When I co-founded SkyKick in 2011, we set out to solve a clear problem: help MSPs migrate and protect their clients' Microsoft 365 environments. Over the next thirteen years, we built the platform, scaled to $44M in ARR, earned six US patents in cloud migration and data management, and in 2024, completed our merger with ConnectWise.
It was a meaningful chapter. But the whole time we were building backup and migration infrastructure, I kept seeing something else. Something bigger. Something that had nothing to do with backup.
The small businesses living inside these M365 tenants - 10, 20, 50-person companies - had access to one of the most powerful productivity suites ever built. And they were still drowning. Not because the tools were lacking. Because nobody was connecting the dots between them.
The richest dataset nobody was using
As CTO of SkyKick, I spent over a decade deep inside the Microsoft Graph API. I built systems around its data model, pushed the boundaries of its permissions architecture, and developed an intimate understanding of how signals flow between Exchange, Teams, SharePoint, and OneDrive. Many patents came out of that work - all focused on making cloud data more useful, more portable, more manageable.
And that depth is what made the problem so maddening. The data was right there.
The email thread that had gone cold for six days, with a client waiting on a response? Visible in the Graph. The recurring meeting where three of five attendees declined last week? Right there in the calendar API. The SharePoint document uploaded for review but never opened by the person who needed to approve it? Tracked in the audit logs.
Every signal that a smart operations person would catch, if they had the time and attention to watch for it, was already sitting inside the tenant. Microsoft 365 was generating an incredible amount of operational intelligence. Nobody was reading it.
I kept thinking: this is the most comprehensive record of how a small business actually operates, and it's just sitting there.
The gap that a decade of building couldn't close
Over the last two decades, we've built extraordinary tools for small businesses. Email, chat, file sharing, video conferencing, project management, the capabilities available to a 25-person company today would have been unimaginable for an enterprise 15 years ago.
But we solved the tooling problem without solving the operations problem.
Small businesses weren't failing because they lacked the right apps. They were failing at the seams between apps. The proposal that sat in someone's inbox for two weeks. The onboarding checklist that stalled because one person was out sick and nobody picked it up. The quarterly business review that got rescheduled three times and eventually just didn't happen.
These aren't software problems. You can't set a threshold alert for "this client relationship is quietly deteriorating." You can't write a script that detects "your team agreed to do something in a meeting and then nobody did it."
We had world-class tools for doing work. We had almost nothing for watching whether the work actually got done.
That gap has always required a person - a chief of staff, an operations lead, an executive assistant with great instincts. Someone who could hold the whole picture in their head and catch the things that were about to fall through. Small businesses couldn't afford that person. So the CEO did it themselves, badly, on top of everything else.
After years of building tools that made M365 data easier to move, protect, and manage, I was convinced the next frontier wasn't making the data more accessible. It was making the data intelligent.
Why Microsoft Graph is the key to something much bigger
Most people in the tech industry think of the Graph API as plumbing - a way to manage users, read emails, or sync calendars. Useful infrastructure, not particularly exciting.
I see it completely differently. After building many patents worth of systems on top of it, I've come to believe the Graph API is the closest thing we have to a real-time model of how a business actually runs. It captures who communicates with whom, what commitments get made, which documents are in play, where attention is flowing, and what's being ignored. All structured. All queryable. All sitting inside the customer's own environment.
For years, the limitation wasn't the data. It was the reasoning. You could write a rule that says "flag emails older than 5 days with no reply." You could not write a rule that says "this email chain is about a contract negotiation, the client's tone has shifted from warm to formal, and the last message contained an implicit deadline that nobody on your team has acknowledged."
That second one requires understanding context, intent, and urgency. It requires reading between the lines. Until recently, that kind of reasoning was exclusively human.
But M365 is only part of the picture
The deeper I looked at this problem, the more I realized something: Microsoft 365 might be the center of gravity for most small businesses, but it's not the whole story.
A 30-person company doesn't run exclusively on M365. Their sales pipeline lives in HubSpot. Their financials are in QuickBooks. Their product team tracks work in Jira or GitHub. Their knowledge base might be in Notion or Google Drive. Some of them have homegrown line-of-business applications with SQL databases that hold the real operational data - client records, project statuses, inventory, billing history.
The operational intelligence trapped inside M365 was just the beginning. There was equally valuable signal locked inside every other system these businesses depended on. And none of these systems talked to each other in any meaningful way. Not at the level of understanding what the data meant.
