US startup Intelus is hosting an open beta for companies seeking a no-code, no-data-science platform it says can be used to build custom artificial intelligence models quickly and at a fraction of the cost of incumbent practices.

Intelus was founded in 2020 by former executives from Microsoft and Salesforce. The company is privately held, and is headquartered in Seattle.

Among current industry challenges it aims to solve include:

- Small companies can’t compete – need large stores of data; major computing power

- Large companies can’t scale – machine learning & AI can’t solve problems for which there is no data

- Inaccurate data hurts business – poor modelling costs the average U.S. company up to $14.2 million annually (source: Gartner)

- Faulty models in research – “universal clinical failure” cited in select COVID-19 studies in which faulty models generated the same “methodological, dataset, and bias issues” (“State of AI 2021,” slide 110)

- Inefficient workflows – tech can be hard to use, difficult to understand, and costly to maintain; building models requires collaboration between software engineers, data scientists, and subject matter experts

“We are here to address all these issues, and to place the power of data back in the hands of business owners and domain experts,” said Patrice Simard, CEO & Co-founder, Intelus.

“For the past decade, organisations have been at the mercy of specialists. It’s a situation that prevents smaller companies from competing and hampers innovation at large firms, where incumbent technology is placed before informed business decisions.”

“The cost and complexity of machine learning has long been a barrier too steep for most enterprises,” said Gary Flake, former CTO Search & Data Science, Salesforce.

“Intelus makes a new paradigm possible with machine teaching, a framework that democratises machine learning by empowering anyone to teach, test, and deploy state-of-the-art models with no code, no hardware, low complexity, and at low cost."

Duet is Intelus’ software-as-a-service (SaaS) cloud platform, allowing users to leverage the “Machine Teaching,”  method to build, deploy, integrate, and maintain their artificial intelligence models.

At a click of a button, users can deploy their models on live traffic to classify and extract data from digital text documents and text streams, like email, customer support tickets, social media feeds, legal documents, and other text-based unlabelled and unstructured data.

Using a graphical user interface, the user can:

- Author: interactively create artificial intelligence models using a combination of schema editing, document labeling, feature creation and editing, and pre-defined features vetting from curated feature repositories. (Available now)

- Deploy: perform continuous integration and deployment functions of those models; check-in and store changes for the model and associated data, labels, and features; create versions of a model; test the versions; consume model versions deployed on REST endpoints. (Available now)

- Reuse and Grow: build a library of reusable features and models as building blocks for more elaborate and performant models. Unlike labels which are dataset-specific, curated features and models become a company’s reusable asset. (Full reuse availability in 1H2022)

https://duet.intelus.ai