Ken Van Haren and Chris Stanley, who had both been data scientists for Google and Square, respectively, felt aggravated about how much time they were wasting on setting up back-end infrastructure instead of doing actual work. After asking around, they discovered that this was a frequent issue. To estimate how widespread the challenge was, they conducted a survey and discovered that data scientists spent more than half their time cleaning and arranging data and the majority of the remainder collecting datasets.
Van Haren and Stanley created Patterns as an efficient way to do AI model engineering. With the support of Y Combinator and the angel investor Lenny Rachitsky, this platform raised $2.5 million during its pre-seed round.
“Patterns is a great resource for executives to get ready for the introduction of AI, to get in front of the changes it will cause to their company and to begin integrating core AI abilities into their item and processes,” said Van Haren in an e-mail exchange with TechCrunch. “We assist companies to manage AI’s astonishingly fast rate of progress, which entails adapting to new models and paradigms quickly.”
The Patterns platform enables individuals to construct combinations, automation, and flows incorporating synthetic intelligence from a selection of component pieces. Van Haren states it essentially binds workflow logic and structure in a software layer.
At first, customers can link an app to Patterns using a ready-made collection of connectors. Next, they create a particular application using code in the Patterns internet-based tool. Lastly, they can ship the end result and, if desired, observe its performance with the analysis and debugging tools Patterns has available.
Van Haren demonstrated what can be done with Patterns by creating a question-answering bot based on a CrunchBase database of investors, companies, and fundraising rounds. Additionally, he fine-tuned OpenAI’s GPT-3 language model with a dataset of over 6.5 million Hacker News comments, in order to bring together the collective knowledge of the HN community in one AI.
Van Haren noted that individuals are enthusiastic about AI and wanting to do more than just simply experiment with it. He explained that Patterns provides an effective and speedy way for people to build and implement AI into realistic tasks.
Patterns is a MLOps platform, which means it can be used as a tool for creating, testing, and sending machine learning models into production. MLOps is a rapidly growing business, with a lot of businesses in competition for capital and market share.
It has been predicted that the MLOps sector could be worth $4 billion by the year 2025.
Examples of other AI development platforms in the same space are Galileo, Qwak (which joins together machine learning engineering and data management), Diveplane (which is tailored towards businesses), Tecton, Arize, Iterative, Comet and Weights & Biases.
Despite the other competitors in the market, Van Haren reports that Patterns has seen an easy increase in users and their user base has grown to around 1,500 up until now. He declined to give the percentage of paying customers, but did confirm that Patterns anticipates completing a deal with a government agency sometime this year.
Patterns’ immediate objective is to expand its workforce, currently featuring four full-time staff members that include Van Haren and Stanley.