Limitless Academy is based in Bengaluru, which means a large share of the sessions I run are for companies within a short drive. Over the past year that has produced a fairly clear picture of what local teams want, and it is noticeably different from what companies elsewhere ask for.
Bengaluru teams start further along
In most cities, the first block of an AI training has to establish that these tools are useful at all. In Bengaluru that block can usually be compressed, because a meaningful share of the room is already using AI tools daily, often without any company policy telling them how.
This changes the useful shape of a session in three ways.
Less demonstration, more construction. A room that already believes the tools work does not need to be convinced. It needs floor time to build something on its own data. The build block, where each person ships one working workflow, becomes the centre of the day rather than the finale.
Deeper down the stack. The questions here go past prompting quickly: how do I connect this to our internal systems, how do I make it run without me, how do I stop it doing something stupid at 2am. That is agent and automation territory, which is why n8n, Lovable and Replit come up more in Bengaluru sessions than anywhere else.
Wider internal gap. The same company that has engineers running local models also has a finance team that has never opened ChatGPT. The gap inside one building is often larger than the gap between cities, which makes role specific tracks less of a nicety and more of a requirement.
What local teams are actually asking to build
The requests that come up most often in Bengaluru rooms, roughly in order:
- Internal document Q and A. Policy manuals, past proposals, product documentation, all loaded into something that answers with citations. NotebookLM does most of this with no engineering effort, which surprises people.
- Inbox and ticket triage. Something that watches a shared inbox or queue, extracts the fields that matter, and routes or logs them. This is usually the first n8n automation a team builds and the one most likely to still be running at day 30.
- Meeting and call analysis. Transcripts into structured summaries, action items, and in sales teams, objection patterns across accounts.
- Internal tooling prototypes. Small tools that would never survive a sprint prioritisation meeting, built in an afternoon with Lovable or Replit. Often the highest morale outcome of a session.
- Reporting drafts. Variance commentary, weekly updates, board pack first drafts, grounded in the actual numbers rather than invented.
Notice that none of these are exotic. The value is not in sophistication, it is in the fact that somebody built the thing and it now runs every week.
How sessions are usually structured here
Most Bengaluru engagements fall into one of three shapes.
Single campus, single day. On site at the client office, a shared foundation in the morning, role specific tracks and a build block in the afternoon. This is the most common and generally the most effective.
Leadership offsite. Shorter, higher altitude, focused on where AI actually changes the operating model rather than on tool mechanics. Usually runs at a hotel as part of a larger agenda. Works best when paired with a hands-on session for the teams afterwards, otherwise leadership leaves convinced and nothing downstream changes.
Bengaluru flagship plus virtual elsewhere. For companies with offices in Hyderabad, Pune, Gurugram or Chennai, running the flagship on site here and delivering the rest live virtual keeps cost sensible without losing the energy of a real room for the largest group.
What to plan for before booking
- Confirm laptops and accounts. The single most common thing that damages a hands-on day is participants without working accounts for the tools, or laptops locked down such that nothing can be installed or accessed. Sort this a week in advance with IT, not on the morning.
- Decide the data rules first. What can go into a public tool and what cannot. This conversation needs to happen before anyone touches anything, and having a position already agreed makes the session considerably better.
- Pick the functions, not the headcount. Two or three functions trained deeply beats the whole company trained thinly. The second group will ask to be included once they see the first group's output, which is a much better position to be in.
- Book the day 30 reconvene now. Before the training, not after. Scheduled afterwards it never happens, and week two is where adoption dies.
- Bring real tasks. Ask every participant to arrive with one task from their own job that they currently dislike. That task is the curriculum.
Context on what I run
Sessions from here have included Coca-Cola in Guwahati, Lenovo Tech World at the Jio World Convention Centre in Mumbai, HP in Delhi, EPAM in Coimbatore, IIFL Capital at the Datum AI conference in Mumbai, the ICAI AI Innovation Summit at Bharat Mandapam, the Department of Economy and Tourism in Dubai, and Mesa School of Business here in Bengaluru. Roughly 25 companies and more than 10,000 people so far.
The Bengaluru sessions are consistently the ones where the build block runs long, because people keep going after the timer. That is usually a good sign.
If you are planning something for a team here, the scoping call is the fastest way to work out which of the three shapes above fits.