How AI is changing B2B prospecting

By Dave Curran, Co-Founder, Firmbase | March 2026 | 10 min read

There's a quiet revolution happening in B2B prospecting, and most sales teams haven't noticed yet because they're still thinking about AI the way they think about email: as a tool to use, not as a way to fundamentally change how they work.

When ChatGPT arrived, the first wave of "AI for sales" was obvious - use AI to write cold emails faster. Automate the copy. That's useful. But it's not why AI actually matters for prospecting.

The real change is subtler: AI can now understand what companies actually do, synthesise signals across multiple data sources, and tell you which accounts are worth your time - without you having to configure anything.

Why filter-based prospecting is reaching its limit

For the last 15 years, the prospecting model has been the same: you pick variables, you configure filters, you export a list.

Looking for software companies in London with 50 - 200 employees that use HubSpot? Filter: industry = software, location = London, employee band = 50 - 200, technology = HubSpot. Export. Done.

The problem is that this model assumes precision in how business data is classified. It assumes everyone categorises themselves the same way. It assumes that SIC codes (Standard Industry Classification) tell you what a company actually does.

They don't.

A company with SIC code 6201 (software development) could be a pure-play developer tools vendor. Or it could be a management consulting firm that happens to write code. Or it could be a SaaS company that uses custom development internally. Same SIC code, completely different buying priorities.

A company's description says "we're an agency." Are they digital? Design? PR? Media? Search? All of the above? You need to read their website, their job postings, their LinkedIn profile, their Companies House filing. One data point isn't enough.

Filter-based systems assume you can be more precise than reality allows. So you end up exporting lists with a lot of noise - accounts that technically match your filters but aren't actually relevant.

Or you end up creating so many filters that your target list becomes impossibly narrow, and you lose coverage entirely.

What natural language search changes

Natural language search flips this on its head. Instead of configuring filters, you describe what you're looking for in plain English: "media agencies in London that have hired a new creative director in the last six months."

The AI reads that instruction and understands:

It cross-references Companies House filings (which have director appointment dates), job posting data (which shows seniority and role), and company descriptions (which tell you what they actually do). It synthesises that information and surfaces accounts that match - not because they hit a filter, but because they actually fit what you described.

The difference feels small. It's not. Filter-based systems find you accounts that match your configuration. Natural language systems find you accounts that match your intent.

Why AI synthesising signals matters more than AI writing copy

This is where the real opportunity sits. Prospecting isn't actually about finding accounts. It's about finding accounts that are about to buy something.

Every company leaves signals: they hire people, they appoint executives, they file accounts, they receive funding, they post job listings. But no single signal is dispositive. A company might hire a CFO because they're preparing for acquisition. Or because they're just replacing someone who left. You can't tell from the hire alone.

But if you see a CFO hire, a new finance director appointment, a fundraising announcement, and increased job postings in revenue roles all within a 60-day window? That's a signal. That's a company in growth mode, evaluating tools, potentially raising capital, building out infrastructure.

Traditional prospecting tools show you single signals. A job posting here, a funding round there. You have to synthesise them yourself. You're reading 20 accounts and manually trying to figure out which ones are actually worth your time.

AI agents can do that synthesis at scale. A seller describes their ideal account profile and their buying signals. The AI continuously monitors signals across multiple data sources. It weighs them, prioritises them, and surfaces the accounts that matter most. The seller's job becomes: "pick up the phone" instead of "decode which leads are actually hot."

How this maps to the evolution of prospecting tools

The prospecting tech landscape is reorganising around this shift:

Old model: Database companies (ZoomInfo, Apollo) sell you access to their data warehouse and filters. You query it. You get lists.

New model: Continuous systems (like Firmbase) sell you access to data sources, plus an AI layer that monitors them and surfaces accounts based on signals and intent.

The old model doesn't disappear. But it becomes the commodity layer. You'll always need someone with clean contact data. But the real value isn't the data - it's the intelligence layer on top of it.

The practical upside for your team

This matters because it changes how sellers actually work.

In the filter-based world, prospecting is a bottleneck. You need someone to maintain segmentation, rebuild lists, configure filters, keep things updated. It's administrative work done by someone technical.

