AI & Data Engineer · Co-founder, Ciatel Corp Ltd

Bala Pothabattula

I build AI agents and data systems for UK small businesses — voice receptionists, daily research pipelines, internal tools. Through Ciatel Corp Ltd, the consultancy I co-founded.

linkedin whatsapp ciatel-corp.com

Now

What I'm working on right now

Right now I'm working with UK small businesses through Ciatel Corp Ltd on two main things: AI receptionists that answer the phone 24/7 and book appointments straight into the calendar, and a daily research system that listens to what UK business owners are quietly complaining about online — so we know what's actually worth building next.

I'm also testing a quieter idea — instead of putting your customer data on a shared cloud server (the way most AI products work), I deploy each AI agent inside your own systems, so your data never leaves. Early answer is that the unit economics work, with a 12-month minimum.

Alongside client work, I'm deep in building an internal AI system that runs my business day-to-day — scanning the market every morning for where real AI demand is moving, surfacing which opportunities are worth my time and which aren't, and sharpening my thinking before high-stakes conversations. The kind of strategic intelligence that enterprise firms get from a full research team, running 24/7 on a one-person operation. It's the most ambitious thing I've built for myself, and it's changing how fast I move.

I haven't taken on legal, healthcare, or finance compliance work yet — that needs a specialist partner I haven't sourced. Everything else is in scope.

Work

Selected work

Four systems I've built. Honest about what's running, what's paused, and what's deliberately not built yet.

Arno 2.0

AI voice receptionist that picks up the phone and books appointments

What it is

An AI receptionist that picks up your business phone, has a natural conversation with the caller, and books the appointment straight into your calendar. Each business gets its own setup — your callers never share infrastructure with another business, and your customer data stays separate.

Why I built it

Service businesses lose around 15% of their bookings to unanswered calls — calls that come in after hours, during the school run, or while the front desk is with a customer. For a 20-booking-a-week salon at £50 average, that's around £150 of lost revenue every week. I wanted to see if today's voice AI was actually good enough to capture those calls without sounding like a robot.

Architecture · simplified

What makes it work

Two AI brains work together — a fast one runs the live conversation in real time, a more thoughtful one writes a summary of the call afterwards. You read the summary with your morning coffee.
It recognises your regular customers by phone number and greets them by name.
It checks your calendar live before quoting any time — so it never books an appointment that conflicts with one you already have.
If a call needs a human — an emergency, a high-value enquiry, an angry customer — it transfers to you straight away. It doesn't pretend to handle problems that need you.

What it doesn't do yet

Sending text confirmations needs a separate phone number alongside the voice number — a simple extra step at setup time.
The daily summary email of yesterday's calls is designed but not yet sending automatically — for now you read summaries inside the system itself.
The call summaries could be sharper at flagging the urgent items at the top.

Salesu — Lead Intelligence Engine

Daily research system that finds new business leads and writes the call brief

What it is

A research system that runs every morning before you get to your desk. It finds UK businesses that match the kind of customer you sell to, reads their websites, pulls their reviews, and hands your sales rep a one-page call brief with talking points — so the rep walks into the call already knowing the prospect. Currently paused (no live billing) until a client switches it on.

Why I built it

Most sales reps spend half their day on research before they pick up the phone — finding companies, reading their websites, looking up reviews, drafting the opener. That's mechanical work. I wanted to know what the day looks like if AI does the mechanical part overnight, and your rep walks in already prepared. I wanted to know if AI was finally good enough to do the part of sales that's mostly reading.

Architecture · simplified

What makes it work

Pulls leads from official UK sources — Companies House, Google Maps, healthcare and veterinary registers — so the businesses are real, not scraped contact data.
A safety check runs on every contact before anyone gets approached — no chasing people you've already spoken to, no contacting blocked names, no wrong-time outreach.
The call brief is written by the same AI tools used to power the most reliable enterprise software today. It reads the website and the reviews and writes a tailored opener.
I deliberately built only the research half. I didn't build the auto-emailing half — most businesses already have email tools they like, and choosing the wrong one for them would cost more to undo than to skip.

What it doesn't do yet

It doesn't send emails or read replies. That's deliberate — those steps tie tightly to your existing tools, and choosing the wrong fit early would be expensive to undo.
It doesn't push leads into your CRM yet — same reason, that's specific to whichever CRM you already use.
The 'type of customer' definition is currently a sensible UK default — first thing we'd customise for any new client.

Hero Pipeline / R&D Agent

Daily research system that listens to what business owners are quietly complaining about online

What it is

A system that reads the same online forums I used to scroll through manually — Reddit, Hacker News, accounting forums, job boards — and writes a one-page brief every morning of what UK business owners are quietly suffering through. It runs automatically and lands in Slack before I've had coffee. It's how I figured out which problems are worth Ciatel building products for.

