In-Person AI Training in Sri Lanka: Inside Our Bootcamps and Workshops

By Ammar Ahamed, Founder, Cocoon · 29 October 2026

One Saturday in our first Applied AI cohort at Hatch Works, a few people hit the same wall. The meeting-notes app we were teaching asked for a work email, and not everyone had one they could use.

On a video call, that is where the lesson quietly ends. Cameras go off, someone types "will try later", and later never comes.

In the room, we saw it happen at the tables. That night I built a free alternative, Cocoon Notes, and shared it with the cohort the next morning.

That is the case for in-person AI training in one story. Not that the slides are better. That the problems show up where someone can do something about them.

This post is about the rooms we run in Sri Lanka: the bootcamps, the workshops and the on-site days for companies. Who has been in them, what a session looks like from the inside, what people walk out with, and how to join one.

Why learn AI in person when there are a thousand online courses?

We run online sessions too, and every session in our public cohorts is recorded. So this is not a fight between formats. It is about what each one is good at.

AI is a skill we learn with our hands. We learn it by briefing a model badly, reading what comes back, and trying again. The hard moments are small and practical: a login that fails, an automation that stops at step three, an answer that looks right and is wrong. Those are the moments where most self-taught learners give up.

In a room, a trainer sees the frown before the question. A neighbour from a different industry shows a fix nobody planned to teach. And the time is protected, because nobody can be pulled into "a quick call" from the next desk.

Our view: in person is better for starting and for building. Online is better for reach and for keeping going. The cohorts that work best use both, with the building done in the room and the recordings and office hours in between.

Where we have taught AI in person

Since 2025 we have taught AI in co-working spaces, boardrooms, newsrooms, hotel function rooms and packed halls. A few of those rooms:

The Applied AI Bootcamp at Hatch Works, Colombo. Eight Saturdays, delivered by Cocoon for BuildrLabs. Cohort 1 ran from July to September 2026 with a deliberately mixed room: founders, senior managers, people from sales, HR, finance and operations, and students. Nobody needed to code. By the build day, people were presenting working apps and agents. Cohort 2 is under way now, and the full story is in Applied AI Cohort 1, delivered by Cocoon.

The Derana creative teams. A bespoke programme in Colombo in early 2026, run over several weeks by me and Lishan Peiris and built around the teams' own workflows. It grew into our twelve-week AI for Creatives programme.

Lines Strategy. A hands-on session for the Lines team on how work is changing with AI, and what to do about it on Monday.

ICTA at Global Entrepreneurship Week. In November 2025 we walked a packed room through AI automation workflows and AI agents, live on the big screen.

The Singapore AI Association. Also in November 2025, an AI skills workshop with SAIA in Singapore. Same format, different city.

Build with AI Sri Lanka 2026. The Google Developer Groups event at Hatch Works, where I spoke on the AI for Industries track and our Saturday cohort came along as a class.

SA Ventures. In October 2026, a personal branding and LinkedIn workshop for SA Ventures leaders, taught by Affan Ahamed, where every participant rewrote their own LinkedIn presence before they left the room.

And corporate programmes we cannot name, run in client offices and hotel function rooms around Colombo. You can see more of these rooms in our gallery.

What happens inside a Cocoon session?

Every room we teach follows the same rhythm, whether it is a Saturday cohort or a day at a company's office. Our programmes name the three parts Teach, Studio and Crit.

Teach is short. One idea, shown live on the big screen, on a real task. Not a tour of features. If we are teaching automation, we automate something someone in the room does every week.

Studio is the longest part. Laptops open. Everyone applies the idea to their own work, at their own table, with trainers walking the room. This is where people hit walls, and where being in the same room earns its keep.

Crit closes it. People show what they built and the room responds. Watching a finance manager and a student solve the same brief in two different ways teaches more than another slide would.

A few practical things we have learned from running these rooms. A session runs for at least four hours, because anything shorter ends just as people get going. A fully hands-on day needs a proper break, so we plan six hours or more. Everyone brings a laptop, not a phone. And we check logins before the day, not at nine in the morning, because a room full of people resetting passwords is a lost first hour.

The best hours are sometimes not on the schedule at all. In our first cohort, the official finish time was usually when the second conversation started: about the businesses people were running, the ones they wanted to start, and what AI is doing to jobs here in Sri Lanka.

What do people actually build?

The test of any AI course is what you leave with. A feedback form is not the answer.

From our first Saturday cohort, built by people in sales, operations, HR and management rather than developers: an executive contact engine, written up as a full product spec with contacts pulled from Apollo into Airtable and an app on top. A licence-renewal tracker, the kind of internal tool that usually waits a year for an IT budget. A stock market explainer that helps people follow listed businesses without the jargon, which its builder kept improving after the cohort ended. Sixty-second AI video ads from the creativity week. Agents from the build day.

Underneath every cohort run six threads: foundations and your own AI stack; multimodal AI and custom assistants; creativity; automation, agents and apps; ethics and security; and personal branding, SEO and analytics. The capstones are concrete. A 60-second video ad. A live AI-powered web app at a real link.

What changed most was not the builds, though. It was the question people asked. They arrived asking "which tool should I use?" and left asking "what problem am I solving?" That habit keeps us current long after any course ends.

Which in-person format fits you?

If you are still comparing providers, our guide to AI courses in Sri Lanka lays out the options, and how to choose an AI training provider in Sri Lanka has the questions to ask any of them, including us. For the bootcamp format in general, start with what happens inside an AI bootcamp in Sri Lanka.

Who in-person AI training is not for

We would rather say this now than after you have paid. If you live far from Colombo and cannot travel on a Saturday, an online cohort with recordings will serve you better than a room you reach twice. If you want a formal credential to move into machine learning engineering, a university programme is the right path. And if you already build with AI every week, you need depth in one area, not a general bootcamp. Our free school and the Automation Lab are better next steps.

What people who trained with us say

“With AI evolving so quickly, I would definitely have my team train with them again.”

Anuki Premachandra, Founder & CEO, Lines Strategy

“Their sessions are well structured, engaging, and focused on real-world use cases.”

Nimila Jayawardena, AI Lead, Derana Macromedia

Both are Google reviews. Read them on Google.

Before you book a seat

Find out where you are starting. The free AI Readiness Score takes about six minutes at its quickest and shows whether you need foundations or depth. Bring the result to a free 15-minute call and we will tell you which room fits, even when the answer is a different one from ours.

And one small thing to do this week, whichever room you choose: write down one task you do every week that still runs by hand. That is what you will build on first.

Frequently asked questions