It reads your services pages, manifesto, case studies, and team bios, so the bot's representation of the agency tracks the clarity of the underlying content. Agencies with crisp service definitions get crisp bot answers; agencies whose positioning is vague get bot answers that match. Most teams review the bot's tone after a week of live conversations and tighten either the system prompt or the published positioning where the bot drifts.

If you publish retainer ranges, yes, exactly as written. If you would rather not commit to a public rate card, configure the bot to give project-size context anchored to past work, like "comparable SaaS retainers we run start at 8,000 USD a month," and offer to book a call for an exact figure. Most agencies prefer this hybrid: enough price signal to screen out fundamentally mismatched prospects, without losing pricing power.

Anything marked private or password-protected in WordPress stays out of the bot's context by default. You can also exclude specific case studies by setting a custom field or by scoping the data source in the wizard to only published, public posts in the case-study post type. Sensitive metrics that live in private memos never reach the model.

Yes. The system prompt can encode any qualifying axis the agency cares about: industry fit, content cadence, in-house writer count, SEO maturity, current stack, decision timeline. The bot weaves these into the conversation naturally rather than reading a form-field grid. Prospects who fail the fit check get a polite, honest no rather than a wasted discovery call.

Conversations log to WordPress and can fire webhooks to HubSpot, Pipedrive, Notion, or Airtable when a chat closes. Most agencies forward qualified leads with the full transcript attached, mapped to their pipeline stages. The raw transcript is often more useful than structured fields because it captures the prospect's actual language and concerns, which informs how new business approaches the call.

Yes. Multibot lets editorial, SEO, video, and demand-gen arms each have their own bot scoped to its services and case studies. A prospect on the programmatic-SEO cluster gets the SEO bot's tone and proof points; a prospect on the thought leadership cluster gets the editorial bot. Display conditions in SleekAI handle the routing by URL pattern or taxonomy term.

Yes. The underlying model handles multiple languages, so a German CMO asking about prior SaaS work in German gets a reply in German with case study summaries translated on the fly. For agencies chasing European or LATAM business, this measurably broadens the qualified inbound pool. The conversation log still records the original language for the new business team's reference.

Configure the system prompt to defer commitments on dates, deliverables, and headcount allocation to a discovery call rather than commit on the agency's behalf. A line like "a senior strategist would normally lead a program of this shape, happy to confirm staffing on the call" is honest without losing momentum. Audit conversation logs weekly and tighten the prompt where the model tries to commit too eagerly.