From the AI Agent & Copilot Summit — La Jolla, March 2026
I spent two days this week at the AI Agent & Copilot Summit in La Jolla. What made this Summit a genuinely useful window into the state of agentic AI was who was in the room. The event is built around four pillars of the AI business ecosystem: Microsoft as the central builder of AI platforms and tools; partners bringing those solutions to market; enterprise buyers actively investing in AI to modernize their businesses; and the practitioners and users putting these systems to work every day.
That cross-section: platform, partner, buyer, and user, all in the same room at the same time, is rarer than it sounds. Most conferences optimize for one audience. This one brought the full stack together, which meant the conversations had real stakes and real friction. Not a vendor roadshow. Not an analyst briefing. A working session across every layer of the people actually building, selling, buying, and running agentic AI in production.
Real companies. Real deployments. Real results. Paragon Films. WB Mason. Mater Hospital. Lindsay Corporation. Vixxo. Dozens more.
Which is exactly what made one moment stand out.
During a session on AI transformation, a Microsoft executive turned to a partner on stage and asked directly: “Do you see yourself shifting from selling software to selling outcomes?”
I watched what happened next. The partner, experienced, smart, well-prepared, gave an answer that sounded like an answer. Described their current model. Talked about value delivery. Outlined the roadmap. But never actually answered the question.
“They described what they sell. Not what they deliver.”
It was a small moment on a La Jolla stage. But it captured exactly the crossroads that enterprise vendors and their customers are now navigating in real time.
What makes that moment worth writing about is the backdrop. The market has been pricing in this question for months, and it hasn’t been pretty.
Salesforce is down roughly 40% from its 52-week high. Workday off 40% or more. ServiceNow, which has beaten earnings nine consecutive quarters and is still growing at 21%, down approximately 50% from its highs. Adobe off 35%. The iShares Software ETF down more than 22% year-to-date. Close to $1 trillion in enterprise software market value erased since mid-2025.
These companies aren’t being punished for weak execution. Salesforce just reported $41.5 billion in revenue. The market is repricing their future, specifically, what happens to per-seat subscription businesses when AI agents start doing the work of the humans who were paying for those seats.
That’s not a valuation correction. That’s a business model verdict. And the people building and deploying in this room are the ones delivering it.
What the Summit Actually Showed
The depth and breadth of sessions across the three days was instructive in ways that went beyond any single moment. For business leaders looking to harness agentic AI — not just evaluate it — five things stood out as actionable takeaways.
* Humans as Orchestrators
Vixxo’s COO and CTO presented one of the more compelling frameworks of the Summit: humans are becoming orchestrators of agents, not operators of software. The implication is significant; your workforce isn’t being replaced, it’s being repositioned. The skill shift is from doing the work to directing the agents that do the work. That’s a talent strategy, a training strategy, and a change management challenge all rolled into one.
* The ROI Window Is Compressing
Multiple sessions made clear that tolerance for multi-year AI transformation timelines is evaporating. Enterprise customers aren’t buying vision anymore. They’re buying outcomes they can measure inside a quarter. One session was provocatively titled “Launching Quick AI Wins to Deliver ROI in 90 Days”, and the room was packed.
* Data Is the Actual Moat
Across the Azure AI Foundry, Data and Fabric, and ERP tracks, one theme was consistent: the organizations winning with agentic AI are the ones with clean, connected data. The session on WB Mason’s transformation, with Microsoft and BDO, made clear that agentic transformation isn’t primarily a technology problem. It’s a data readiness problem that technology then solves.
* Governance Isn’t Optional
An entire track was dedicated to governance, security, and agent lifecycle management. The organizations moving fastest aren’t ignoring guardrails, they’re the ones who built governance into the architecture from day one. “From Chaos to Control” and “The Agent PMO” weren’t peripheral sessions. The rooms were full.
* Reskilling Is the Unasked Question
The “Reskilling for Human Relevance” panel may have been the most important conversation of the Summit. Agentic AI doesn’t eliminate human judgment, it changes where and how it’s applied. The organizations that figure this out first will have a durable advantage over those still debating whether AI is a threat.
The Two Timelines Running in Parallel
Here’s the nuance that business leaders need to hold simultaneously: agentic AI is both accelerating the pressure toward outcome-based models and complicating the path to get there.
Accelerating
When an agent autonomously completes a workflow end-to-end; resolves the ticket, closes the loop, processes the transaction, the value is no longer abstract. You can see it, measure it, and price it. The logical unit of value shifts from the user with access to the outcome delivered. That logic is clean and it’s coming regardless of what any vendor decides.
Complicating
Defining “outcome” across complex, multi-system enterprise environments is genuinely hard. Who owns the result when three agents from three vendors contributed to it? The closing session of the Summit, a UCSD professor presenting “What Often Goes Wrong” in AI implementation, was a reminder that the gap between demo and deployment is still wide.
The shift is happening. The question that remains, is who owns the definition of “success”, and who gets to price it.
What I’m Taking Away
The Microsoft executive who asked that question on stage wasn’t trying to corner anyone. I think it was a genuine question from a platform vendor trying to understand how far their partner ecosystem has actually moved in their thinking. The answer they got was the answer most of the industry gives right now: careful, competent, and constructed to protect the current model.
The practitioners in this room, the CIOs from manufacturing companies, the COOs running field service operations, the IT leaders deploying agents into healthcare workflows, they’re not waiting for vendors to answer the outcomes question. They’re building the answer themselves, one production deployment at a time.
That’s the shift. And this Summit was a live look at it in motion.
What are you seeing in your own organization? Are your technology partners answering the outcomes question, describing their business model? Hit reply and let me know.And if you weren’t at the Summit this year, it’s worth putting this one on your calendar for next year.
Keep going: The 25% Problem and From SaaS to Service as Software
The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself.
“Uphoff on Media” is published by Tony Uphoff, founder of Uphoff Advisory, a strategic advisory practice for founders, CEOs and investors in B2B information, marketing and technology. The businesses that drive business.
