The Benchmark Was Wrong: What AI-Native Partnerships Actually Requires

By Rob Moyer, Founder, BlueThread

Sam Gong and I just published a report with WorkSpan. We called it "Running AI-Native Partnerships." We pulled in ten practitioners from across the cloud, ISV, and SI ecosystem to help us make it worth reading: Jay McBain from Canalys and Omdia, Jason Mann from Gong, Joe Estes from Boomi, Swati Moran from Docusign, Vince Menzione from Ultimate Partner, and Mayank Bawa, CEO of WorkSpan.

This post is the operator's summary. What we found, why it matters, and what I think partnership leaders should do with it.

The Headline Number Was the Wrong Benchmark

When you survey partnership teams about AI, you get a predictable answer: near-universal adoption, a few hours saved per week, and a long list of use cases that boil down to meeting prep, document drafting, and knowledge search.

That is real. Those hours are real. The practitioners doing it well, building reusable agents, running morning call-prep automations, eliminating research backlogs, deserve credit for moving fast.

But the benchmark measures whether individuals got faster at tasks they were already doing. It does not measure whether partnerships as a revenue function got transformed.

Faster call prep does not change how a co-sell motion runs. It does not change how sellers activate. It does not move a deal from account overlap to closed revenue. A few hours saved per week is a floor. Partnership operators deserve a higher bar than this.

Why the Underlying Market Makes This Urgent

The pressure to get this right is not about AI hype. It is about a structural shift in how enterprise software is bought.

Two-thirds of technology is now sold on subscription or consumption. That one number changes everything about what a partnership team actually does. When software was sold as a perpetual license, the transaction was the relationship. The partner's job was to move product. Once the check cleared, the work was done.

That model is gone. Today, customers do not commit upfront. They commit to the outcome, renewed every quarter. The transaction is the beginning of the work, not the end of it.

Jay McBain has been tracking this across the largest technology companies in the world. The average enterprise deal now involves 6.3 partners coordinating around a single customer. On large deals, that number exceeds ten. Co-sell is the convergence motion because co-sell is where the deal lives.

McKinsey's estimate for annual ecosystem revenue by 2030 is roughly $80 trillion. Partnerships are not a side function. They are how enterprise software gets sold.

The partnership teams running manual, relationship-mode operations are not just slow. They are structurally mismatched to the market they are operating in.

The Evolution of the Partner Manager

The market shift did not just change what gets sold. It changed the job.

The previous era's partner manager made internal sense: manage a portfolio, maintain the relationship, run the QBR, track certifications, issue MDF, approve deal registrations. That is a coherent job description for a world where the relationship was the product.

It does not scale in a co-sell world. When the average deal involves 6.3 partners and the co-sell motion lives in deals opening and closing every week, you cannot manage your way through it. The work is different: deploying the right partner capability against the right deal at the right moment.

The role is not dying, it is evolving. I have been calling this the rise of the Partnership Operator. The operator does not maintain relationships. The operator runs systems. Inputs, levers, and loops.

My benchmark for operational: under 60 seconds to find the right partner for a deal stuck at VP level in healthcare. If finding that partner requires sending a Slack message to your team, you have a manual system with a chatbot in front of it.

Sales has known this problem for over a decade. Sellers spend more than 70 percent of their time on non-selling activity. Partner managers now have the same problem. The new partner manager is measured on deal velocity, working pipeline, and co-sell conversion. The old one was measured on connections, certifications, and lunch-and-learns.

AI did not create this shift. But AI is what makes the operator model actually executable at scale.

Three Bars AI Has to Clear

In the report, Sam and I identified three specific requirements that have to be met before AI transforms partnerships as a revenue function. General-purpose tools do not meet them. This is not a criticism of the tools. It is a structural observation about what the co-sell motion actually requires.

Bar 1: Live Deal Context

Most AI-assisted partnership work looks like this: the partner manager opens a chat tab, pastes in account information, pastes in opportunity fields, pastes in tech stack details, sends the prompt, reads the response, and carries the output back to wherever it needs to go.

The human is the courier. That is not AI-native. That is AI-assisted with all the friction left in.

The difference in an AI-native motion is that the agent already has the context. It knows the account. It knows the stage. It knows the partner's coverage model and historical win patterns. The human reviews the output. The human does not feed the input.

