Is Your AI Agent Just the Modern User Interface?
The Value of Context and Outcomes Is a More Important Topic
VOLUME 1 - ISSUE 21 ~ MAY 19, 2026
In this edition of the “CIO Two Cents” newsletter, I take a look at why I believe AI will not be defined by who builds the most agents, but by who delivers the most meaningful outcomes through context and execution.
— Yvette Kanouff, partner at JC2 Ventures
The JC2 Ventures team: (John J. Chambers, Shannon Pina, John T. Chambers, me, and Pankaj Patel)
(1)
AI agents are becoming the new interface layer while enterprise context and execution become the real differentiators.
(2)
The market is shifting from per seat SaaS models toward outcome-based value and measurable business impact.
(3)
Companies that operationalize real time context, governance, and workflow orchestration will lead the next era of enterprise AI.
We’ve all heard endlessly about the “death of SaaS,” but I’m more curious about what’s replacing it.
One thing that is certainly a major shift: customers now expect to see and measure results. Outcome based pricing is certainly a customer goal, but—hey—if your software is good, this should be an advantage, not a headwind. Consumption (or token) pricing is another post-SaaS pricing direction. This is an area we’re all watching to see what will stick in the end, but let’s look at a few shifts:
Data is Still King—but Context is the True Crown
Data has always been considered “king,” but I would say that in this AI-driven era, its value has only increased.
In the early AI days, my conversations centered around prompt engineering and LLMs. Now, we’re talking much more about context engineering and SLMs. Having a system that knows something about everything is impressive. But for us to optimize and create outcome-based value, we need to know everything about THIS specific thing. That’s the shift.
The power of LLMs—creating answers instead of search-based links—continues to be game-changing, and personalization will continue to evolve (what a fun journey!). But for the sake of SaaS and enterprise software solutions, context engineering is huge.
The value is not necessarily in the bot or AI agent. It is in having the right transactional context, real-time (RAG) information, and historical context of successful and authorized resolutions. Let me repeat what’s probably a controversial statement in today’s world: the value is not the fact that there is an agent—the value is in the context and what that agent can do given that context.
Agentic systems are great because they simplify interactions, and even in today’s agent-to-agent and multi-agent world, context remains the true differentiator. Splitting up tasks (multi-agent) and interacting on behalf of tasks (agent-to-agent) are only as effective as the context they operate in. Similar to hallucinations, we do not want decisions to be made outside our desired range of risks and expected outcomes. Context is what grounds AI systems and keeps them aligned.
Sometimes we focus too much on having an agent—we should be focused on having a good agent and having a system that solves a problem the way you would solve a problem based on:
• Historical context
• Data specific to you, not generic data
• Embedded guardrails and authorizations
• Tailoring to the customer context
This is the value of software: saving time, reducing cost, improving customer satisfaction, and optimizing revenue.
Pick the use case… in creating and proving situationally specific context engineering, data is even more of a king than it was in the past.
Per-Seat Is History
Need I say more? Value per person is being re-evaluated everywhere in lieu of agentic systems. I will just say this: be careful here. Agentic is a new interaction layer. Building prompts or enabling voice-generated requests may feel like innovation, but this is just the new user interface if implemented without good context.
As Priya Vijayarajendran, CEO of ASAPP, recently noted in a CIO article, “Most enterprise value comes from systems that can execute reliably across complex environments, with governance, integrations, and auditability built into the platform.”
Telling an AI agent to do something is only as valuable as its ability to execute. So, I’d focus less on counting agents and more on:
• How strong your context engineering is
• How effective are the actions that your agents take
• How clearly those actions tie to measurable outcomes
Focus on context and the quality of the action your AI agents take. Think about empowerment instead of having an agent follow only what you can do yourself on a web site and using the years of context to enable more actions/resolutions. Agent-to-agent automation is exciting, but again, outcomes are what matter in today’s agentic world—not seats and general subscriptions.
Metrics are Outdated
Right now, we’re trying to measure AI with metrics that honestly feel a bit dated. You see stats like:
88% of executives plan to increase AI budgets, driven by agentic AI. My question is, why are we centering on agentic AI instead of context engineering and measurable value-based results?
71% of CIOs and CTOs say leadership has unrealistic expectations for an AI ROI. Why is this unreasonable? We should all have high expectations for results instead of being paid simply to experiment with AI. Tough comment, but it’s my opinion. Too many organizations are not aggressive enough with AI-based goals. Lead or lose, IMO.
64% of companies still measure AI ROI primarily through efficiency metrics. That’s a decent starting point. Let’s not totally discredit a savings-based approach. That said, we should also be looking at:
•Speed and time-to-impact
•Accuracy and risk reduction
•Innovation and competitive differentiation
•Business-related advantages as drivers of AI strategy
The bottom line here is that measurement models are still catching up to the technology. AI is here to stay. Expectations are only going to rise. I will say again, lead or lose.
So What?
The first era of SaaS sold access. Now we sell results and execution.
From seats to outcomes.
From features to workflows.
From adoption to performance and industry-specific results.
The shift toward context engineering isn’t theoretical. I’m already seeing it where we’re placing bets. Companies like Uniphore and PeopleReign are not winning because they have “better bots.” They’re winning because they embed deep enterprise context into workflows and drive measurable outcomes.
Similarly, platforms like Together AI are focused on enabling real-time data access and infrastructure – because without the right context layer, the AI simply doesn’t perform.
This is one of those exiting moments where companies will either leap forward or fall behind and wonder what happened.
My vote: let’s be in the group that leaps forward.