This realization expanded the vision entirely. Small businesses don't need another integration platform that moves data between silos. They need an intelligence layer that understands data across silos. An AI that can read a client email, recognize it's about a billing dispute, pull the relevant invoice from QuickBooks, check the project history in Jira, find the original contract in SharePoint, and draft a response that accounts for all of it. That's not an integration. That's operational reasoning across the full surface area of a business.
The convergence that changes everything
I've been skeptical of most "AI-powered" products in the M365 space. Having spent thirteen years building real infrastructure on the Graph API, I know how hard it is to make something genuinely useful versus something that just demos well. I've seen enough hype cycles to know the difference.
But something fundamental has changed. For the first time, AI models can read an email thread the way a chief of staff would - not pattern matching on keywords, but actually understanding that a deliverable is at risk, that a relationship needs attention, that a commitment was made in a meeting on Tuesday and by Friday nobody has acted on it.
This is the convergence I spent my career unknowingly building toward. The Graph API gives us the operational core with the communication, collaboration, and document layer where most business activity happens. Native integrations with systems like HubSpot, QuickBooks, Jira, GitHub, Notion, and even custom SQL databases extend that awareness into every corner of the business. And AI gives us the ability to reason across all of it the way a great operations person would.
Put them together, and you can build something that has never existed before: an autonomous AI workforce that lives inside a business, understands it holistically, and keeps it running.
That's Outermind.
Not another dashboard. Not another integration platform. A team of specialized AI agents that actually understands the operational state of your business and acts on it - drafting email responses, running down research across your documents and knowledge bases, updating your CRM, querying your financial data, and coordinating the work that nobody was hired to do.
The vision: every small business gets an AI workforce
Here's what I believe. Every company with 10 to 100 employees is running with an operational gap that costs them clients, revenue, and the mental health of their founders. That gap exists because the connective tissue between their tools - the part that turns information into action - has always required people they can't afford to hire.
AI changes that equation permanently. Not by replacing your team, but by giving small businesses the operational capacity that used to be reserved for companies with dedicated ops staff, executive assistants, and chiefs of staff.
Imagine a 25-person company where:
- No client email goes unanswered for more than 48 hours - because a Mailbox Assistant drafts replies before you even open your inbox
- A question about company policy gets an instant, sourced answer - because a Subject Matter Expert searches your actual SharePoint, Google Drive, and Notion documents, not the internet
- Your CEO starts Monday morning with a clear briefing - because an AI Chief of Staff has already reviewed the week's email, calendar, CRM activity, and project tracker updates, and distilled it down to the three things that actually need attention
That's not a feature set. That's a transformation in how small businesses operate. And it's possible right now because the data sources, the AI reasoning, and the deployment model have all matured at the same time.
Built where your data already lives
Over a decade at SkyKick taught me something non-negotiable: Outermind had to deploy inside the customer's own M365 environment.
This isn't a philosophical preference. It's an architectural decision rooted in over a decade of learning what actually works for small businesses. Data residency matters. Tenant isolation matters. Deploying via standard OAuth admin consent, without requiring shared credentials or custom infrastructure, matters.
We built Outermind natively on Microsoft technology. Not through a third-party connector. Not by syncing data to an external platform. The AI operates on your data, inside your tenant, under your existing security policies. And when Outermind reaches out to your other business systems like your CRM, your accounting software, your project tracker, it does so with the same respect for data boundaries and security that we learned building enterprise-grade cloud infrastructure at SkyKick.
I've seen too many products in this space take shortcuts on data isolation. We didn't. Because when you're building an AI workforce that reads a company's email, queries their financials, and searches their documents, trust isn't a feature. It's the foundation.
Why this time is different
I've been building in the M365 ecosystem since 2011. I've watched waves of "next big thing" come and go. I've built products that scaled to $44M in ARR and I've built prototypes that went nowhere. I hold many patents in this space. I know the difference between a demo and a deployment.
Outermind exists because three things are true at the same time, for the first time:
- The Graph API and native integrations with tools like HubSpot, QuickBooks, and Jira provide comprehensive operational data about how businesses actually run.
- AI models can now reason across that data in meaningful, contextual ways.
- The deployment model - inside the tenant, via admin consent, with full data isolation - is ready for production.
I didn't start Outermind because the market analysis looked good on a slide. I started it because I spent thirteen years staring at the richest operational dataset in small business technology and watching it go to waste. Every frustrated founder compiling Monday morning status updates by hand. Every client relationship that quietly deteriorated because nobody was watching. Every meeting that generated commitments that evaporated by Friday.
The data to prevent all of that was always there. Now the technology is too.
If you run a small team and you've ever wished someone was watching the whole picture so you didn't have to, that's exactly what we built with Outermind.