#76: AI vs. Security: The Unfiltered Reality with Saahil Sheenoy

·1 hr 4 min·5 clips

Imran reveals he created an AI clone of himself using HeyGen that was so convincing his friends couldn't tell it wasn't him.

The clock feels loud. The host opens with Riverside, a ticking-bomb joke, and the slightly chaotic live feel that keeps the conversation from getting too glossy. Then Sahil's background lands: data scientist, founding AI scientist at Bedrock Security, podcast host, RSA speaker, and someone working right where AI meets security. The tone stays loose. The host wants candor, and their shared background lets them talk without sanding everything smooth. Then comes the blunt question. If a non-engineer can use a few prompts to build something with embeddings and knowledge graphs, why does anyone still need an engineer or data scientist? Sahil does not dodge it. He says it depends on the application, the accuracy needed, and whether the person is mostly wiring together standard techniques. Small tasks can look like magic. A function, a script, or a narrow implementation can work surprisingly well when the model has a tight problem. Big systems are where the shine starts to wear off. Once the ask turns into a full application that touches many systems, context, scope, and scale start causing trouble. Knowledge graphs make that concrete. Sahil points to redundancy as a real scaling headache when uploaded data repeats itself inside the graph. That is the useful tension here. AI can make the jump from idea to prototype feel tiny, but serious products still need judgment about accuracy, scale, and redundancy. The ending turns more hopeful. Sahil points to passwordless flows, Face ID, and thumbprints as places where security can remove friction instead of adding another annoying layer.

As heard by us

A grounded AI-security conversation that keeps the tradeoffs in view.

Sahil keeps the discussion on one plain question: what AI actually changes in security, and where human judgment still has to stay in the loop.

Read the full review in PlayNext →

Why you'd press play

Hear the blunt case for where AI helps security and where it still breaks on scope and accuracy.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Imran's Podcast