Why we built Kydosa

Before Kydosa had a name, we were building and running AI agent fleets on challenging business problems. Some days it felt like superpowers, but other days it fell short. We were boundary-sensing on what the models were capable of, and the boundary moved every week or even every day. That process, plus the harness that we put around AI, taught us how to take advantage of the constantly-advancing capabilities and at the same time, create dependable results. The payoff is huge; traditional approaches are being up-ended.

I've seen a moment like this once before. I spent three decades at Accenture, where I built a start-up practice for Salesforce into a multi-billion dollar business. I remember what the early days of cloud felt like - the skepticism, the hype, the leaders who moved early and the ones who never grasped the underlying change. The pattern is repeating, except the cycle that took a decade then is taking months now. That pattern recognition is why Kydosa exists. It's also why we built it AI-native from day one, rather than retrofitting an old model.

James Mattison saw the same shift. I've worked with James for years, and I've never met another profile like his. He is a builder, a technical architect who has run giant infrastructure for global clients, and a client account lead who shaped some of the most innovative outcome-based programs our industry has produced - deals where the firm's fees depended on the client's results. For many that was just a sales-pitch; James made it happen. Alongside all of it, he founded a school in West Africa that has educated more than ten thousand students over thirty years. I continue to be in awe of what James builds.

Both of us are hands-on, not because we're nostalgic for it, but because at this boundary, being close to the work is the only way to know what's actually possible. We're working with our first clients now and being deliberately selective, targeting the use-cases where AI moves a real business outcome. Our commercial model matches: we're paid when our clients' outcomes are achieved. Case examples will be posted here as they mature; we'd rather show you than tell you. We're scaling the team the same way: carefully, with people who want to build and who share our passion for business outcomes.

Contrary to those who see AI diminishing human innovation, it's the people-side of this that we’re most ambitious about. We're setting up an in-person hub in New York where young talent learns with AI embedded from day one, alongside experienced leaders and forward-deployed engineers working inside client teams. Just as important: we help people in client organizations unlock their own potential. We harness new research into the human skills it takes to work well with AI.

We named the company after kudos - recognition earned through excellence. We share everything we learn with our clients, and we celebrate what they do with it.

Saideep Raj and James Mattison