Click him, ask a question out loud, and the site moves with you: he answers from our own content, scrolls to the right section, opens a live demo, and opens the booking calendar — all without ending the call.
Tarun is an AI assistant, not a person, and he says so at the start of every call. He is live on every page of this site — look for the widget in the corner.
A prospect asks what we do, gets taken to the right part of the page, opens the live ticket-triage demo, triages an inbox, filters it to what needs a human — and books a call. Eighty-five seconds, one continuous conversation.
The voices are Tarun's own ElevenLabs voice and a second voice for the prospect. The pages, the scrolling and the demo are the real site.
Most website chatbots answer in a box. Tarun answers out loud and then operates the page in front of you — which is the part prospects remember.
Services, use cases, how we work, what we've built, who we're not for. His answers come from this website's own content, indexed as a knowledge base — not from general internet knowledge.
Ask for "customer service" or "your health app" and the page scrolls straight there. He can reach 25 named sections and cards on the homepage.
Case studies and the six interactive AI demos open inside the site while the conversation carries on. A normal page change would end a voice call — this doesn't.
When you want a call, he opens the real booking page inside the site and steps back. He doesn't ask for your name, email or time zone — the calendar already does that, and does it correctly.
Drag him anywhere on screen, or shrink him to a small photo bubble. The call keeps running while he's minimised.
He qualifies gently, never pressures, and passes you to a human call when it's worth one. Everything he says about DM AI Labs is drawn from pages you can read yourself.
The same conversation as a silent, click-through replay — handy on a phone or in a meeting room with the sound off. To hear the real thing, open the homepage and click Tarun.
Illustrative replay, not a recording. The panel on the right is a simplified stand-in for the real page so it stays readable on a phone.
Nothing here is off the shelf. The voice, the instructions, the conversation flow, the knowledge and the browser-side tooling were all built for this site.
A calm, professional Indian-English voice, picked for advisory delivery rather than energetic sales, with speed and stability tuned for a business conversation.
A written brief covering who he is, what he may claim, how briefly to answer, how few questions to ask, and what to do when a tool fails. Rewritten repeatedly against real call transcripts.
Guide first, qualify only if the visitor wants to solve something, then hand off. Each step has its own brief, so he doesn't interrogate a browser or waffle at a buyer.
The website is crawled and indexed, so answers come from what we've actually published. Re-index the site and his answers update with it.
One scrolls the page to a named section; one opens a page, a demo or the booking calendar. Both report back, so he only claims to have moved you when the page really moved.
Pages open inside the site so the call survives, the widget rides along on every page, and he can be dragged or minimised. That plumbing is as much of the build as the agent itself.
The same pattern works for inbound sales, front-desk and support triage, product walkthroughs, onboarding, dealer and distributor queries, and internal helpdesks. Two ways to get there.
Discovery and scripting, voice selection, prompts and conversation flow, a knowledge base from your content, custom tools for your site or systems, testing against real calls, and handover with the transcripts dashboard.
A working session where your team builds an agent on your own content: prompt structure, workflow design, knowledge base, tools, and what to watch in call transcripts afterwards. You leave with something running, not slides.
One implementation credit equals ₹1,000 of approved DM AI Labs professional setup work. Third-party subscriptions and usage are not included.
₹13,500 GST (18%) · ₹88,500 total
₹9,000 GST (18%) · ₹59,000 total
The same defined initial scope as the standard package: a 25-credit pre-GST discount, with no hidden reduction.
3 credits per month: ₹3,000 + ₹540 GST = ₹3,540 total. Maintenance is required while DM AI Labs provides active monitoring and routine support. The recommended support period is the first 12 months after go-live, but billing is month-to-month with no annual prepayment. You may cancel with 30 days' written notice; paid maintenance periods are non-refundable except where required by law. After support ends, Tarun may continue on your provider accounts, but DM AI Labs monitoring, support and tuning stop.
The actual invoice will show CGST/SGST or IGST as applicable. New workflows, integrations, redesigns or major prompt and knowledge-base changes require a separate written credit estimate and approval before work begins.
Want this as a deck? The same story in twelve slides, including what we learned running Tarun in public: PDF · PowerPoint
Book a free 30-minute fit review. We'll confirm whether Tarun fits your business, what content and systems are required, and whether the founding offer applies.
Request a Tarun fit review