Enterprise AI keeps failing for a reason nobody usually schedules a conference around. The model works. The business context underneath it does not exist in a form the model can use. Atlan hosted three hours on that problem online on October 28, 2026. It was announced as Context/26 and took the name of the thing it is about.
The problem is not academic. According to S&P Global Market Intelligence (2025), 42% of companies scrapped most of their AI initiatives in 2025, up from 17% the year before. So the agenda asked one question the industry has not settled: what does it actually take to give AI systems the business context they are missing? Eight sessions worked through it, from the business case for context to a debate on open context interchange. For what the speakers said and what Atlan released, read the Context Conference 2026 recap.
- A virtual, three-hour community conference on AI context, hosted by Atlan on October 28, 2026
- Separate from Atlan Activate, which is Atlan’s product and customer conference
- Eight sessions, from the business case for context to an open-standards debate
- Built for AI leaders, CIOs, CTOs, CDOs, and enterprise architects
- Every session is now on demand, with a full transcript
| Field | Detail |
|---|---|
| What it is | A virtual community conference on AI context, hosted by Atlan |
| When | October 28, 2026, about three hours; now on demand |
| Format | Eight sessions: keynotes, an executive panel, an architecture deep dive, product demos, a debate |
| Previously called | Context/26, renamed to the Context Conference |
| Not to be confused with | Atlan Activate, Atlan’s separate product and customer conference |
| Who it is for | AI leaders, CIOs, CTOs, CDOs, enterprise architects |
| Recap | Context Conference 2026 recap |
Why does the Context Conference exist?
Model quality stopped being the constraint on enterprise AI a while ago. Frontier models write production code, pass graduate exams, and power consumer products millions of people already trust with real work. What breaks is what happens next, inside one specific company, when the model has to act on that company’s own knowledge instead of the open internet’s.
The numbers say so. S&P Global Market Intelligence (2025) found the share of companies scrapping most of their AI initiatives before production rose to 42% in 2025 from 17% a year earlier, across a survey of more than 1,000 enterprises in North America and Europe. MIT NANDA’s “The GenAI Divide” (2025) put roughly 95% of enterprise generative AI pilots at no measurable profit-and-loss return. That second report is preliminary and has been criticized over its measurement window, so treat the exact figure as directional.
Neither number describes a model problem. Both describe a context problem: the model has no reliable way to know how a business works, what its terms mean, or what has already been decided.
In practice that gap is very concrete. An agent asked about churn cannot tell which of four revenue tables is the certified one, which customers the business counts as active, or that the definition changed last quarter. Closing it is why teams now build a context graph or an enterprise context layer, and why the question of who owns that layer keeps arriving on a CIO’s desk unresolved.
Atlan calls the answer to this contextual intelligence. The conference exists because the people solving it are solving it alone, in parallel, without a shared vocabulary, and repeating each other’s dead ends.
How is the Context Conference different from Atlan Activate?
Atlan already runs Activate, its product and customer conference. The Context Conference is not a rebrand of it or a second edition. They do different jobs.
| Context Conference | Atlan Activate | |
|---|---|---|
| What it is | Community conference on AI context as an emerging discipline | Product and customer conference |
| Focus | Industry-wide vocabulary, ownership, and production patterns | Product announcements and customer stories |
| Sponsored content | None, by design | Product-focused, as you would expect from a vendor event |
| Audience | Anyone building or thinking seriously about AI context | Primarily existing customers and prospects |
The distinction is the point. The Context Conference centered on debates, panels and open architecture walkthroughs rather than a keynote-and-breakout format, though Atlan also used the day to release Atlan Trident, the AI Registry spec and the Forward Deployed Playbook. For Atlan’s broader product news, read the enterprise AI questions Activate 2026 surfaced. If you want the industry’s current best thinking on context, this is the one.
A simpler way to choose: Activate answers what the vendor is shipping. The Context Conference answers what the industry has not agreed on yet. If your open question is how to implement an enterprise context layer, or why agents need one at all, rather than which Atlan feature covers it, the recordings are the place to start.
What was on the Context Conference agenda?
