Redefining Kaeso

Kaeso was not defined wrong. Sometimes a product changes. Not because the idea was wrong, but because working in such problem-solving areas naturally leads to discovering more things you want to change. I personally decided that, since we haven't even started yet, it makes sense to turn the ship around as long as it is not too expensive to do so.

At the start, I didn’t really strive for anything specific. I mainly wanted a motivation, a binding reason to work on such a project and to enter this environment: working with AI.

I already knew what the market was seeking. Almost all current inventions regarding AI revolve around one major topic: AI agents. Whatever it is, a support agent in a telecommunication company, a coding agent, everybody is thinking about agents.

So I thought about centralizing this idea by creating a platform that could use all kinds of models created by large companies to execute workflows with various tools and integrations. I already knew that agents need data. A construction worker cannot do what you want without a plan given by you. A lawyer cannot represent you without knowing your case.

While working on this platform, I realized that it is extremely annoying to build a tool for the agent to use for every single small connection you want to create to any other data source. The agent only works with data, and there is a design principle that many developers, designers, and creators follow: "The User Is Stupid." An agent requires idiot proof design.

This leads to a simple question: how should the agent get data without the user manually giving it every time?

The solution is simple: the agent should be able to do it on its own. That is what users expect. This is also why people liked projects such as OpenClaw (ClawdBot).

You connect the agent to a connector and it retrieves whatever data it needs by itself.

The average human is lazy. Most people do not even want to set up a single gateway for an agent to access their data, not to mention several different ones. Every service has its own login, its own email, password, two factor authentication, and so on.

At the same time, people still expect privacy while often ignoring the complexity behind it.

This is where Kaeso solves the problem.

Kaeso creates a unified API gateway that agents connect to once. It provides a well documented and structured API where you can define permissions once and control which API keys are allowed to access which services.

It also allows multiple accounts per service, making it possible for an agent to access significantly more data than it would normally be able to on its own.