Arunly · AI Conversation Studio

Most AI agents read lines.
Ours hold a conversation.

A founder-led studio that designs, advises on, and builds the voice and chat systems behind AI that actually sounds human — and ships its own products along the way. Multilingual conversation, a specialty few can match.

live agent turn · multilingual
வணக்கம்! நான் உங்க senior living assistant. உங்க requirements கொஞ்சம் சொல்லுங்க — perfect community-ஐ நானே suggest பண்றேன்.
Tamil shown — agents also built in Hindi, Kannada & English, code-switching the way people really speak.

Agents that qualify a family in Tamil, sell a course in Tanglish, and stay in character when a caller pushes back — running today across senior-living care, overseas education, and premium food brands.

Why the studio exists

A dentist who became a founder, and learned to build the thing itself.

I trained as a dentist, then spent years as a health-tech founder — living the gap between a bold idea and a product people actually use. That taught me the discipline this studio runs on: understand the human on the other end first, then engineer the system that serves them. Today that means conversation AI — designed and built to hold up in the real world, not just the demo.

What the studio does

Design. Advise. Build.

Take one, or the whole arc. Founder-led on every engagement, with specialist collaborators brought in per build. The through-line never changes: conversation systems that hold up in the real world, not just the demo.

01 — Design

Conversation design

Personas, dialogue flows, prompt systems, language rules, knowledge-base structure. The blueprint that makes an agent coherent and on-brand before a line of it ships.

02 — Advise

Strategy & advisory

Agent architecture, build-vs-buy, guardrail and escalation strategy, and QA that tests against real calls — not just the path where everything goes right.

03 — Build

Deployment & iteration

Shipping agents on modern voice platforms, wiring post-call data and CRM pipelines, then tuning from live-call QA until the conversation truly holds.

The craft

Not scripts. Engineered conversation.

Every agent comes from the same discipline: a defined persona, a language system, guardrails that hold under pressure, and a data layer that turns talk into records.

01 · language

Native multilingual voice

Agents that move between Indian languages and English the way people really speak — Tamil, Hindi, Kannada — natural, consistent, never robotic. A genuine edge for brands serving Tier-2 and Tier-3 India.

02 · control

Guardrails that hold

Logic that stops an agent guessing pricing or promising outcomes; budget-flip sequences; disqualification paths; escalation by department. In-bounds even when the caller pushes.

03 · rhythm

Human turn-taking

Barge-in-safe openings, sub-second bridges for latency, interruption recovery, silence handling. The details that separate a conversation from a bot reading lines.

04 · data

Post-call intelligence

A modular LLM layer after every call — summary, outcome tagging, structured extraction into the CRM — turning conversations into clean, queryable records.

Selected work

Shipped. Tested on live calls. Iterated with clients.

Client names withheld by preference; described by sector. Each ran a full delivery cycle — build, live-call QA, client review, iteration.

Case 01
Senior-living real-estate developer · multi-project
● Live in market

A bilingual concierge that qualifies families and books site visits

The problem

Enquiries from seniors and their adult children arrived across WhatsApp and web chat, mixing Tamil and English freely. The team needed warm, patient qualification — budget, location, intent — without a human on every first touch.

What I did

Designed the full persona and conversation system — empathy-first flow, budget-based routing across four project tiers, objection and competitor handling, CRM write-back of name, relation, intent, budget and visit slot — deployed across WhatsApp and web with automated visit confirmations.

Outcome — Markedly faster lead qualification. The client ran structured QA against real call recordings and returned line-level feedback that shaped the next iteration.

Case 02
Overseas-education & language-training firm
● Live · client demo cycle

A recruitment-manager agent that screens leads and sells a course

The problem

Social-media leads for overseas job and study programs needed eligibility screening and routing to the right human expert — while fully handling the sales conversation for language-course leads, in natural Tanglish.

What I did

Designed a dual-path agent: triage-and-handover for complex programs, and an end-to-end qualification and sales flow for course leads — with mandated CRM capture, disqualification rules, and WhatsApp registration handoff.

Outcome — Delivered through a live client demo cycle with recorded calls and transcripts; the client returned detailed QA that shaped v2.

Case 03
Premium vegetarian-protein FMCG brand
● Built · showcase demo

A product-education agent with a genuine kitchen-buddy voice

The problem

Callers — home cooks, chefs, first-time buyers — needed product, recipe and cooking guidance in warm colloquial Tamil, with a hard line: never guess pricing or bulk terms, always escalate cleanly.

What I did

Designed a brand-grounded persona over a structured knowledge base, caller-type branching, and credibility cues used only when trust is questioned — so the agent reads as a person, never a catalogue.

Outcome — Used as a showcase demo; the clearest expression of brand-voice and multilingual craft in the portfolio.

What the studio is building

We don't only build for clients. We're building the platform, layer by layer.

Client work sharpens the craft; our own products are where it compounds. Two modules of the healthcare platform, in build now — more as the studio grows.

module · in development

The front-desk layer

A multilingual voice and chat agent that manages patient communication end to end — booking, rescheduling, cancellation, post-visit follow-up — across Tamil, Hindi and English. Proving the model in dental clinics first, then scaling across healthcare.

module · in design

The growth layer

The system that helps clinics get discovered and grow — local visibility, patient reactivation, review generation, feeding the agent layer above. Vernacular-first, built for Indian healthcare.

How I think

What you're really hiring.

Tools change every quarter. Judgment is the thing worth paying for — here's the judgment I bring to every engagement.

principle 01

The human comes before the model

I start with the person on the other end — anxious patient, busy front-desk, first-time caller — and design the conversation around them. A trained clinician's instinct for how people actually talk when something matters.

principle 02

Safety is a design input, not a disclaimer

In healthcare especially, an agent that oversteps is worse than no agent. I build in clear boundaries, escalation to humans, and honest limits from the first line — so the system earns trust instead of risking it.

principle 03

Systems, not scripts

Anyone can write a prompt. I architect the whole system — persona, language, guardrails, data, escalation — so it holds up across thousands of real, messy, interrupted conversations, not just the demo.

principle 04

I build to own, not to rent

Every engagement leaves you with architecture you control — no lock-in, no black box. The same discipline I use to build the studio's own products, applied to yours.

Where Arunly is headed

An AI operating system for healthcare.

The client work and the products point somewhere. The studio is building toward an all-in-one AI automation platform for healthcare — voice agents, chat agents, and agentic workflows working together to support patient communication, clinic operations, and service delivery, from single clinics to hospitals and PHCs.

A system that listens, understands, and communicates back — with empathy, across Indian languages, and always inside the boundaries that healthcare demands: medical-AI safety guidelines, clear disclaimers, and human clinical oversight at every step. Ambitious on purpose. Built carefully, on purpose.

Listens. Thinks. Understands. Communicates back.
Foundations

Built on modern voice-AI infrastructure.

Voice-agent platforms
Conversation architecture
Multilingual & code-switching
Agentic workflows
Post-call intelligence
Healthcare-safe design
Work with me

Building something that has to hold a real conversation?

I take on conversation design, advisory, and build work for teams shipping voice and chat agents — multilingual ones especially. Written-first, async, fixed-scope.

Start a conversation →
or connect on LinkedIn