Harald Collet isn’t an engineer. He earned his tech management stripes the old-fashioned way, working his way up through roles at various large technology companies.
Yet in 2017, he and three entrepreneurial-minded colleagues co-founded Alkymi, a venture capital-backed tech startup that works to use artificial intelligence to improve enterprise business workflows across a number of industries, most recently landing on financial advisors as an appropriate market.
In April, the firm announced its Data Acceleration Platform would use OpenAI’s GPT-4, and Microsoft’s iterations of the technology, with the aim of helping advisors, and others in financial services, “summarize any document, any piece of content they receive in over 95 languages, and then integrate that into their workflows,” said Collet.
Previously they built their own machine learning tools based on open-source platforms. In 2020, they launched their first product, Data Inbox, and gathered clients in enterprise businesses in diverse verticals including trucking and the broader financial services industry.
They designed the tools to integrate with the most common enterprise software systems, from email and office productivity tools to cloud storage, content management systems and product management software.
Within months they added Patterns, another machine learning tool for building out business workflows, and later Patterns Studio, a no-code, automation toolkit.
“One of our top use cases right now is brokerage and custodial statements and their related use for new client onboarding,” he said. Some of the largest firms in the space to sign on for the service are Interactive Brokers and SimCorp—there are others that Collet could not name.
Late last year Alkymi raised $21 million in Series A funding led by Intel Capital, along with existing investors. Intel Capital’s Investment Director Dave Mueller joined Alkymi’s board of directors.
With the GPT update Collet said his team has created a bridge for those using the technology, but with business-specific protections and guardrails built in.
That includes meeting a client’s security and compliance policies, incorporating pre-defined prompts specific to investment advisory clients or other customized options.
Collet spent almost 10 years at big technology companies including Oracle and HP before joining Bloomberg where he spent six years as a general manager for the company’s professional-service governance and compliance tools. His team launched the Bloomberg Vault cloud service in 2011.
“We were really focused [at Bloomberg] on wealth, asset management, and private markets, and the people in the front office [at Bloomberg] were taking all these diverse documents and bits of data, whether pdfs, emails, you name it, and trying to turn it into actionable information,” he said.
Taking similar unstructured data and feeding it into their own AI—and in turn generating actionable information for their clients—sums up the mission of Alkymi.
“I’ve spent 20 years on this kind of unstructured data,” he said. Previously, that’s been “easy for humans to create but hard for computers to understand. And it very much influenced the product strategy we have put in place.”