In the natural language + signal synthesis world, prospecting is a continuous system. You define what you're looking for once. The system finds it. You respond to what it surfaces.

For a small team without a RevOps person, this is transformative. You don't need someone dedicated to list-building and maintenance. You need a system that does that work for you.

For a larger team, it frees up your ops person to focus on actual strategy - which segments should we prioritise, which signals should we weight more heavily - instead of configuration and maintenance.

The change in skill required

There's an assumption that AI makes prospecting easier because "anyone can use it." That's partially true, but it misses the real skill shift.

The new skill isn't "how do I configure filters." It's "how do I describe what I'm looking for clearly." It's "what signals actually matter for my business."

A seller who just knows "I want to target software companies" will get worse results from an AI system than one who knows "I want to target companies that have recently appointed a new VP Engineering or are actively hiring DevOps roles."

The system is only as good as the intent you give it. That requires thinking clearly about your ICP, your buying signals, and why you're reaching out.

That's harder than it sounds. A lot of teams haven't done that work. When you force them to articulate it - not for a filter, but as an instruction to an AI agent - it often surfaces that they haven't actually thought through their targeting carefully.

But once they do, everything gets sharper.

Where the industry is heading

The industry is moving toward three things:

First: naturalisation of search. Fewer filters, more instructions. Instead of configuration, conversation.

Second: signal synthesis at scale. Less "here's a database of companies," more "here are the accounts in your universe that are showing buying intent right now."

Third: continuous re-prioritisation. Lists don't age because they're not static. They're live, continuously updated based on real signals.

The companies that build for that world are going to win. The companies trying to bolt AI onto list-building models are going to struggle because they're trying to solve two different problems at once.

What this means for you right now

If you're evaluating prospecting tools, ask:

The answers to those questions will tell you whether you're looking at a legacy system trying to stay relevant, or a system actually built for the AI era.

The future of prospecting isn't better filters. It's continuous discovery based on real buying signals.

Firmbase uses natural language search to understand what you're looking for, then continuously surfaces UK accounts matching your ICP and showing buying intent.

Start your free trial and see the difference that signal synthesis makes.

FAQ

Q: Does natural language search actually work, or is it just marketing?

A: It works, but with caveats. The better you describe what you're looking for, the better the results. "Media agencies" is vague. "Media buying agencies in the South East with revenue over £5M that have appointed a new business development director" is precise. Natural language handles that precision better than filters because it understands context and nuance.

Q: Won't AI get my targeting wrong?

A: It can, but not in the way you'd think. The issue isn't accuracy of data - it's clarity of intent. If you haven't clearly defined what your ideal customer looks like, no system (AI or filter-based) is going to make that work. AI just forces you to think about it more clearly.

Q: How is this different from just using ChatGPT to research accounts?

A: Using ChatGPT to research one company is helpful. Using an AI system to continuously monitor thousands of accounts, synthesise signals, and surface the ones worth your time is transformative. It's the difference between a research tool and a decision system.

Q: Do I still need contact data if I'm using an AI system?

A: Yes. The AI system tells you which accounts matter and why. You still need verified email and phone numbers to reach out. They're different layers.

Q: Won't this just mean more emails to more people, faster?

A: Only if you use it that way. The point isn't to reach more people. It's to reach the right people with better context. If you're using signal synthesis to understand which accounts are actually worth your time, you should be reaching fewer people with more relevance.

Q: How much setup is required?

A: That depends on the system. Ours: 15 minutes. You describe your ICP, highlight your buying signals, and the system starts running. No configuration required.

Author Bio

Dave Curran is the co-founder of Firmbase, a UK B2B sales intelligence tool that helps sales teams find, prioritise, and reach the right accounts without needing a RevOps team to make it work. Before Firmbase, Dave co-founded Love Mondays (acquired by Glassdoor, where he went on to serve as VP of Product) and Openvolt. He writes about UK B2B sales, prospecting, and go-to-market strategy.

Firmbase helps UK B2B sales teams discover their complete account universe, prioritise based on real buying signals, and reach out with genuine relevance - without the complexity of enterprise tools. Start your free trial