Insights
The technology changes quickly. The operating questions are more durable: what should be automated, what has to be governed, where human judgment belongs, what Microsoft capabilities already exist, and how an organization measures whether AI is creating real value.
The three pieces most boards ask us about first.
A Copilot rollout fails or succeeds in the seven decisions made before the license switch flips. A board-level checklist: data exposure, sensitivity labels, conditional access, responsible-AI policy, training, output review, and the kill switch.
Most boards we talk to assume shadow AI is a future risk. It isn’t. It’s a present condition. Three failure modes, the seven-day diagnostic question, and what the governed response actually looks like.
In a default Microsoft 365 tenant, Copilot Studio authoring access belongs to every user in Microsoft Entra. Nobody granted it. Nine agent governance controls across the Microsoft 365 admin center and the Power Platform admin center, the SharePoint carve-out that catches everyone, and the order to set them in.
The model behind all of it: What is a Managed Intelligence Provider? and how it differs from a traditional MSP.
Who can build what, what the board signs, and why pilots stall after launch.
The agent works. Ninety days later, four people use it and two are on the project team. Six documented sources of resistance and how each lands in a forty-person nonprofit, how to pick a first scenario worth finishing, and the three-rung ladder that separates usage from impact.
Microsoft just announced Scout, an always-on Autopilot with its own Entra identity that lives across Teams, Outlook, OneDrive, and SharePoint. It is experimental, Frontier-gated, and explicitly enterprise-pitched. The mission-driven sector is not in the press release. The use cases are everywhere.
Most foundation boards approved an AI policy from a template they never modified. The problem is that template policies do not survive contact with deployment. Here is a charter built from real deployments: yours to adopt verbatim, modify, or use as a starting point.
What the tools cost to run, how they compare, and what the tenant has to carry.
Microsoft 365 Copilot Cowork is generally available, and it brings usage-based billing to Copilot through Copilot Credits. The feature is impressive. The billing model is the bigger story: the shift from paying per seat to paying per seat plus usage, and what that means when the budget is tight and the board asks hard questions.
In a cloud-first world, your Microsoft 365 tenant holds identity, devices, email, and data, and it earns the same operating discipline once reserved for the server room. Microsoft ships new security capabilities constantly, but they are not switched on automatically. Why a documented baseline is the unit that matters now.
Dell just posted a record quarter selling the servers that run AI. As workloads move from pilot to production, that compute lands on your budget as LLM token spend. Why production AI economics will rhyme with the cloud era, and how to chart the right balance.
The marketing for both tools makes the choice sound obvious in opposite directions. The honest answer is more nuanced. Three factors that decide it, the cost reality, and where each genuinely shines.
Funding and automation for critical-access and rural hospitals.
Boston Children’s reported 50+ AI automations, tens of thousands of staff hours saved, and roughly $7M in redeployed labor. The same operational automations run on Microsoft 365, Copilot, the Power Platform, and Azure OpenAI, the stack a rural hospital is likely already licensed for, governed before it’s turned on.
The One Big Beautiful Bill Act put the largest one-time investment in rural health in a generation on the table. Cybersecurity, IT modernization, and AI are named, allowable uses. What the program funds, how to frame the ask, and how to position your hospital before the window closes.
Why the managed-services model is changing, and how we work inside it.
Pax8’s new report argues the divide that decides the next decade is between organizations that use AI and those that operationalize it. What its findings (the three-hour dividend, the governance emergency, the internal-first flywheel, and the $1.3 trillion shift from MSP to MIP) mean for nonprofits, foundations, and rural hospitals.
During the main keynote, Microsoft’s Chief Partner Officer, Nicole Dezen, named Centered Networks on stage as a leading Frontier partner, next to Infinity Group, NexusTek, and Sourcepass. What being a Frontier Firm actually means, and why we made ourselves customer zero before bringing AI to a single client.
I’m building AutoGrant as a company separate from Centered Networks, even though I founded both. Why the split made sense, what AutoGrant’s grant agents actually do, and why the next wave of useful AI is teams of specialized agents built for a single domain.
In the past three weeks, the two largest AI labs in the world have launched standalone enterprise services firms. The pattern matters. So does the gap they’re leaving open for mission-driven organizations.
Two weeks. Six phases from intake through readout. What we do, what you do, and what you walk away with on day 14.
Occasional perspectives from Centered Networks on the Managed Intelligence Provider model, governed AI on Microsoft, and what changes for nonprofits, foundations, and rural hospitals. No marketing fluff. Unsubscribe any time.
We’ll send the next perspective when it’s ready. No filler in between.