AI
From prompt to product: How the Event Strategy Bot was born
Dr Barış Onay shares a unique AI journey of discovery with EW readers
didn’ t set out to‘ build
I a bot’; I set out to stop repeating myself.
With my consulting hat on, I find myself working on multiple projects across organisers and sometimes private equity investors. One day, I’ m double-checking the outside-in analysis of an M & A target, and the next, I’ m helping an organiser assess a launch or the reposition of a mature show. For my internal thinking, each brief starts the same way: I need to produce a quick and dirty, outsidein view of an event’ s competitive position, credible enough to provide a starting point, and fast enough to keep moving at scale.
For years, I did this mostly manually. I had my checklists and a mental model I could run in my sleep: What’ s the job to be done at this event? Who is selling? Who is buying? How has it evolved? Who else plays in this space? Where is demand shifting? Which sub-sectors are the real engines of growth? What would we do in the first 90 days? I found myself rewriting versions of the same questions and frameworks, project after project.
That was the itch.
A helper, not a‘ product’ The first version of what became the Event Strategy Bot was remarkably basic: a set of prompts I used inside my own AI workspace to structure those outside-in sprints. I’ d paste a URL or two, ask very specific questions, and force the system to answer in my
Above: Dr Barış Onay
“ Over time my prompt became an ingrained process, part checklist, part script, that reliably produced the framework of an event strategy brief in minutes rather than hours”
Below: Barış and Bot
preferred format and syntax. It wasn’ t elegant, but it saved me time and, crucially, standardised my starting point. The outputs were rough and erroneous, but they were consistent.
After many iterations, these separate prompts merged into a single one, creating a significantly longer process. One day, I wrote the entire thing from scratch and asked my AI space to memorise and name it. With hindsight, that was the breaking point.
I kept returning to that prompt. I would reuse it, adjust a line here, add a constraint there. Over time, it became an ingrained process, part checklist, part script, that reliably produced the framework of an event strategy brief in minutes rather than hours.
Refining the structure I iterated on that prompt every time it met reality. When I noticed inconsistencies in competitor tables, I tightened the schema. When sectors appeared fuzzy, I added rules for parsing product categories and determining where to look for available data. When it fabricated critical information, I imposed more limits on its‘ creativity’. I introduced a short‘ Where to play / How to win’ synthesis to turn diagnosis into direction.
During this accelerated iteration phase, two things made the difference.
First, structure. By insisting on a single, tight format, I could compare very different shows on a like-forlike basis. That made portfolio conversations easier. PE partners could stack opportunities; organisers could see white space; marketing teams could spot the content angles that convert.
Second, guardrails. The Bot is designed to use only publicly available information, to cite what it’ s looking at, and to surface uncertainty rather than gloss over it. It’ s not there to replace actual strategic judgment. It’ s there to accelerate it, reduce variance, and make sure we never forget to ask the obvious next question. www. exhibitionworld. co. uk Issue 5 2025 51