After a few months in stealth mode (we’re not completely ready yet), we’re pleased to launch Angus.ai to help developers make their industrial or consumer objects smarter, based on robust AI perception.

What are we trying to solve?

For R&D folks, embedding algorithms that locate sounds, recognize people, or detect gestures remains highly challenging:

– searching the right algorithm within the AI jungle that matches a given feature is a huge work. It must handle the dirty data of dynamic environments, far different from those of clean labs.

– once it is identified, tuning and integrating the algo is another complexity. It’s rarely an easy copy-paste from other cases.

– this algo will sometimes be a commercial product. Contracting will require additional weeks and effort, spent with papers and lawyers.

– finally, the sequence will be repeated as much as the features’ count: voice biometry, face recognition, speech recognition, etc.

Bottomline: all this requires strong expertise from scarce resources, and takes way too long – average 24 months for a complete AI product development.

Frequently, developers and product managers end-up reducing the scope of product features in order to launch within 6-12 months and meet business cycle times.

Possible deceptive experience for designers and end-users.

At angus.ai we look at reducing these complexity, bureaucracy and time-to-market, all at the same time. How?

We’re building a catalog of 40 AI perception algorithms that make sense, tuned for real life use, packed and orchestrated as cloud APIs, licensed under a single contract, to help with rapid prototyping + development before launch, and product scaling after launch.

Where do these 40 APIs come from?

3 sources:
– commercial products we’ve been testing from all around the world,
– opensource software or academic projects, which we clean and robustify to meet industrial standards,
– our own IP, when we’re not happy with what we’ve found out.

We constantly review the latest improvements, and welcome anyone willing to submit his own code for evaluation and potential addition to Angus.ai catalog.

How do we know what it takes to build smart objects?

We’ve been through these pains for the last decade, packing state-of-the-art AI algorithms into leading edge IoT products: humanoid robots, teleconference robots and self driving vehicles, both embedded and cloud.

“Angus.ai” : how come this company name?

The answer is somewhere between this and that

Our values : brave and practical!

Comments are closed.