Day 3 showed partners how to build, bill and govern AI coworkers across their client base.

Our take: the agents are well covered, and the people working alongside them deserve the same attention.

Every AI coworker an MSP deploys will sit next to someone who can still be phished, fooled by a deepfake or tempted to paste client data into an unapproved tool. That's no reason to slow down. It's the part of managed intelligence MSPs are best placed to own.

Managed intelligence is ready to build

Pax8 Chief Product Officer Libby McIlhany and Chief Technology Officer Avery Moon showed what running a managed intelligence practice actually looks like. 8os and 8bot are designed to give partners a repeatable, billable way to deliver AI to every client, with agents governed across the whole customer base from one place. 8os also learns each client's business over time.

McIlhany's message to partners, as she told CRN, was short and clear:

"Stop talking, let's get building."

Attendees received early access, with general availability expected at the start of 2027.

That governance layer covers the agents. The people prompting them, approving their actions and deciding what data to feed them need guidance too. That's where your security practice earns its place: helping clients agree who is responsible, where the boundaries sit and when someone needs to step in.

Attackers start with trusted identities

CrowdStrike's Titanium Sponsor keynote showed how attackers use trusted tools and identities to move from endpoints into cloud and SaaS environments. Its answer was connected visibility and fast response across AI tools, identities and cloud.

We'd take that a step further. Behind every identity is a person, and their behaviour, credential exposure and access all shape how much risk that identity carries.

Think back to the two employees from our Day 2 breakout. Both clicked the same phishing link, but only one had privileged access and credentials that had already been exposed. Same result, very different priority.

That's where Human Risk Intelligence (HRI) comes in. It builds on awareness training by connecting behaviour, credential exposure and access, so you can see who is most at risk and what to fix first. uHealth, the newest part of the usecure platform, brings those signals together in one view for every client.

From there, turn insight into a recommendation: explain what's exposed, agree the first fix and show the client how you'll track progress.

Customers want guidance, not more alerts

In Owning the Last Mile: Where Insight Becomes Income, Pax8 Co-Founder and Chief Strategy Officer Ryan Walsh pointed to a big gap in the market. For the first time, SMBs are outspending enterprises on IT, while MSP readiness to guide them on governed AI is falling sharply. Walsh sees that gap as a major opportunity, and so do we.

The What Would You Do? workshop showed one way to start: turning security discussions into business conversations. Groups worked through real-world scenarios, challenged assumptions and uncovered risks together.

It's a useful approach for client reviews too. Bring your evidence, then ask how each risk affects the client's business. Their answer might change which action matters most.

Pax8 summed up the day neatly. Customers don't want more alerts. They want trusted guidance.

For the wider event highlights, read Pax8's official Day 3 recap.

Three questions to open your next QBR

Combine human risk data with a wider conversation about how your clients use AI. Try these questions in your next quarterly business review (QBR):

• Which employees have access to sensitive systems, and are any of their credentials exposed?
• Which AI tools are people using, and what information are they sharing with them?
• What risks have we reduced since the last review, and how can we show that progress?

Each question connects technical risk to a decision your client can act on. Agree who owns the next step, then come back to it at the following review. That gives both sides a clear way to see whether the work is making a difference.

Trust in people

Peter Schmeichel's closing fireside chat, The Championship Mindset, gave us our favourite summary of the event: be clear on the job, trust your people and own your mistakes.

Over three days, Pax8 made a strong case for trust in technology. For us, that trust also depends on people: how they use AI, the access they hold and the decisions they make. Governed AI needs governed users.

A huge thank you to Pax8, to every partner who stopped by our booth during the event, and to everyone who made our EMEA Collaboration Award possible. See you in Madrid, 3 to 5 October 2027.

Want to give your clients the guidance they're asking for? Request a free MSP NFR or book a tailored demo.