Ask most agency owners where their new business comes from and the answer is some combination of referrals, word of mouth, and the occasional inbound enquiry. That works until it does not. And when it stops, there is no pipeline to fall back on.
Building a systematic new business pipeline used to require a dedicated BD person, or at least a significant chunk of the founder’s time. AI changes the economics of that equation.
The AI-powered pipeline
A new business pipeline has four stages. AI can accelerate each one.
1. Prospecting: finding the right targets.
AI can process large lists of potential clients and identify the ones most likely to need your services. Feed it your ideal client criteria (industry, size, location, current marketing activity) and a list of companies. It will rank them by fit and explain why. Pair this with an AI lead scoring model and you can prioritise systematically rather than on gut feel.
For most agencies, LinkedIn Sales Navigator combined with AI analysis is the most effective prospecting approach. Export your saved leads, feed the data into Claude, and ask it to prioritise based on signals: recent funding, new hires in marketing roles, website changes, or content gaps.
2. Research: understanding each prospect.
For your top 10-20 prospects each month, AI generates a research brief: their business model, recent news, competitive landscape, digital presence assessment, and likely pain points. This is the context that turns a cold outreach into a warm conversation. We cover the full process in our guide on researching prospects with AI.
The prompt: “Research [company]. Summarise their business, target market, main competitors, current marketing approach (based on their website and social presence), and three likely challenges I could help with as a [your specialism] agency.”
3. Outreach: starting the conversation.
AI drafts personalised outreach messages based on the research. Not template emails with a name swapped in. Genuinely tailored messages that reference specific aspects of the prospect’s business and connect them to your capabilities.
The key is specificity. “I noticed your website does not have structured data markup, which is likely costing you featured snippet opportunities for [specific keyword]” is infinitely more effective than “I can help with your SEO.”
4. Proposals: winning the work.
Once a prospect is in conversation, AI accelerates the proposal process. Research is already done. The tailored executive summary writes itself from the discovery call notes. The case study selection is informed by the prospect’s industry and challenges.
The numbers
An agency using this system consistently can expect to:
- Research and qualify 50-100 prospects per month in 2-3 hours (vs. 15-20 hours manually)
- Send 20-30 personalised outreach messages per month in 1-2 hours (vs. 6-8 hours manually)
- Convert outreach to conversations at 8-15% (vs. 2-5% with generic outreach)
- Produce proposals 60% faster with higher quality tailoring
The total time investment is roughly 5-8 hours per month. For most agencies, that is manageable even without a dedicated BD hire.
Getting started
You do not need all four stages running on day one. Start with stage 2: research. Take your next 5 prospects and run the AI research process. Walk into those calls better prepared than you have ever been and see how the conversations change.
Once you see the difference that preparation makes, the rest of the pipeline builds naturally.
This is part of The Pitch Stack, a series on using AI to win more agency work. Subscribe to the newsletter to get new articles weekly.
