Over the last two decades, I’ve spent a lot of time building in data and software. Looking back, one thing keeps nagging at me: we made huge progress, but a lot of it was incremental.
We shipped better analytics, better dashboards and better ways to activate data. It all helped, but most of it lived “on top” of the stack. The underlying machinery stayed largely the same.
AI is not just another feature layer, it is changing what the stack needs to look like in the first place. Right now, most enterprises are experimenting where it’s easiest to start: the application layer. Copilots, assistants, AI-powered workflows. The demos look great, but once you try to put these systems into real environments, the cracks show up fast.
Because underneath, many of the basics are still messy:
- Data is scattered across systems that don’t talk to each other.
- Context is missing or stale.
- Workflows are fragmented.
- And the infrastructure was never built for real-time, model-driven decision-making.
So AI ends up feeling impressive in a prototype, and not yet there in production. That gap is what I keep coming back to, the distance between what AI promises and what enterprises can reliably use.
Closing it won’t come from bolting on more apps. It needs a new foundation system that unifies data and context in real time, makes models dependable, and plugs decisions directly into the workflows where work actually happens. In other words: we need an operating system for enterprise AI.
Why did I choose Team8?
The reason I joined Team8 is straightforward. Team8 isn’t a traditional VC. It’s a company-building platform. It starts with real, concrete problems which are validated with operators who are living them day to day and builds companies around solving those problems properly.
That approach matters even more in AI, especially right now.
Because the biggest opportunities aren’t always where the hype is. They’re where adoption breaks: where constraints are real, the plumbing is missing, and teams are stuck patching things together just to get something working.
At Team8, I’ll be focused on AI and software infrastructure, working with founders and tech leaders to pinpoint where enterprise AI falls apart in practice, turn those gaps into strong theses, and build companies to solve them from day one.
I’m particularly excited about areas like multi agent systems, small (local) models, neoclouds, multi modal processing, evaluations, SDLC with AI, and post-attention model architectures.
It still feels early. The companies that define the next decade won’t only build better models, they’ll build the systems that make those models usable, reliable, and embedded in how enterprises actually operate.
That’s the layer I care about. And that’s what we’re building at Team8.
Operating Partner
Nitay Joffe is an Operating Partner at Team8, focused on building and investing in SW & AI Infrastructure.
