Most discovery calls start cold. The prospect arrives knowing nothing about you. You arrive knowing nothing about their situation. You spend the first fifteen minutes on context-gathering that should have happened before either of you joined the call.

We built the Discovery Assistant because the founder of Augself faced that exact problem in his own consulting practice — and got tired of solving it manually.

Why we built it

Jose Lazo-Flores runs an independent AI consulting practice in Zurich. Every inbound discovery call burned thirty to forty-five minutes of his calendar. The first fifteen were always the same: "Tell me about your business. What's prompted you to look at AI? What have you tried?" The next fifteen were where the real work happened.

A non-trivial fraction of those calls weren't even from real prospects. Recruiters, vendors, researchers wanting to chat about AI — same screening cost, zero return.

Pre-call intake forms didn't fix it. Completion rates were poor, and a form is a one-way artefact. What he needed was something that could have a conversation with a prospect before the call. Not a chatbot reading from FAQs. A structured intake conversation that surfaced exactly the context he'd otherwise spend the first half of the call digging out.

So we built one.

What it does

The Discovery Assistant is an AI chatbot trained on the consultant's services, methodology, and pricing structure. It runs at a hosted URL (Jose's lives at augself.ai/t/discovery) and has a structured conversation with each prospect who lands on it.

A few things it does specifically:

Opens warm in the consultant's voice — third-person about them, peer-to-peer with the prospect, no salesperson register.

Asks adaptively. A prospect with a clear ask gets a quick path; an exploring prospect gets more open-ended questions.

Explains services and pricing in context — not as a brochure, but in response to what the prospect is actually asking.

Hands off cleanly with the booking link when the prospect is ready, then stops.

Detects non-prospects. Recruiters and vendors get handled gracefully without notifying the consultant.

After the conversation, the consultant receives an email summary: who the prospect is, what they're trying to do, what they've tried, what they need. The call starts at minute fifteen of where it used to start.

What we learned designing it

A working prototype was easy. Making it useful in production took a week of iteration. A few of the patterns we learned:

Reciprocity matters. Bots that interrogate from message one feel like forms. A structured reciprocity rule — the bot shares specific things in proportion to what the prospect shares — changed conversation quality more than any other single change.

Must-ask checklists beat free-form conversation. Before the bot offers a booking link, a small set of questions has to have been answered. The check is silent; the prospect doesn't see it. But it ensures every summary email arrives complete.

Anti-patterns belong in the prompt. A list of "things the bot must never do" prevented entire classes of failure that would otherwise have shown up unpredictably.

The handoff is the product. A great conversation that ends with "we'll email you" wastes the warmth. Routing prospects to the booking link while the conversation is still hot is what closes the loop.

Now available to other consultants

We built it as part of the Augself platform, then made it productizable — same template, customized for each consultant's services and methodology.

It's a new product. Jose's bot is the only one running in production today. We're onboarding our first external customers now.

Two ways to evaluate it:

Live demo: chat with Jose's Discovery Assistant — it's the same product you'd get for your practice.

Learn more: Discovery Assistant product page →

Built for ourselves first: how the Discovery Assistant earned its name

Why we built it

What it does

What we learned designing it

Now available to other consultants

Curious if the Discovery Assistant fits your practice?