Eight sessions in about three hours. The order moved from why context is worth funding, through one real architecture and Atlan’s own releases, to whether context can move between platforms. The final agenda differed from the one announced in August; this is what aired.
| Session | Who |
|---|---|
| Opening keynote: Enterprise AI is missing a context layer | Prukalpa Sankar, Atlan |
| Executive panel: Why invest in context? | Cindy Hoots, Leigh-Ann Russell, Prakash Kota |
| Inside Verizon’s context architecture | Steven Huang, Verizon |
| The company brain for shared skills and ontologies | Rohan Goel and Ashwin Murali, Atlan |
| The Post-AI Data Stack: a blueprint for analytics agents | Zach Michel, Atlan, with Workday, Upwork and DocuSign |
| Open sourcing a spec for AI registries | Ben Hudson, Atlan |
| Debate: Will open context interchange ever be real? | Bob Muglia with Snowflake, Google, Microsoft and Anthropic |
| Recap of the day | Austin Kronz, Atlan |
Three throughlines stood out.
The business case came from the buyer’s side. Cindy Hoots moderated Leigh-Ann Russell of BNY and Prakash Kota of UKG on what context is for: shared meaning across a business, people who stay accountable for AI outcomes, and bets that can be reversed when tools change.
One architecture got opened up. Steven Huang walked through the Enterprise Context Platform Verizon is building to serve context to every agent studio it runs, including the gaps, such as the missing standard for authoring agent operating procedures.
The vocabulary debate became an interchange debate. The closing debate put Snowflake, Google, Microsoft and Anthropic speakers on the same question: whether governed definitions belong in a semantic layer, a knowledge graph or an ontology, and whether they can move between platforms. The panel agreed on open formats and split on whose engine should hold the governed version.
Who spoke at the Context Conference?
Eighteen speakers across eight sessions, from practitioners running context in production to the people building the open standards. Several speakers announced in August did not appear on the final agenda; the list below is who spoke.
| Speaker | Role at the conference |
|---|---|
| Prukalpa Sankar | Atlan co-founder. Opened the day with the keynote |
| Cindy Hoots | Former Chief Digital Officer and CIO at AstraZeneca. Moderated the executive panel |
| Leigh-Ann Russell | CIO and Global Head of Engineering at BNY. Executive panel |
| Prakash Kota | Chief Transformation Officer at UKG. Executive panel |
| Steven Huang | Lead Enterprise Architect at Verizon. Walked through Verizon’s context platform |
| Joe DosSantos | VP, Enterprise Data and Analytics at Workday. Analytics agents panel |
| Bob Muglia | Former CEO of Snowflake. Coined “open context interchange” and moderated the debate on it |
| Josh Klahr | Snowflake. Debate panelist on semantic models and Apache Ossie |
| Prajakta Damle | Google. Debate panelist on Open Knowledge Format |
| Dipti Borkar | VP, Microsoft IQ and OneLake at Microsoft. Debate panelist |
| Amandeep Khurana | Member of Technical Staff at Anthropic. Debate panelist |
| Austin Kronz | Field CDAO at Atlan. Hosted the day and closed it with the recap |
The executive panel is the session to watch if you are trying to get context funded. Hoots, Russell and Kota talked about meaning, accountability and reversible decisions rather than models, which is a different conversation from what the technology can do.
Who should watch the Context Conference recordings?
This was a leadership agenda. It assumed you had already tried to put AI into production and hit the context wall, so nothing on the schedule stopped to define an LLM.
| Role | What you get from the day |
|---|---|
| AI leaders | The keynote’s case for managing context like code, and the open-standards debate |
| CIOs, CTOs, CDOs | Enterprise technology leaders on meaning, accountability and reversible bets |
| Enterprise architects | Verizon’s context platform architecture in detail, and the open interchange debate |
| Data and platform leaders | What analytics agents need: metrics, an ontology, skills and an accuracy bar per use case |
Worth doing before you watch: write down your own one-line definition of a context layer. The open-standards debate is far more useful if you already know which camp you have been arguing from, and whether you treat a context graph and an ontology as the same decision or two.
If you are earlier than that, the context layer glossary and what context engineering actually is are better starting points, along with four context engineering strategies and a context engineering framework to test them against. The recordings are not going anywhere.
Why Atlan is the one hosting this
Fair question to ask of any vendor-hosted event. Atlan’s answer is that it has been arguing the context case in public for long enough to have earned the room, and it gives that room away rather than filling it with product.
Atlan builds the context layer for AI: the active metadata platform that gives agents the business context they need to work with enterprise data. Its MCP Server is how agents reach that context in practice, and even there the interesting part is contested rather than settled: what MCP actually delivers to an agent, and when an MCP server beats an API, are both live arguments. That is the same territory the agenda covered. Bob Muglia coined “open context interchange” at Atlan’s Great Data Debate, an earlier Atlan-hosted argument, and came back to moderate the debate on whether the idea can actually work.
