The Context Gap: What It Takes to Make AI Work at Scale
Anyone can get AI to do something impressive. Almost nobody can get it to do their actual job.
The distance between those two things is the most expensive problem in tech right now, and the people who have truly closed it can be counted on one hand. Three of them are spending Friday afternoon in Boston telling you how.
An official Boston AI Week 2026 session, presented by Misneach.
Why Your AI Is Stuck
It is not the model. Your AI does not know how your business works. Where the numbers come from, what your own terms mean, which version of the truth is the right one. That knowledge sits in people's heads and old spreadsheets, not anywhere the AI can reach. Nine out of ten companies are in the same position.
The missing piece is called context. It is the difference between AI that is impressive in a demo and AI you can hand real work to.
Who Is On Stage
Robert Caulk, Founder and CEO at AskNews
Robert is the founder and CEO of AskNews, which hosts the largest licensed news knowledge graph in the world. He started AskNews in 2023 to build an AI-first news data delivery system for frontier AI labs, governments, hedge funds, insurers and intelligence groups. He coined the term context engineering in early 2024, more than a year before the field formalized it. He holds a PhD in Computational Mechanics and has more than 1,600 academic citations.
Mo Aidrus, Co-Founder and CEO at ekai
Mo is the co-founder and CEO of ekai, a platform that turns enterprise analytical data and expert knowledge into business context for AI to work. He and his team are focused on the meaning gap and the context rot that keep capable models from understanding how a business really runs. Before ekai he was co-founder and CEO of Rayn, building production-grade AI platforms for large enterprises, and spent 20 years at Accenture, most recently as a Managing Director in its Global Communications, Media and Technology practice in Boston and Singapore.
Mark Coffey, Co-Founder and Managing Partner at Misneach, Co-Founder and CEO at Cambridge Artificial Intelligencer
Mark is the co-founder and managing partner of Misneach, a founder propulsion platform that pairs capital with the next generation of capabilities for early-stage companies, including an in-house earned media studio that builds founder narratives from day one. He is also co-founder and CEO of the Cambridge Artificial Intelligencer, a civic technology nonprofit building AI infrastructure for local newsrooms. Before Misneach he spent two decades across media and startups in Europe and the US, and he wrote The Model Is the Easy Part, the essay this session is built on.
Judah Phillips, Founder and Board Chair at Boston AI Week and SmartCurrent, Chief AI and Product Officer at Market Holdings
Judah founded Boston AI Week and chairs its board. He is Chief AI and Product Officer at Market Holdings, a PE-backed global holding company with thirteen brands and thousands of employees, and the founder of Squark AI, the no-code predictive AI platform he sold to Market Holdings in 2024. He has worked in AI since 1997, has written three books on AI and analytics, has taught at Boston University and Babson College, and founded the Boston Generative AI Meetup, one of the largest in-person AI communities in the country. He is moderating.
Two short keynotes, a panel, then the room, and the last stretch is yours to ask any of them anything. Nobody on stage is selling anything. No product pitches, no demos, no vendor slides. They also do not fully agree with each other on how much human oversight AI still needs, and Judah is going to push on exactly that.
What You Leave With
Three questions to ask your own team on Monday that tell you whether your data is ready for AI at all
Why asking AI the same question twice can give you two different answers, and what actually fixes that
Why a smarter model will not fix it, and where the money should go instead
How to get what your best people know out of their heads and into something AI can use
What this work costs and how long it takes, from people who have done it
How much human checking AI still needs, and who in your company should be doing it
The Last Stop Before the Party
This is the final session of the day, and Boston AI Week's opening party starts at 5:30 at Island Creek Oyster Raw Bar in the Seaport, running until 8:30. We wrap at 5:15 and head over together, about fifteen minutes on foot or by car, which lands everyone right at the start.
The party is free and takes its own RSVP, so grab a spot for that too.
So you get an afternoon worth your Friday, and you arrive at the party with a room of people you have just spent it thinking alongside.
Friday, September 25. 4 PM - 5:15 PM. 99 High Street, 3rd floor, Boston. Free, 100 seats, RSVP required.
Any questions? Email nandini@misneach.ai.
The model is the easy part. Build the rest.