Many AI tools today can perform individual tasks with high accuracy, often surpassing 98%. However, in complex multi-step workflows common in financial services, legal, and insurance sectors, small errors accumulate, causing overall accuracy to drop to 30–40% by the final step. This “compound error” issue prevents companies from adopting AI at scale due to unreliable results.
Curvestone AI, a London-based startup, addresses this critical problem by providing a workflow automation platform that maintains consistent accuracy throughout entire document-heavy and compliance workflows.
The company just secured $4 million in seed funding led by MTech Capital, with participation from Boost Capital Partners, D2 Fund, and Portfolio Ventures, to accelerate product development and expand go-to-market efforts, growing its library of proven workflows.
Before raising external funding, Curvestone AI reached profitability and grew revenue 7x in 12 months. The platform now processes billions of tokens quarterly across legal and financial services workflows.
Curvestone was founded by brothers Dawid and Sebastian Kotur, who previously led AI automation programs at PwC, Metro Bank, and GKN, and were involved in Facebook Developer Circles London. Recognising the potential of generative AI in professional services, they decided to pivot from consulting to building a solution that fundamentally improves AI reliability in regulated industries.
Read full article: https://techfundingnews.com/curvestone-ai-raises-4m-seed-to-fix-ai-compound-errors/
