The AI Automation Summit 2026 was held on Friday at the Regency Hotel in Dhaka, with 250 entrepreneurs, business owners and professionals attending. The day-long event was organised by MyAI and the Idean Center for Business Excellence, and its stated aim was to show how artificial intelligence can be used across business functions — and how a company moves from first using AI tools, through AI agents and automation, towards an AI-led business.
The trainers included Mohammad Saif Noman Khan, associate professor at the Institute of Business Administration at Dhaka University and co-founder of MyAI, and Mark Anupam Mallick, founder of Idean Consulting and co-founder of MyAI. They were joined by Adnan Rafid, consulting partner and COO at Idean Consulting and a director of MyAI; Mehedi Hasan, founder of Quick Growth and co-founder of ICBE; and Shafat Kadir, founder of Hatbaksho and co-founder of ICBE.
Sessions covered content planning and marketing, sales and customer management, business research and intelligence, product development, team productivity, AI agents, workflow automation and the AI tech stack, with worked examples. An evening segment from six o'clock brought industry figures to discuss AI's use and prospects in their own fields.
What the curriculum tells you
Read the session list as a diagnosis rather than an agenda and it says something useful about where Bangladeshi firms actually are.
Nothing on it is technical. There is no model training, no fine-tuning, no infrastructure. It is content, sales, research, product and workflow — the ordinary operating functions of a small or mid-sized company. That is the correct level, because the binding constraint for most firms here is not model quality. It is that nobody inside the business has mapped which of its repeated tasks could be handled differently.
The framing — tools, then agents, then automation, then an AI-led business — is also the right order, and it is the order most companies skip. Firms commonly try to automate a process they have never written down. Automation applied to an undocumented process produces a faster version of whatever mess was there.
The question a 250-person summit cannot answer
A day of training produces enthusiasm reliably and change unreliably, and the gap between them is where most of this spending goes.
The useful measure is not how many attended but how many have a workflow running sixty days later that they did not have before. That number is almost never collected, and it is the only one that distinguishes a training business from a training outcome.
For an owner who attended, the step that converts the day into something durable is narrow: pick one task that happens at least weekly, is written down, and has a clear definition of done — invoice chasing, order confirmations, first-draft responses to supplier emails — and automate only that, this month. One working automation teaches more about the second than any further session would.
The tasks that actually suit this, in a Bangladeshi firm
Automation pays where a task is frequent, rule-shaped and tolerant of a human check. In most firms here that points at a short and unglamorous list.
Drafting replies to repeat customer questions, where the same twenty questions arrive weekly. Turning a supplier's price list into the format your system expects. Producing the first version of a product description in both Bangla and English. Reconciling an order sheet against a delivery sheet and flagging only the rows that disagree.
What does not suit it, and is attempted constantly: anything where being wrong is expensive and nobody checks the output. A model drafting an invoice is useful; a model sending one is a liability.
Where this sits
This is the demand side of a market whose supply side was in Jakarta the same week, at the regional ASOCIO summit. Locally built AI products aimed at exactly these buyers are appearing — the Bangla AI interview platform is one. The argument about where the sector should be heading is in the BPO summit piece.
