By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
At a glance
WHAT IT’S REALLY ABOUT
Founders Reveal How to Embed AI, Differentiate, and Protect Data
- Cresta's Zayd Enam argues for a 'creative' approach to AI that reimagines work, such as turning high-attrition contact center jobs into relationship-building roles and mining customer conversations for business insight, rather than just automating existing processes.
- Hex's Barry McCardel and a Sourcegraph speaker both stress that the context and prompts fed into a model determine output quality, and that the hard work of AI products happens before the API call.
- On differentiation, the guests say model access is a commodity and UIs get copied, so advantages come from proprietary or personalized data, polished user experiences, and workflows that deliver business outcomes.
- Privacy, security, and cost shape architecture choices, with Sourcegraph offering self-hosting and a pluggable, bring-your-own-model design, while Hex avoids training models on customer code.
- The episode closes with Cresta's vision of a 2028 contact center where machines handle data entry and routine requests so humans can focus on strategic relationship building.
IDEAS WORTH REMEMBERING
5 ideasCreative AI beats lazy automation
Zayd Enam contrasts automating an existing process end to end with using AI as a new building block to reimagine the business, including mining customer conversations for product, pricing, and market insight.
Context is king
The Sourcegraph speaker says the context given to a model dictates output quality, and that most of Cody's quality issues come from failing to fetch the right context into the context window.
More context is not always better
Barry McCardel says Hex was tempted to stuff in as much context as possible but found that too much can confuse a model, so it iterated to find the right level.
The real work happens before the API call
McCardel notes anyone can get an OpenAI API key, but building context, tuning prompts, chaining models, and giving users feedback is the hard part of embedding AI.
Models are commodities; moats come from elsewhere
McCardel says sending a request to a model is already a commodity and UI elements get copied, so he points to data-driven personalization, strong user experience, and being first to adapt to changed assumptions.
WORDS WORTH SAVING
5 quotesthe lazy approach is to basically go say, "Hey, this is an existing process. Let me just go automate it end to end." But then the creative way to look at it is that, hey, no, my job or my role as a business is to deliver the best possible customer experience or the best possible product experience.
— Zayd Enam
internally, we've started saying, uh, this phrase, essentially like context is king.
— Unattributed
I think the models themselves and being able to send a request to one of them is, is already a commodity.
— Barry McCardel
Like we will generate the code, we will show it to you, but we wanna keep the human in the loop.
— Barry McCardel
It is not true that language models make search engines unnecessary. If anything, they make the search engines more valuable because now all that data that you can search becomes like 10X more powerful
— Unattributed
High quality AI-generated summary created from speaker-labeled transcript.
