Stack AI co-founders Antoni Rosinol and Bernardo Aceituno are MIT PhD college students who accomplished their levels in 2022, simply as massive language fashions have been changing into extra mainstream. ChatGPT will likely be launched to the world on the finish of this yr, however even earlier than then they acknowledged there was a difficulty throughout the firm of mixing information with fashions with out a number of experience and data – and so they wished to alter that established order.
After commencement, they moved to San Francisco and joined Y Combinator’s Winter 23 workforce, the place they launched Stack and refined their concept. In the present day, the corporate has constructed a low-code workflow automation instrument designed to assist corporations construct AI-driven workflows, similar to chatbots and AI assistants. The corporate has raised $3 million thus far.
“Our platform permits individuals to construct workflows that require connecting completely different instruments to work collectively. We specialise in connecting sources and LLMs as a result of doing so permits you to construct highly effective workflow automations. We additionally provide many different instruments and options to automate advanced enterprise processes. Their product has solely been accessible for six months, however it’s reported that greater than 200 clients are already utilizing it.
Primarily, this entails dragging elements onto the workflow canvas. This typically contains information sources similar to Google Drive and LLMs in addition to different workflow components similar to set off components or motion components for creating workflows, permitting clients to construct generative AI applications with out intensive coding. The coding itself shouldn’t be AI-driven, however the duties throughout the workflow typically are and will require some guide coding to make the workflow run easily.
A few of their earliest clients are from the healthcare trade, and Aceituno admits they should be cautious with purposes involving docs and sufferers, particularly when inside information sources aren’t at all times dependable or could comprise conflicting or outdated data.
In these circumstances, he stated, it is essential to depend on human consultants, docs, to evaluate the standard of the solutions. As one other stage of safety, they embrace supply citations in every reply so healthcare professionals can test the supply earlier than accepting a solution.
“That being stated, it is true that you may put in rubbish after which the citations will likely be rubbish, which is why these assistants are requested to not fully take over the method,” he stated.
Rosinol, who just lately graduated from MIT and based a startup, stated going to YC actually helped them perceive the enterprise aspect of issues and how one can refine their startup concepts by working with clients.
“We began with an preliminary model of this API that was extra developer-focused. We began with some clients and their concept was that we wished to make use of AI to automate RFP responses or automate gross sales. By working with clients, we found the true challenges It is not about coaching fashions, it is about effectively querying the sources and connecting them to those language fashions.
The corporate at present has six staff however is recruiting engineers in addition to gross sales and advertising and marketing professionals.
The $3 million funding was accomplished a couple of yr in the past. Buyers embrace Gradient Ventures, Beat Ventures and True Capital, in addition to LambdaLabs, Y Combinator, Soma Capital and Epakon Capital.
