We serve media clients, museums and publishing houses, telecom operators, customer service centres, manufacturing enterprises, trade associations. They don't lack the will to change. What they lack is a path they can afford and actually use.
In 2003 our founding team began working in software development. In 2013 they turned one conviction into a company: that people who don't write code should still be able to build the systems they need. For the decade that followed we built low-code, then no-code, making that act simpler each time.
In 2023 we judged that the road had reached its next junction — the point was no longer to make it easier for people to write code, but to let AI write it. So we turned the company's centre of gravity toward rebuilding systems with AI.
It was not a comfortable decision. It meant setting down product lines that were already stable and rebuilding from the foundation. But direction matters more than pace.
Most organizations don't lack the desire to use AI. They have systems everywhere, data scattered across them, and rules that live only in a few people's experience. So our first move is always to find out what's actually there.
A general-purpose model has no idea which conditions add up to "approved" in your organization. The first layer we build is the one that tells it.
We rebuilt our own delivery around AI before offering it to anyone else. We can commit to a fixed price and schedule because we proved it on ourselves first.
Agents in an organization, like the people in it, need defined headcount, permissions, collaboration and accountability.
The system keeps getting better after launch: agents in tsanClaw learn from real business feedback, so a year on clients are renewing and adding new use cases rather than shelving it after sign-off.
The ontology and business rules the first department builds in tsanCenter are reused by the second, and tsanCode turns every delivery into an asset for the next — so each new system costs less and ships faster than the last.
Data, ontology and agents stay with the client: business rules are changed without code in tsanCenter, agent roles and permissions are set in tsanClaw, with full on-premises deployment and China-developed models by default.