When I started my podcast back in 2019, my early guests were based in London so interviews were recorded face-to-face, in a small studio in Soho. As the show grew, my guests started coming from all over the world, so it was less likely I’d meet them in person. Not only do I enjoy meeting and recording in person, it’s even better when you get to meet your guests again after a recording. This week I had the pleasure of once again meeting Umesh Sachdev CEO & Co-founder of Uniphore.
Uniphore is a true AI company, not a recent launch on the back of ChatGPT. They started 16 years ago working on voice recognition using AI. Since then they’ve become truly multi-modal, working with images, video and text.
One question I’m asked all the time as a futurist is – what’s the next frontier for AI? My answer is that one of the most challenging, and exciting spaces for AI will be the enterprise.
Umesh and I had a great chat about the readiness of companies he’s speaking with and the things he’s seeing such as :
– companies struggling with data quality and format, leading to confusion and delays in decision-making.
– the lack of good quality (anonymised) data, leading to the need to synthesise data to be AI-ready – something Uniphore’s platform can do
– we also looked at the new topic of instruction tuning – training models to recognise and mitigate bias in data, creating a knowledge lake that can be used across the company.
I predict that over the course of this year, companies will prioritise budgets for “AI projects” only to realise their data is not yet AI-ready.
Data quality and access is something I’ve been shouting about from stage for years now – and perhaps finally the necessity for high-quality AI-ready data will bring companies up to where they should have been 5 years ago.
The slide below is one I show in my talks – and is often photographed. How many of these questions can you answer?
You can listen to our podcast episode below
Listen or subscribe via your favourite app
