan includes 1,000 credits a month.

Firecrawl does offer a keyless tier with no account at all, but it does not cover this workflow. Keyless access covers only a few core tools such as search and scrape. The people-data providers need a signed-in team that has accepted their terms.

Also read: Introducing Firecrawl Keyless: Search, scrape, and interact without an API key.

The free plan returns an error once its monthly credits run out, rather than charging you. Paid plans on the Firecrawl pricing page add more credits if you scale the process up.

Firecrawl's Maps search finds businesses from a text query such as "accountants Brighton UK". Each call returns up to 20 places with name, address, phone, website, rating, review count, and coordinates. In the demo, one prompt returned 40 listings in 43 seconds: 20 accountants and 20 independent financial advisers around Brighton.

The prompt names the categories, the town, the radius, and the data source:

@Firecrawl find accountants and independent financial advisers in Brighton
within a 5 mile radius using google maps

An independent financial adviser, or IFA, is a UK financial adviser who is not tied to one product provider.

ChatGPT's reply listed each firm with its address, phone number, and Google rating. It also linked the Firecrawl Maps search URLs it used, so you can open them and check the source.

Firecrawl's Maps search has no radius parameter. It takes the query text and an optional centre point. ChatGPT centred the search on Brighton and then checked that each listing's coordinates fell within roughly 5 miles.

For a wider area, ask ChatGPT to search each nearby town separately and filter the combined results by distance. Putting "within 20 miles" in a single query does not widen it.

Note: the same Maps provider has a reviews tool that returns up to 60 reviews with the owner's replies, at 5 credits per business. Reviews that complain a firm is slow to respond make a strong hook for a chatbot pitch.

Qualifying a lead means dropping businesses that do not need what you sell. For a chatbot business, a firm that already runs a chat widget is a weak lead. Firecrawl's Scrape loads each website, returns its content, and can take a screenshot, so ChatGPT can look for a chat bubble on the rendered page.

The demo capped the list at the first five firms of each type, 10 in total, to keep the run short:

@Firecrawl scrape the websites of the first 5 ifas and accountants using
screenshots to see if they have chatbots on their site show me the
screenshots if necessary

ChatGPT worked for 5 minutes 15 seconds and found no chatbot on any of the 10 sites. It checked for visible chat bubbles and for widgets resembling Intercom, Drift, Tidio, HubSpot Chat, Crisp, Zendesk, LiveChat, and WhatsApp-style chat. Enquiry forms, phone buttons, and contact links did not count.

ChatGPT also flagged its own blind spot. A chat tool can appear only after a delay, a cookie consent click, or a user action, and a single page load misses it.

Firecrawl Interact closes that gap. Interact opens the site in a cloud browser that can click, scroll, and wait. It costs more than a scrape, so use it only on the sites a scrape could not settle. Scrape is the cheap first pass at 1 credit per page.

A decision maker is the person at a firm who can approve a purchase, usually a director or partner at a small practice. Ask ChatGPT to use "the relevant Alexandria providers" and it searches Apollo, FullEnrich, and Data Legion. It returns each company's most senior person and their work email. In the demo, this step took 15 minutes 15 seconds for 10 firms.

@Firecrawl for each of the companies look up their information using the
relevant Alexandria providers find the most senior decision maker and give
me their work email

ChatGPT returned a table with each company, its senior decision maker, their current role, a work email, and a confidence level. The table below is redacted because it holds real people's names and email addresses.

You do not need to know which provider holds data on which firm. Firecrawl Alexandria lists every provider's tools, and ChatGPT picks the ones that fit, which avoids separate subscriptions to each data vendor.

Order the calls by price. Apollo's people search is free and shows who exists at a company before anything is spent. Pay for a work email only for firms that passed every earlier check. Expect some lookups to come back empty, and check each name against the firm's own website before you rely on it. To look up one person rather than a whole list, see how to find someone's work email in ChatGPT.

Wondering which provider has the most accurate company data? Our company enrichment API benchmark roundup compares Apollo, FullEnrich, Data Legion, and others on company-record accuracy from two public benchmarks. Neither benchmark tests work emails.

Every Firecrawl Alexandria tool lists its price in credits before it runs, and discovery is free. The cheap steps come first in this workflow and the expensive one, buying a work email, comes last. That order means you only pay 20 credits per email for firms that already passed every check.

Prices are as listed in Firecrawl Alexandria on 6 October 2026. ChatGPT chooses the tools for each prompt, so a real run can use more calls than the minimum. Check Firecrawl's billing docs for scrape and Interact pricing.

The demo's own timings show where the minutes go:

The run was split into three prompts so each step's data is visible. One combined prompt can run the same steps end to end.

ChatGPT can turn the research into a personalised cold email, meaning a first message to someone who has not contacted you. The demo stopped at the contact list and suggested two next steps. Connect ChatGPT to Google Sheets to save the list. Then have it draft each email, plus a follow-up for seven to ten days later.

A prompt for that step, not shown in the video, might read:

@Firecrawl for each firm on the list, draft a short cold email to the
decision maker. Mention one specific, verifiable thing about their website
and one small ask. If you mention what having no chatbot might cost them,
frame it as a question, not a claim. Draft a follow-up for 7 days later. Do
not send anything.

Review every draft yourself before sending, and keep sending manual. Each email should say who you are, stay relevant to the firm, and include a way to opt out.

UK rules depend on the prospect's legal form. Under PECR, the UK's electronic marketing rules, limited companies, LLPs, and Scottish partnerships count as corporate subscribers, and the email consent rule does not apply to them. Sole traders and other partnerships count as individuals and need consent. UK GDPR still applies whenever you hold a named person's details. The ICO's business-to-business marketing guidance sets this out.

Note: this is a summary of published guidance, not legal advice. Financial advice firms are regulated, so pitch a chatbot that answers common questions and books consultations. Do not pitch one that gives advice.

The same three steps work for any business that can name a visible gap on a prospect's website. Swap the Google Maps category and the website check, and keep the work email step. The process works in any region the providers cover.

• Web designers: find local firms whose sites have no mobile version.
• Accountants: find restaurants with no online booking, a sign they may need help with bookkeeping too.
• Marketing agencies: find firms with no reviews page.
• Anyone hiring a contractor: find tradespeople on Google Maps, read their reviews, and pick the best one.

If your prospects are not local businesses, how to generate B2B leads in ChatGPT builds lists from Apollo and FullEnrich company data instead of Google Maps. Once you have meetings booked, how to use ChatGPT for sales covers account research and call prep.

To build the same pipeline in code instead of a chat, see how to build an AI SDR that researches companies in real time. For a Python version that pulls leads from a list of URLs, see the bulk sales lead extractor tutorial. Fire Enrich is Firecrawl's open-source tool for enriching a spreadsheet of companies.

Google Maps finds the firms, a website scrape qualifies them, and Alexandria's people providers find who to email.

Check out the Firecrawl Alexandria docs and the ChatGPT install guide to go deeper.