Why I built it

Manually trawling forums for real customer pain was eating two to three hours of every morning before I'd done a single useful thing. I built this to give myself the headlines. The same system would work for any business that wants to know what their market is actually worried about — competitive research, customer sentiment, voice-of-the-market work.

Architecture · simplified

What makes it work

It only reads public sources — no scraping, no privacy issues, no legal risk.
A pre-filter throws out 60–70% of the noise before the AI even reads the post — so the system doesn't waste money on irrelevant content.
It groups similar complaints together, so you don't see the same problem ten times in a row.
If a topic suddenly spikes — five people complaining about the same thing in a single day — it sends an alert.
The end-of-day brief is an actual one-pager in your inbox or Slack — not another dashboard you have to remember to check.

What it doesn't do yet

The dashboard view (where you can browse by topic) currently shows sample data while we focus on getting the daily brief right. The plumbing is built; we'll switch it on once the daily brief has earned its keep.
Two extra features are designed but not yet running: a deeper competitive analysis when a topic gets interesting, and an auto-drafted social-media post about findings. Both are deliberately on hold until we're confident the daily brief is the right shape.

Database Assessment

A senior engineer spends a week inside your database and tells you what's broken — in plain English

What it is

A 5–7 day audit of your business's database. I get read-only access (so nothing can be broken during the audit), spend the week looking at seven specific things — how the database is structured, where it's slow, where security gaps are, where AWS spend is being wasted, whether your backups would actually work, where storage is leaking, where you're flying blind — and write you a plain-English report ranked by what's actually costing your business money.

Why I built it

A pattern I keep seeing — your database started simple and now runs eight things nobody fully understands. The engineering team senses something's off — slowness, creeping bills, a backup nobody's tested — but they don't have a clear week to step out of feature work. So I do.

Architecture · simplified

What makes it work

Findings are ranked by what they're costing your business in pounds, not by technical severity. The report has to make sense to your finance director as well as your engineer.
I never modify anything during the audit — it's read-only. If you want me to fix what I find, that's a separate engagement with explicit sign-off.
Every finding has a 'if we don't fix this, what happens' line. Severity without consequence is just opinion.
The cost review uses AWS's own attribution data, not just the bill — so we can see exactly where money is leaking.

What it doesn't do yet

Currently it's all manual delivery. I'd like to automate the easy diagnostics so you get the boring findings in the first day, leaving the rest of the week for the parts that need real judgment.
I'm considering open-sourcing the diagnostic checklist so any business can run a basic health-check themselves.

Defaults

Defaults I keep

Things I do on every project, regardless of brief — because these are the principles that prevent the kinds of failures I've seen most.

Single-tenant by default

Each customer gets their own deployment. Your data never sits on a shared server with someone else's.

Row-level security on every table

Even when records live alongside another tenant's, the database itself enforces that they can't cross over.

Strict checking at every data handoff

Bad data gets rejected at the boundary, before it can spread into the rest of the system.

Hard cost limits on every external service

Every paid API has a budget cap built in. No runaway spend, no surprise bills.

Tests written first where it matters

Not as a religion — but on the parts of the system where a bug would cost real money.

Stack

Stack & tools

The actual tools and services I build with. Deliberately short — anything that's promised the moon and then sent a surprise bill has been dropped.

Languages

TypeScript
Python
SQL

Frameworks

Astro
Next.js 15
Fastify
Tailwind CSS

AI / LLM

Anthropic Claude (Haiku 4.5, Sonnet 4.6, Opus 4.7)
Voyage AI embeddings
Anthropic SDK

Voice

Twilio Voice + ConversationRelay + SMS
ElevenLabs streaming TTS

Data & infra

Supabase (Postgres + pgvector + RLS + Realtime)
AWS (RDS, S3, Cost Explorer)
Vercel
Trigger.dev v4

Tooling

Doppler (secrets)
Lucide
Zod
pgAnalyze

Background

Double Master's in Engineering and Project Management. Spent most of my career inside production systems that have to keep running when nobody's watching — that's where my intuition for what's actually load-bearing comes from.

Co-founder and Technical Director of Ciatel Corp Ltd, registered in England & Wales. The company is the commercial vehicle; this site is the engineering one. They share a stack, a brand, and a phone number.

Most interested in: building AI products that actually work in real businesses (not just demos), keeping each customer's data inside their own systems, the unglamorous engineering that makes the difference between a demo and a working product, and what UK small businesses are quietly putting up with that no software vendor has bothered to fix.