Podbit · The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
Most enterprise AI deployments fail not because models are bad but because agents spend most of their time and tokens searching for the right context. Treating AI like a brute-force tool burns money and delivers slow results. The solution is investing in context infrastructure before deploying agents.
Where this was said
At 24:50 · chapter starts 20:53
Harry declares that 2026 H2 and 2027 will be the years when enterprises stop writing AI checks on faith and demand ROI accountability. Arvind doesn't disagree but provides texture: customer support is the clearest win because productivity is directly measurable (cases resolved per day per agent). Coding is where the biggest AI spend has gone, but the paradox is real — coding speed increased dramatically, but product shipping speed hasn't followed, because coding is only one component of shipping. At Glean itself, Arvind admits it's hard to isolate AI's contribution from team growth and tenured experience. He then delivers the core insight: AI ROI is a throughput problem. [1] — Arvind Jain "Most enterprise AI deployments fail not because models are bad but because agents spend most of their time and tokens searching for the rig…" 24:50 Enterprises deploy AI in a brute-force way, letting agents assemble context from scratch for every task, burning tokens and time on setup rather than actual work. The fix is investing in context infrastructure upfront.
At Glean, essentially 100% of initial code is now written by AI, though the company enforces mandatory human code reviews before any code is merged.
Glean now has essentially 100% of its initial code written by AI — but enforces mandatory human code review before any commit. The paradox: review is now the bottleneck, not writing. Some companies eliminate reviews entirely, but Glean believes maintaining oversight is worth the cost while the industry is still in a learning phase.
Most enterprises connect AI to their systems and let it brute-force its way through context assembly — burning tokens, slowing down, and delivering poor results. The real ROI unlock is investing in context infrastructure so agents start with the right information rather than searching for it.
By 2006, AI algorithms were stuck because they were being trained on almost no data. Fei-Fei Li's insight: human children see tens of thousands of object categories by age 6 — so machines needed massive data too. ImageNet's 15 million images, combined with GPU power and better algorithms, triggered the modern AI revolution in 2012.
The ImageNet challenge pitted machines against humans on recognizing 1,000 object categories. Humans clocked a ~4% error rate. In 2012, a neural network smashed previous AI performance — and by 2016, machines had surpassed humans entirely. That single benchmark created the modern AI era.
When video was added to AI training data in 2023, something clicked: machines could generate plausible motion without knowing muscle anatomy — just from watching millions of cat videos. Sora's January 2024 release proved AI had crossed into temporal, physical understanding.
AI is trained on the internet — the largest archive of human behavior ever assembled. But the most profound human thoughts, Picasso's creative flash, a private childhood memory tied to a gray cup, have never been uploaded anywhere. That's the gap AI cannot close.
AlphaGo's Move 37 against Lee Sedol shocked Go masters — no human had ever conceived it. But Fei-Fei Li urges caution: Go has fixed mathematical rules, and AI's bigger compute simply found a configuration human memory couldn't retain. That's creativity in a constrained space, not the open-ended kind.
Surface-level intuition — the kind you can describe in words — is just context, and AI already handles that. But the deeper kind: the feeling shaped by what you ate, your hormones, and a mood you can't name? That has no sensory apparatus feeding it to any machine. It's inaccessible, and will remain so until brainwave-level sensors exist.
Self-driving cars already exist. But the real robot revolution — robots assisting the elderly, fighting wildfires, supporting overworked nurses — is a 20-30 year arc, not a 2-year one. Hardware plus AI moves slower than software alone, but the impact will be civilizational.
Language AI is powerful, but humans evolved in a spatial, physical world. WorldLabs is building foundational models for spatial and 3D intelligence — letting people generate entire environments from a sentence or sketch. The applications span filmmaking, robotics training, architecture, and healthcare.
Modern AI didn't emerge from one breakthrough. It took three things converging at once: mature neural network algorithms, the ImageNet large-scale dataset, and fast GPU computing. When all three lined up around 2012, the revolution was inevitable.
We use essential and analytics cookies to run Vuci. To understand how the site is used: Privacy Policy.
Install Vuci on your phone
Add it to your home screen for a faster, app-like experience.
Install Vuci on your phone
Tap the Share button, then “Add to Home Screen”.
A new version is available
Reload to get the latest Vuci.