How we shipped an AI consultant on the landing page for ~$12/month

Problem

We wanted the TenetLab landing page to work around the clock: answer service questions, qualify leads with a lightweight BANT-style flow and push contacts to Telegram without a human on call after hours. Budget was tight—enterprise chatbots and heavy CRMs were out. We needed composable SaaS with predictable pricing and room to iterate quickly.

Criteria: voice and text (some visitors prefer speaking), a clear dialogue script, lead storage in a database and a webhook for notifications. Speed mattered—ship an MVP in days, not a quarter.

Solution

Three services covered production needs:

• ElevenLabs Conversational AI for voice and text—roughly $5/month at our volume, flexible prompt and behaviour tuning.
• FastAPI on Railway as a thin backend for webhooks, validation and lead routing—about $5/month.
• Supabase for PostgreSQL plus APIs—the free tier handled early traffic.

The pipeline is simple: a visitor talks to the on-site agent; when the goal triggers (contact captured, qualification complete) the backend writes a row and pings Telegram. We refined the prompt from live logs—which questions repeat, where people drop and which phrasing converts best.

Outcome

The AI consultant runs on tenetlab.agency—you can try the flow yourself. Total infra spend at our load lands around $12/month, no long contracts and no owned GPU.

Next up: CTA copy experiments and deeper CRM integrations when clients ask.

Want a similar bot on your landing page? Book a strategy session—we will show how to wire a comparable loop in a single working day.