There's a line from a recent panel on partner tech that I keep coming back to. Stephen Kellum from Structured said it plainly: no partner actually wants to go into anybody's portal. Not mine, not yours.

He wasn't being dismissive. He was being honest about something the industry has danced around for years. We've spent enormous energy building portals, loading them with content, deal registration flows, MDF tools, training modules. And partners still don't show up. The portals are fine. That's not the problem.

The problem is that portals ask partners to change their behavior. Log in here. Learn this system. Come to us. For a partner managing relationships with six or eight vendors, that's not a workflow. It's homework.

The Model Is Flipping

There's a macro current running underneath all of this. At Davos earlier this year, David Sacks, who oversees US AI policy, was asked how you know when you've won the AI race. His answer was simple: you win when everyone around you uses your products and your services. The companies that get huge are the ones that create ecosystems, get the most developers building on their APIs, have the most apps in their store. That framing is shaping how CEOs think about partnerships right now, whether they say it explicitly or not. The platform economy isn't an abstract concept anymore. It's a boardroom directive.

What's starting to change, and I think it's moving faster than most program leaders realize, is that the burden of engagement is shifting from the partner back to the vendor.

Instead of building a place where partners have to go, the platforms gaining traction are figuring out how to show up where partners already are. Email-based deal registration. AI assistants that work through a chat interface instead of a dashboard. Content that gets personalized and delivered rather than catalogued and hoped for.

Harbinder Kira from Mind Matrix described it as the Netflix effect. You used to drive to Blockbuster. Then content came to you, on demand, matched to what you actually wanted. Partner enablement is heading the same direction. The question isn't whether a partner will visit your portal. The question is whether your tools can reach them.

That's a real operational shift. It means the architecture of how you deliver information, content, and support has to be rethought from the partner's perspective outward, not from the vendor's system inward.

Easy Isn't the Finish Line

Making things easy matters, but it's becoming table stakes. The next layer is telling partners what to do, not just giving them the tools to do it.

The difference is real. An AI assistant that helps a partner find a campaign template is useful. An AI assistant that says "here's the campaign you should run, here's who you should send it to, here's what similar partners in your tier have done, and here's how much MDF you have left" is a different thing entirely. One answers questions. The other removes the need to ask them.

This is where the agentic conversation becomes relevant for partner programs, not as a buzzword but as a practical shift in what good support looks like. The partner comes in on a Tuesday morning and needs to do something. The system should already know who they are, what they've done, what they're likely to need. That's the bar being set right now.

The Data Problem Nobody Has Fully Solved

The honest caveat, and several people on that same panel said versions of this, is that none of it works without the underlying data being in order.

James from 360 Insights made the point clearly. Any one platform can build impressive AI, but if it's only operating on its own data, it's missing most of the picture. The co-sell motion he demoed required CRM data, account mapping, marketing execution, deal registration, and buying intelligence all talking to each other. That's not a single-vendor problem. That's an orchestration problem.

Pat McNalen, who spent nearly 30 years building partner programs at Intel, framed the same issue from a different angle. AI is only as good as the data you feed it. If you don't have one version of the truth, you're not optimizing a program. You're optimizing a slice of one.

That's worth sitting with if you're evaluating where to start. The clearest AI wins in partner programs right now are in contained, well-defined processes. Deal registration. MDF claims. Content localization. The data is relatively clean, the workflow is understood, and you can actually measure what changes. That's the foundation before you get into more complex orchestration.

There's a related point about sequencing that I think gets missed. A lot of programs buy tooling before the motion is working, then wonder why adoption is low. The better order is to get the flywheel spinning first, then add tooling to remove friction from something that's already moving. A PRM doesn't create partner momentum. It can support momentum that already exists. That distinction matters a lot when you're deciding where to invest time and budget.

What This Means for How You Build

If the direction is toward workflows instead of portals, some assumptions about how programs are built need revisiting.

Partner-facing tools should be evaluated not just on features but on how much behavior change they require from the partner. If adoption depends on partners learning a new system, the value has to be proportionally obvious.

Content strategy also shifts when delivery is personalized. Building a library and hoping partners find the right asset is a different problem than building modular content that a system can assemble and target on the fly.

And the PAM role changes. Less time managing portal access and answering basic questions, more time on the strategic work where a human actually adds something the system can't replicate.

Where the Trust Stack fits in

Which raises the question of what that strategic work actually is. Sudhir Kumar's Trust Stack framework is useful here, even though it was written for B2B marketing. He breaks trust into three layers: credibility (do you know your stuff), reliability (do you show up consistently), and empathy (do you understand their world). Applied to partner programs, the breakdown is pretty clean.

Layer 01

Credibility

The layer AI handles well. Surfacing the right content, data, and co-sell context at the right moment. A sharp, accurate answer in seconds builds confidence in the program.

Layer 02

Reliability

Where automation earns its keep. Consistent deal registration confirmation, predictable MDF processing, on-time QBR prep. Partners notice when these work, and definitely when they don't.

Layer 03

Empathy

Still requires a person. Reading the subtext in a QBR, knowing when to push and when to back off. The partners who matter most can tell the difference immediately.

The PAM's job in a workflow-first program is to operate at that empathy layer while AI handles the other two. That's not a diminished role. It's a more focused one.

The Harder Question

There's also a harder question sitting underneath all of this. A lot of ecosystems grew by adding partners, adding tools, adding program complexity, and then adding headcount to manage it. That math is getting harder to defend. Scaling expenses alongside performance isn't a growth strategy anymore. The programs that are going to hold up are the ones focused on yield: fewer partners doing more, tighter workflows with less overhead, AI that handles volume so people can focus on what actually requires judgment.

The same pressure is hitting services companies and SIs from a different angle. Their entire business model is built on billing manhours, renting out the time of skilled people. When a junior consultant with an AI assistant can complete a task in 20% of the time it used to take, the hourly model starts to crack. McKinsey reportedly generated 20 to 25% of their services revenue from outcome-based pricing last year. That's not a small number, and it signals where things are heading. The services companies that survive this are the ones that find their moat in deep industry knowledge, build reusable IP around it, and stop competing on headcount.

The Salesforce State of Sales 2026 report is worth noting here. They surveyed 4,000 sales professionals and found that 94% are now participating in partner selling, up from 86% the year before. Partnerships are growing even as the model shifts. That's an important distinction. The concern isn't whether partnerships matter. It's whether the infrastructure we've built to support them is keeping pace with how partners actually want to work.

Which brings it back to the role itself. Anthropic recently listed a global partner sales manager for SIs at $250K OTE, responsible for revenue ownership, program design, ops, and enablement. One person. The expectation, reading between the lines, is that this person runs their patch with agents handling the volume work across all those functions. That's not a future-state prediction. That job posting is live now. The partner professional who builds those agents and knows how to direct them is a different animal from the one who manages a portal and chases login rates.

Von Moray, CRO at Mind Matrix, put the five-year view plainly in a recent conversation: partners don't want to log into platforms. They want to sell. The platforms that last are the ones that figure out how to get out of the way and into the workflow.

That's the shift worth building toward now.

Want to talk channel strategy or partner operations? Connect on LinkedIn.