Design agency sites are tuned for taste, not for self-qualification. A prospect with a 120,000 USD product design project lands on a beautiful case-study grid, skims four projects, and still has the same three questions: have you done my kind of work, what does this kind of engagement actually cost here, and who would lead it. The contact form does not answer any of those, and prospects either move on or send a vague brief that turns into a recruiter-style scheduling exercise. SleekAI reads every case study, every service tier, and every team bio, then answers "have you done B2B product design" with specific links and rough scope.

The bot pulls from wp_posts, ACF fields, and custom taxonomies via the data-source wizard. Tag case studies by discipline (brand, product, web, packaging), industry, and engagement size, and the bot filters answers on any combination. Display conditions run discipline-specific bots per landing page so the product design cluster gets a product-focused bot and the brand cluster gets a brand-focused one.

Pre-qualification keeps the new business team focused. The system prompt encodes the studio's minimum engagement, the disciplines the team does and does not service, and rough ranges per service tier. Off-fit prospects get a polite no; qualified prospects book a discovery call with the transcript attached. Conversation logs include model name, token usage, and page URL so the agency can audit which clusters convert and tune the prompt over time.

Design retainers and project engagements close on confidence, and most studio sites do less than they should to build it. A product lead at a fintech weighing three studios for a 130,000 USD dashboard redesign has a short list of questions blocking a call: has this team shipped comparable work in B2B fintech recently, what does this engagement actually cost at our scope, who would be the day one design director, and is the design system maturity assumption realistic for our stack. The portfolio answers part of the first question.

The services page answers part of the second. Nothing else closes the loop, and the prospect either submits a brief and waits or quietly moves on to the studio whose site let them get the answers in conversation. The friction layer carries more weight in conversion than most agencies realize.

A bot that reads every case study and quotes published ranges removes that friction without committing to a public rate card. Pre-qualification in conversation protects the new business team's calendar, which is the most expensive resource in the funnel after senior design time. Filtering out projects under the minimum engagement, or in disciplines the team has no proof in, before a call books, raises close rate on calls that do happen.

The transcript is the compounding payoff. Reading a week of conversations tells the studio which disciplines and industries inquire most, which case studies actually pull weight on cluster pages, and which scope questions recur often enough to deserve a dedicated FAQ. That feedback loop is hard to get from contact form submissions, where the prospect's real question gets compressed into a one-line subject before anyone reads it.