Oct 8, 2026 · 1h 19m
Personal agents challenge AI’s business models and venture playbook
What VCs Really Think About Personal AI Agents | E2347
The panel connects advances in frontier AI to a practical question: whether agents can reshape consumer behavior, business economics, and startup investing.
- 1Frontier models may commoditize intelligence while application context, permissions, workflows, and real-world experimentation retain value.
- 2Personal agents could remove small frictions across travel, finance, insurance, and services, threatening businesses built on those frictions.
- 3AI investors increasingly weigh technical depth, direct product use, customer adoption, and conflicts among converging portfolio companies.
Don't miss
The panel’s discussion of agents eliminating the small frictions that support financial, insurance, travel, and service businesses is the episode’s clearest strategic warning.
The brief
Deedy Das moderates a roundtable with Kanu Gulati, Chris Farmer, and Sheel Mohnot on what recent AI progress means for science, startups, and venture investing.
The panel argues that better models will not erase every source of advantage: proprietary data, evaluations, application context, permissions, and real-world experimentation may matter more as intelligence commoditizes.
Personal agents such as Muse and Instinct become the test case, with travel, research, and complex planning showing the upside while sensitive data, security failures, and consumer distribution expose the limits.
The sharpest economic question is whether agents can remove the small frictions behind financial, insurance, travel, and service businesses, forcing companies to adapt to agent-mediated customers.
The discussion closes on venture conflicts, AI interdependence, SaaS’s uncertain future, and a diligence principle: use products directly and look past startup jargon.
What was said on this episode
18 statements · 8 positive · 8 negative · 2 neutral
Applying model intelligence with context and permissions is critical for effective agents.
“how you apply that intelligence and how you give it context and permissioning is going to be critical”
Listen at 12:34
Vertical AI companies with proprietary context can dramatically outperform general models.
“you can build data density in lots of different use cases where the model may have intelligence, but it doesn't have anywhere near the context that you have because you've... you've built that up, you're going to outperform dramatically”
Listen at 13:18
Current large language models are highly inefficient in training and serving.
“these models are very inefficient”
Listen at 14:59
LLMs and transformers will not be the only path to AGI or SI.
“I also don't believe that LLMs and transformers will be the only way for us to get to whatever we consider. AGI or SI”
Listen at 16:11
Physical-world data collection creates flywheel opportunities for AI companies.
“there is a physical world data flywheel opportunities as well”
Listen at 16:59
Compute will eventually stop being the main bottleneck for AI progress.
“at some point compute stops being the bottleneck in all of these things”
Listen at 17:26
Laboratory capacity, trials, manufacturing, and patents may constrain future AI-enabled drug development.
“it might be lab capacity the ability to do clinical trials manufacturing patents all that kind of stuff might be the constraint”
Listen at 17:46
Muse's high entry valuation carries significant investment risk.
“I don't think, you know, paying that kind of entry valuation, to me, the risks are significant”
Listen at 39:17
Investing at Instinct's price requires believing it becomes a major consumer interface.
“I think you have to believe if you're investing at that price, you have to believe that this is going to be at least one of the major consumer interfaces”
Listen at 39:38
Large technology companies have a major advantage in personal agents.
“I personally think that the large companies have a huge benefit here”
Listen at 41:13
Muse will not reach one billion daily active users.
“I don't think that there's a billion DAUs for this product”
Listen at 42:06
Personal-agent products can reach hundreds of millions of daily active users.
“there are certainly hundreds of millions of DAUs that this is a product people can use”
Listen at 42:46
The personal-agent category can become very large.
“I do think the category can be very, very large”
Listen at 44:16
A $10 billion entry valuation for personal-agent companies is concerning.
“A 10 billion price as an entry point, I would be worried about that”
Listen at 44:48
Silicon Valley's coopetition generally benefits competition and innovation.
“It's generally a net positive because it drives competition and innovation”
Listen at 57:42
AI companies targeting labor can significantly expand the addressable market beyond SaaS software budgets.
“the ai companies are often going after the labor on top of it so it expands the tam pretty significantly”
Listen at 1:06:26
AI will probably cause seat-based software pricing to shrink.
“seat-based pricing probably shrinks”
Listen at 1:08:35
AI has substantially reduced switching costs between software products.
“switching costs have gone away”
Listen at 1:08:51
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.