When AI has to rely on pasted context from a human, the human is doing the work to make the agent effective. It is supposed to go the other way.

Bar 2: Execution, Not Generation

The most common AI use case in partnerships is drafting. Draft the co-sell brief. Draft the referral. Draft the partner outreach.

Drafting is output. The partner manager still has to carry that output somewhere, log into a portal, re-key 12 fields, handle validation errors, and close the loop. Every one of those steps is a point of failure. Sellers are busy. Partner managers are stretched. The referral sits in a tab and never gets submitted.

Boomi grew marketplace revenue more than 3,000 percent year over year. That did not happen because their team drafted better referrals. It happened because ClearScale automated 100 percent of their referral submissions across 350+ referrals, generating $5M in partner-originated pipeline. Automation, not faster preparation.

The standard I use: if a human is still carrying AI output across screens, the problem is not solved. AI that does not act is not a co-sell tool. It is a research assistant.

Bar 3: Cross-Company Operation

This is the bar the industry has not fully articulated yet, and it is the most important one.

A co-sell motion involves two companies. Two CRM systems. Two security postures. Two data governance policies. Two views of the same account. AI that operates inside Company A's Salesforce instance cannot see what Company B knows about that deal. It cannot submit a referral to Company B's portal. It cannot receive the acceptance and write it back to Company A's CRM.

An alliance leader at a major global systems integrator, a firm that has trained tens of thousands of professionals in Claude, described exactly this problem. When an AI agent tries to access partnership data, the firm's IT team blocks it. He put it plainly: the agent is an outside force acting on data they consider sacred. He asked the right question: how could we use any AI tool, even one we built ourselves?

The blocker is not vendor trust. It is the absence of a sanctioned shared environment where both parties have explicitly authorized access, where data governance from both companies is enforced at the field level, and where AI can act on partnership data because both parties signed off on the space.

This is architecturally different from anything general-purpose AI provides today. ChatGPT cannot hold simultaneous field-level access to two companies' CRM records. It cannot execute a referral workflow that requires authentication in two partner portals. It cannot generate a private offer that passes compliance review at both the ISV and the cloud marketplace at the same time.

Until partnerships has a purpose-built execution layer that operates between two companies, not inside one, the joint forecast stays a spreadsheet and the referral motion stays manual.

What It Looks Like When It Works

The report documented $513 billion in shared pipeline executed in a sanctioned cross-company environment. Not inside any one organization's AI stack. Between them. Of that, $196 billion is already closed.

Win rates move from 22% to 47% when cloud field sellers are activated inside the deals they are already working. That is not a marginal improvement. That is the operator's prize.

These are not predictions or projections. They are the early operating model, the one that defines partner-led efficiency for the AI era.

What Operators Should Do With This

I am not going to tell you to go buy a tool. This is not a product pitch.

What I will tell you is how to diagnose where your motion is stuck.

Run the 60-second test. Pick a live deal in your pipeline right now, something at VP level, something that has stalled. How long does it take to identify the right partner for that deal, confirm coverage, and initiate the co-sell motion? If the answer requires a Slack message, a spreadsheet lookup, or a meeting to figure out, you know which bar you have not cleared.

Then ask the harder question. How much of your co-sell output, briefs, referrals, joint pipeline updates, is your team actually carrying from a chat tab to a CRM or portal manually? That is your true cost of Bar 2. Not the hours spent prompting. The hours spent moving the output somewhere that matters.

The operators who are going to win the next three years are not the ones who adopted AI the fastest in 2023 and 2024. They are the ones who built the infrastructure underneath the prompt. Context that persists. Execution that closes the loop. A shared layer that both companies trust.

That is the shift from system of record to system of action. It is achievable. Some companies are already there. The question is whether you are building toward it with intention or still measuring success in hours saved per week.

---

The full report is live at workspan.com/ai-native-partnerships.

It includes primary research from nine companies across enterprise software, cloud platforms, professional services, and industrial OEMs, plus practitioner video interviews from Jay McBain, Jason Mann, Vince Menzione, Joe Estes, and Swati Moran.

Worth reading. I would not have put my name on it otherwise.

Questions or pushback, find me on LinkedIn or drop a note at bluethread.io.