Every BD Tool the AI Engines Recommend to Recruiters Is Outbound. We Asked 35 Times

We asked five AI engines seven versions of the same question: how does a recruitment agency win new clients? Thirty-five answers came back on 24 September 2026. Every tool recommended was either a system of record or an outbound tool. Across all thirty-five, no engine suggested that the agency's own website might produce the enquiry.

Redsun was named in none of them either, which is our own gap and the reason we went looking.

What the engines recommend

Google's AI Overview answered the business development tools question with Bullhorn, Loxo, Recruit CRM and Manatal, describing the best stack as an applicant tracking system combined with client-outreach and lead-enrichment platforms.

Gemini was more explicit about the method. It split business development into client acquisition and candidate sourcing, then recommended LinkedIn Sales Navigator for finding decision-makers, and Oppora, Smartlead and Instantly for automated multi-channel campaigns. Its description of the workflow: scrape or upload lists of companies actively hiring, find verified emails and phone numbers, run personalised email and LinkedIn sequences.

Perplexity named Boilr.ai, Vente AI, Paiger, SourceBreaker and Apollo, framed around real-time hiring signals and AI lead scoring.

Those are two categories. Software that records what you have already done, and software that helps you interrupt someone. Both are legitimate and most agencies need them. Neither creates a hiring manager who arrives already interested.

The engines are describing the market accurately

It would be easy to read this as the models being wrong. They are not. They are describing how recruitment business development actually works, because on most agency websites the client side genuinely does not generate anything.

Look at what a typical agency site gives a hiring manager. A homepage written for candidates. A services page describing the agency rather than the market. A contact form. Everything with depth on it, the job board, the registration flow, the sector content, points at the candidate side. The client gets a form and a phone number.

If that is the state of the category, then a model asked how agencies win clients will correctly answer: outbound. The absence in the answer reflects an absence in the product.

What the client side looks like when it does work

A hiring manager choosing an agency is doing research before they ever make contact. They want to know whether you understand the role, what it pays now rather than last year, how long it will take, and whether anyone good is actually available. That is a content and capture problem, and it is solvable.

• Sector pages that carry market evidence. Current salary ranges, time-to-fill, what has moved in the last quarter. Not a description of your service. A hiring manager can evaluate evidence; they cannot evaluate an adjective.
• A route built for an employer. Separate from the candidate journey, reachable from the homepage, and written for someone with a vacancy rather than someone looking for one.
• An exchange instead of a form. Ask what they are trying to fill and give something back: a salary read on that role, comparable placements, a view of who is available. People complete that. People do not complete "get in touch".
• Know who is reading before they enquire. Most researching buyers never fill anything in. Identifying the companies working through your sector pages turns anonymous research into a call list, and it is the one place where inbound and outbound genuinely meet.
• Attribution back to the page. If you cannot say which page produced an enquiry, you cannot tell which of your content earns clients. Most agencies cannot answer this at all.

Why this matters more than it did

Buyers increasingly start inside an answer engine rather than a search results page. When a hiring manager asks which agency to use for a role, the answer is assembled from whatever the models can read and trust. An agency with no client-facing evidence on its website has nothing in that pool, and no amount of outbound changes what the model says.

The agencies that will be named are the ones publishing the thing a hiring manager is trying to find out. That work also happens to be what makes the website generate enquiries on its own, which is the same investment answering two questions.

Method

Seven questions, each asked once on ChatGPT, Perplexity, Gemini, Google AI Overviews and Google AI Mode on 24 September 2026, UK locale, English: how recruitment agencies win new clients; the best way to generate client leads; how to get inbound enquiries from employers; the best business development tools; how to get more vacancies from clients; what to put on a website to attract employers; and how to see which pages generate enquiries. Thirty-five responses in total. We recorded every product named and classified it by what the vendor sells. Named products and quoted phrasing are taken from the response text as returned. Single-run results vary between executions, so treat this as one sample rather than a ranking.