The Founder to Fortune Podcast Building AI Products for Enterprise Users
With the release of ChatGPT, AI has caught everyone’s attention and we are seeing it in more and more consumer and enterprise products. As leaders in the field of AI who have been working with AI before it became a rage, we have some of the best minds to share advice to founders who are building products using AI.
Our guests today include:
* Saurabh the VP of engineering at Uniphore, which is a leading conversational AI company. Saurabh shares his experience building emotion AI, which is an industry-leading multi-modal AI model that helps their end customers literally “read the room”.
* Tamar, the VP of engineering at Box, a well-known document storage company. Tamar has deployed LLMs in interesting use cases that unlock productivity benefits as well as drive business processes at a scale that was not possible before–at least not at scale.
* Mandar, the head of Machine Learning at Doordash in the Ads platform team. They are famously customer-obsessed in his team and are using LLMs to deliver value to their end users in innovative ways.
Key Points From This Episode:
* At Uniphore, Saurabh talks about how it is hard enough to use computer vision models to accurately identify emotions real-time but they actually combine facial expressions along with other aspects such as tone and run them through multiple specialized models before “stitching” them together to help users “read” the sentiment in the room.
* At Doordash, Mandar’s team has automated at least a big part of the cumbersome data labeling process by using LLMs. Another very exciting area they are exploring is whether they can generate more data to feed their recommendation model using LLMs. Both these areas have proven to be quite promising for them. Beyond that, Mandar touches upon some foundational work he and his team are doing at Doordash including shoring up basic processes around model serving, GPU access, cost management among others.
* Tamar at Box underscores the importance of being realistic with AI within the enterprise context by making sure they meet the requirements of businesses for security, access control, compliance, and auditability. They started by publishing an “AI policies for Box” that they adhere to for everything that they build. Not only that, they shared it with the community and customers alike. Their commitment to giving customers control and assuring privacy is core to their strategy. Finally, Tamar brings up the importance of building sound observability into these systems so that engineering teams can effectively manage costs.
Links Mentioned in Today’s Episode:
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.foundertofortune.org