A focused team building AI agents — and the toolkit to build them faster.
We design and ship custom AI agents, voice agents, and automation for teams around the world — from a small engineering studio in Matale, Sri Lanka.
Most AI demos don't survive contact with production.
We started CleverHub in 2024 after enough client projects to see the same pattern repeating: AI demos look great in a sandbox and fall over the first time they touch a real customer, a real database, or a long-running task.
What ships in production usually isn't a model — it's the boring infrastructure around it. Eval pipelines that catch regressions before users do. Tool-calling that retries correctly when an API hiccups. Cost controls that don't surprise you at the end of the month. Observability that tells you which prompt path failed, not just that something failed.
Agents are non-deterministic, so the discipline around them matters more than the model choice. We treat an agent like the distributed system it really is: scoped narrowly, instrumented end to end, and tested against real cases — so it keeps working after launch, not just in the pitch.
So that's what we build. Custom agents and voice agents for clients today, and a small toolkit of products that bake those patterns into software tomorrow.
What we believe
Four principles that decide how we work and what we ship.
Agents that actually ship
Production observability beats demo polish. We build for the messy real world, not the sandbox.
Honest timelines
3–6 weeks means 3–6 weeks. Padding is dishonest; missed dates are worse. We scope tight and deliver.
Boring tech that works
Battle-tested infrastructure under the model layer — Postgres, FastAPI, Redis. New and shiny only when it is genuinely better.
Customers over investors
Funded by client work, not raises. Decisions optimise for client outcomes, not growth-deck metrics.
How we work
We're async-first and written-first. Decisions live in docs, not in meetings nobody remembers. Progress is shared in short recorded updates and clear written notes, so you always know where your project stands without sitting on a call to find out.
That style protects deep work — the long, uninterrupted focus that good engineering actually needs — and it keeps a small team moving fast for clients in very different timezones.
Where we are
We're based in Matale, Sri Lanka, and we work with clients globally. Being here is an advantage, not a footnote: our hours overlap comfortably with US East, Europe, and APAC, so a question asked in one timezone is usually answered before the next workday starts.
Sri Lanka also has a deep, growing pool of engineering talent and a cost-to-quality ratio that lets a focused team deliver senior-level work without enterprise-agency overhead. If you've worked with distributed teams before, you already know the model works — we just run it tightly.
We hire occasionally
If you're a strong engineer in Matale or remote-async and this sounds like your kind of team, reach out — we're always glad to meet good people.
Let's build something that runs in production.
A 30-minute call to scope your project. If we're a fit, we'll send a proposal within 48 hours.