How to Use ChatGPT, Claude and Other AI Tools to Find B2B Leads (and Where They Fall Short)

Quick answer: AI chatbots such as ChatGPT and Claude are excellent at researching companies, defining your ideal customer, spotting signals in public news and drafting outreach. They're poor at producing reliable lists of contacts, can't see behind logins like LinkedIn's, don't check the web for you every day by default, and sometimes invent details. Use them as a research assistant, not a lead database.

AI is now part of how both sides of B2B work. G2 found that 51% of B2B software buyers start their research with an AI chatbot more often than with Google. Sellers are using the same tools to find buyers. Here's how to do it well.

What AI tools are good at

1. Defining your ideal customer

Give an AI tool your website, a description of your best clients and why they bought, and ask it to find the pattern. It's very good at turning a messy list of clients into a clear profile.

Here are my 10 best clients, with their industry, size and why they bought from us: [list]. What do they have in common? Write an ideal customer profile I can use for prospecting, including the events or changes that usually come before they buy.

The last part matters most. It turns your profile into a list of buying signals to watch for.

2. Researching a named company

This is where AI saves the most time. Ask it to research a specific company before you reach out:

Research [company name, website]. Summarise what they do, who they sell to, any news from the last six months, any leadership changes, hiring activity and stated growth plans. Then tell me the most likely reason they'd need [what you sell] right now. Link every source.

Always ask for sources, and click them. More on why below.

3. Spotting signals in news

AI tools with web search can scan recent news for events that suggest a company needs what you sell:

Search for UK news from the last 30 days about [industry] companies with 50 to 250 employees that have announced expansion, new sites, funding or senior hires. List each company, the event, the date and the source link.

You'll get a useful starting list. It won't be complete, and it'll miss anything that didn't make the news.

4. Drafting outreach

Once you have a real signal, AI is good at turning it into a short, relevant first email:

Write a cold email to [name, job title] at [company]. The reason for contacting them is [signal]. We help [type of company] with [problem]. Keep it under 80 words, mention the signal in the first line, no flattery, and end with one easy question rather than asking for a meeting. Use British English.

Then rewrite it so it sounds like you. Our guide to signal-led cold emails has before and after examples to compare against.

5. Preparing for calls

Before a first meeting, ask for likely priorities, objections and good questions based on what the company has published. It's a fast way to turn up informed.

Where AI tools fall short

They can't see most of the data that matters

Much of the most useful prospecting information sits behind logins or paywalls: LinkedIn profiles and posts, many job boards, paid databases and trade publications. A general AI chatbot searching the open web usually can't see it, or sees only fragments.

Their coverage is patchy

AI search finds what's well indexed and widely reported. A large company's funding round will show up. A 60-person manufacturer hiring a new operations director often won't, yet that might be your best prospect this month.

They sometimes make things up

AI tools can produce confident but wrong answers: a person who's left the company, a job title that doesn't exist, or an email address that's simply a guess. Never send outreach to an email address an AI chatbot has guessed. It hurts your deliverability and can break data protection rules. Verify every contact with a proper email finding and verification tool.

They don't watch the market for you

Signals are most valuable when they're fresh. Unless you set up a scheduled task or agent, a chatbot only finds what you ask for, when you ask. Running the same searches by hand every morning across your whole market quickly becomes a part-time job.

They don't remember what you've already done

A chatbot doesn't know who you contacted last month, who replied, or which signals led to meetings. That feedback is how prospecting improves over time.

Data protection still applies

Using AI doesn't change UK GDPR and PECR. You still need a lawful basis for processing personal data, a clear way to opt out, and care around sole traders and partnerships. Check the ICO's guidance before scaling up.

A simple AI-assisted prospecting routine

If you want to do it yourself, this works:

1. Once: use AI to write your ideal customer profile and list of buying signals.
2. Weekly, 30 minutes: run news searches for those signals in your market. Save the companies that fit.
3. For each company: ask AI to research it and suggest the best angle. Check the sources.
4. Find the right contact yourself, and verify their email with a dedicated tool.
5. Draft with AI, edit by hand, then send from a properly set-up inbox. See our guide to cold email deliverability.
6. Log what you sent and what happened, so you learn which signals work.

When to use a dedicated tool instead

AI chatbots are general tools. Dedicated prospecting tools are built for the parts chatbots struggle with: watching many sources every day, matching signals to your ideal customer, finding verified contacts and tracking outcomes.

Go7 Radar does exactly this. It reads your website to understand what you sell, brings you companies showing a relevant buying signal every morning, explains why each one matters, finds a verified contact and drafts the outreach. It also brings press opportunities with suggested comment angles, so you can build credibility at the same time. You review and send from your own inbox.

For a wider look at the options, see our guides to buying signal tools and client acquisition software.

If you'd rather not do any of it yourself, our Managed Growth service runs signal-led outreach and digital PR for you.

Not sure where your best opportunities are? Get a free Growth Report.

FAQs

Can ChatGPT find B2B leads? It can help. ChatGPT is good at researching companies, spotting signals in public news and drafting outreach. It isn't a reliable source of contact details, and it can't see behind logins such as LinkedIn's.

Can I use Claude for sales prospecting? Yes, in the same way: researching companies, analysing your best customers, summarising news and drafting emails. Treat it as a research assistant and verify everything it finds.

Is it safe to email addresses that AI finds? Not without verifying them first. AI tools sometimes guess email addresses. Sending to unverified addresses increases bounces, damages your sender reputation and can create data protection problems.

What's the best AI prompt for finding leads? Be specific about the industry, size, location and buying signal you're looking for, ask for recent events only, and ask for a source link for every company. Then check the links.

Will AI replace sales prospecting tools? AI is changing how prospecting tools work, and many now use AI under the hood. General chatbots still lack daily monitoring, verified contact data and a record of what you've sent, which is why dedicated tools remain useful.

Do buyers use AI to research suppliers too? Yes. G2's 2026 research found 51% of B2B software buyers start their research with an AI chatbot more often than with Google. Being mentioned by credible sources, including the press, helps you show up in those answers.