The other reason is coverage. Atlan co-commissioned MIT Technology Review Insights’ first independent study of the AI context gap, now published as What’s really holding enterprise AI back?. Independent research on your own thesis is a slower way to make a point than a keynote, and a harder one to dismiss.
Real stories from real customers: context as a strategic asset
Two enterprises already treating context as something to own deliberately, ahead of any conference telling them to. They prove different halves of the argument. Workday’s is language: a vocabulary already agreed internally, then exposed to agents rather than rebuilt for them. Mastercard’s is scale, and the metadata lakehouse underneath it holding hundreds of millions of assets.
"Atlan captures Workday's shared language to be leveraged by AI via its MCP server. As part of Atlan's AI labs, we're co-building the semantic layer that AI needs."
Joe DosSantos, VP Enterprise Data & Analytics, Workday
"AI initiatives require more context than ever. Atlan's metadata lakehouse is configurable, intuitive, and able to scale to hundreds of millions of assets. As we're doing this, we're making life easier for data scientists and speeding up innovation."
Andrew Reiskind, Chief Data Officer, Mastercard
Context is becoming its own discipline, and this is where it gets named
Every enterprise moving from a handful of agents to dozens hits the same wall. The model works. The business context underneath it does not exist in a form the model can use. The Context Conference put the people solving it at different companies in one room for three hours. The day closed with a recap rather than a ratified Context Contract, and the open-standards debate ended with its panelists agreeing more than they expected. What they did not settle, whose engine holds the governed definition, is the question to watch next; the recap covers each session.
FAQs about the Context Conference
1. What was the Context Conference?
The Context Conference was Atlan’s virtual community conference on AI context, held on October 28, 2026 over about three hours. Eight sessions brought together AI leaders, CIOs, CDOs, enterprise architects and platform vendors to work through what a context layer is, who owns it, and what it looks like in production. Every session is now on demand.
2. Is the Context Conference the same as Atlan Activate?
No. The Context Conference is a separate community conference about AI context as an emerging discipline, built around debates, panels and architecture walkthroughs. Atlan did use it to release Atlan Trident, the AI Registry spec and the Forward Deployed Playbook. Atlan Activate is Atlan’s own product and customer conference, built around Atlan’s roadmap and customer stories.
3. Was the Context Conference previously called Context/26?
Yes. The event was announced as Context/26 and renamed the Context Conference. The date and format did not change: October 28, 2026, virtual. Older links and coverage that use the previous name refer to the same event.
4. Who spoke at the Context Conference?
Speakers included Atlan co-founder Prukalpa Sankar, former AstraZeneca Chief Digital Officer and CIO Cindy Hoots, BNY CIO Leigh-Ann Russell, UKG Chief Transformation Officer Prakash Kota, Steven Huang of Verizon, Joe DosSantos of Workday, and Bob Muglia, former CEO of Snowflake, with panelists from Snowflake, Google, Microsoft and Anthropic.
5. Who should watch the Context Conference recordings?
The recordings suit AI leaders, CIOs, CDOs, enterprise architects and data platform leaders deciding how their organization handles AI context, ownership and production architecture. The agenda assumed you had already tried to put AI into production, so it is a leadership session rather than an introduction for newcomers.
6. Can I still watch the Context Conference?
Yes. All eight sessions are on demand on the event page at atlan.com/context-conference, each with a full transcript. The MIT Technology Review Insights study on the AI context gap is published, and the Context Conference 2026 recap summarizes what each session covered.
Sources
- Atlan, “Context Conference,” event page, agenda and session recordings, accessed October 2026. https://atlan.com/context-conference/
- S&P Global Market Intelligence, “Generative AI shows rapid growth but yields mixed results,” October 2025. https://www.spglobal.com/market-intelligence/en/news-insights/research/2025/10/generative-ai-shows-rapid-growth-but-yields-mixed-results
- MIT NANDA, “The GenAI Divide: State of AI in Business 2025,” August 2025 (preliminary research). https://www.forbes.com/sites/jasonsnyder/2025/08/26/mit-finds-95-of-genai-pilots-fail-because-companies-avoid-friction/
- Foundation Capital, “AI’s Trillion-Dollar Opportunity: Context Graphs,” Jaya Gupta and Ashu Garg. https://foundationcapital.com/ideas/context-graphs-ais-trillion-dollar-opportunity
