Last week I saw a stunning AI tool demo. With incredible ease, the expert tied together user stories from Linear, user research, and the entire codebase into a context rich enough to generate prototypes in a snap.
I smiled to myself, remembering that just a couple of hours earlier I had failed to connect Claude to PowerPoint – because the connector simply couldn’t figure out how to talk to the German localization.
AI tools can speed you up significantly. They can enable you to build things you were never able to build before. But the real power lies in context – in workflows that draw on what you know about your users, your business model, your data, your problem framing, and your live product. Workflows where you don’t have to explain yourself over and over, and where you don’t have to copy & paste everything from Tool A to Tool B to a markdown file and back.
Yet this seamless connection is exactly where larger organizations hit a wall. Standing in the way is a series of deep organizational moats:
- Fragmented data & tool ownership, paired with varying willingness or technical means to share data and access.
- A jungle of licensing & configuration constraints, from rigid enterprise-wide licenses to isolated teams with niche tools.
- Legacy tools and custom builds without MCP servers, APIs, or any other way to hook AI into them.
- Procurement processes that can’t keep up with the pace of AI development, limiting access to AI models, tokens or new tools.
To be fair: some of these moats exist for good reason. Data privacy, security, and governance aren’t just bureaucracy – they protect the business and its customers. But many other hurdles that did little harm in the past are becoming a liability.
This is the work in front of us. Not picking the next tool, but answering the harder questions: How does context actually flow across our processes, roles, and toolchains today – and where does it break? Which of our silos protect something real, and which are just relics of decisions made long ago? Are our procurement paths fast enough for the pace we say we want? And who owns this question in the first place – IT, the business, or does it fall between the cracks?
AI is holding up a mirror to our organizations. If we don’t bridge these internal moats, even the most powerful models won’t move the needle.