Sander Saar, founder of AgentCDN, making videos work inside AI

Transcript

1. 0:00

   We're massively underusing today's AI, not because the models are weak, but because our ambition is. Microsoft CTO Kevin Scott calls this capability overhang.

2. 0:08

   So the models are more powerful than what we collectively are using them for.

3. 0:12

   This overhang exists at every level. At the research and model level, Ilya Sutskever, the co-founder of OpenAI, points out that models now crush hard benchmarks. They're surpassing humans in law with bar exam, competitive coding, data analysis.

4. 0:24

   They are doing so well on evals, but the economic impact seems to be dramatically behind.

5. 0:30

   At the application level, Kevin Wilder, the chief product officer of OpenAI, says that models today are not intelligence limited, they're eval limited. They can actually do much more and be much more correct on a wider range of things than they are today. And at the user level, while about half of Americans have tried AI, only about 1 in 8, less than 12%, use it daily. In businesses, only 25% of the CIOs have deployed anything significant, and 40% aren't planning it until 2026. Why did our AI pilot fail, or your friend telling AI doesn't work?

6. 0:57

   It's almost never the AI, it's the CTO and the product question. Legacy systems, messy data, security, privacy worries that still consume most of the tech budget. Classic tech deployment problem. The same reason why cloud still runs only about a third to half of the enterprise workloads 10 years later. And when we do use AI, most of us just do the same job a bit faster. Ben Evans showed the tech adoption in 3 stages. First, absorb. Automate the obvious stuff. Type in a prompt, get a paragraph, and done. Marshall McLuhan called this the horseless carriage syndrome.

7. 1:23

   Using advanced tech for the most basic things. It's useful, but it's limited. But this is where most teams are. Next level is innovation. Means shipping something that only works because AI exists. This we start unlocking true value. Finally, disrupt. Build like you started from zero. If you rebuilt the product from scratch, what would it look like? The real opportunity isn't doing your job faster, it's doing a different job entirely. Moving from doer to orchestrator, engineering to engineering manager, writer to editor.

8. 1:48

   Kevin Scott set the challenge: think about how you can set your ambition level to 11, target some things that are, you think are barely possible.

9. 1:55

   And Steve Jobs gave us the playbook for new tech: you can't start with the technology and try to figure out where you're going to try to sell it. And you've got to start with the customer experience and work backwards to the technology.

10. 2:09

    The models are ready. We're the ones lagging.