Kognitos at TechCrunch Disrupt 2023: Binny Gill Fireside Chat
Binny Gill (Kognitos) and Jim McCullen (Century Supply Chain) on TechCrunch Disrupt: humanizing computers, the pilot's checklist analogy, and ending the 20-year SQL-for-business-users mistake.
About this fireside chat
Recorded on the TechCrunch Disrupt 2023 stage, this 21-minute conversation features Kognitos founder and CEO Binny Gill in conversation with Jim McCullen, CIO at Century Supply Chain Solutions. The discussion ranges from the limits of traditional automation to the very specific pain of running global logistics — and what changes when business users can talk to the machine in English.
Binny's framing — the “dark ages” argument
- 0.5% literacy: only one in 200 people in the world can teach a computer something new — “the dark ages” despite 100% computer dependence.
- Wrong direction: seven decades of trying to teach humans to speak machine languages has not worked. The fix is to train machines to speak human.
- The pilot's checklist analogy: a pilot's flight plan is short and clear (in English). The troubleshooting manual handles the infinite edge cases. Today's RPA tries to encode the troubleshooting manual into the flight plan — which is why business automations get more fragile, not less, over time.
Jim's pain — supply chain reality
- No equivalent of FlightAware for ships: kayak.com gives every airline the same answer for a flight. Three logistics sites can give three different answers for the same vessel.
- Thousands of vendors, hundreds of carriers: every bill of lading is technically a standard document, but every carrier has a different variant. Straight templated extraction breaks.
- Offshore document teams: until now, the answer has been humans in Asia reading the documents manually. Generative AI is the first realistic alternative.
Where Binny and Jim agree
- Developers should handle the “happy path” (the first 80%). The complex last 10-20% (every edge case) should be handled by the business user in English, during runtime, with the platform learning as it goes.
- IT keeps governance, security, and the connection layer; business users own the process logic and the exception responses.
- The shift from punch cards → assembly → symbolic languages → today's automation languages is followed by a jump to natural language as the new automation language. Kognitos is built for that jump.
The closing prediction
Binny closes by saying AI will accelerate innovation by at least 100x because today's 25 million developers are the bottleneck for all innovation. When anyone with an idea can “program” a computer the same way they'd brief a teammate, that bottleneck disappears — but only if the resulting program is human-readable, auditable, and runs on a deterministic interpreter rather than relying on a black-box LLM at runtime.