Tink Founders Debut Agentic Compliance Platform Freda

The founders of open banking platform Tink have launched a new compliance platform.

Freda, which came out of stealth Wednesday (Sept. 23), has its roots in what its founders saw at their former company, CEO Daniel Kjellén said in a LinkedIn post announcing the launch.

“At Tink, we’d watched compliance become the speed-limiter on growth, innovation, and ambition,” Kjellén wrote.

“Every corner of our business had winning software, except compliance – the rules to run a company were simply beyond any technology. With Freda we are about to change that.”

According to Kjellén, Freda’s agentic compliance platform lets users “discover every framework and obligation” that applies to their business and create tailored programs to meet them.

From there, the agents “run the work of getting and staying compliant – end-to-end workflows in your connected systems – right where your work actually happens,” he added.

In a manifesto on the company website, Freda describes the limitations it says plagued pre-artificial intelligence (AI) compliance.

“A tool could hand a company a checklist built for businesses like it – never one built for that business specifically – and even then, only say what needed doing, never go do it, across every framework, every contract, every vendor, every system, every person,” the company said. “Compliance required action at a scale that has never been possible to systemize.”

But with AI, systems can for the first time read a rule, glean what it requires and act on it “at scale,” meaning that a compliance system “is no longer just a place to record the work that people did, but a system that actually does the work,” Freda said.

Sweden-based Tink was founded in 2012 by Kjellén and Fredrik Hedberg, who is now co-founder and chief technology of Freda, also based in Sweden. Tink was acquired by Visa in 2022 in a deal worth $2 billion.

PYMNTS looked at the role of agentic AI in banking compliance in a recent conversation with Christopher Phillips, director of financial crime industry engagement at Flagright.

“The AI doesn’t get bored. It doesn’t get fatigued. It doesn’t forget to include that link for the one search it did,” Phillips told PYMNTS. The result, he said, can be “a robust output that is easy to audit.”

He added that humans still need to be on hand to provide “effective challenge,” or understanding why an AI agent came to the conclusion it did and documenting why they agreed or disagreed.