You probably already know this but sou can get a lot of Levine’s writing without a paid Bloomie subscription by subscribing to his Money Stuff Newsletter.
As soon as tokens stop stop being subsidized, heavy agentic use will become as least as expensive than paying an (entry level) employee. When this happens many companies will trade off havy tolen usage for (maybe a bit slower, bit less accurate) employees again.
Tis is very simple and very cool! Just installed it on my Hetzner box where I run a remote controlled local agent so now I can basically chat/email a video link to get a summary and/or ask questions. The only issue was YouTube's PO Token requirement (web/mweb clients refuse to serve formats from datacenter IPs without a valid Proof-of-Origin token.) So I had to find a client that still work without PO Token first. Thanks for sharing!
I appreciate Evans’ work and wrote an “antithesis” to the Nov 2024 iteration of this. Given the pivot to “models look likely to become infrastructure” I might want to update my take.
Commenting here since this is somewhat related to the research (1) I have been doing lately:
UMA is one transparency layer.
The bigger one is the public WebSocket feed itself, which broadcasts post-match book state but hides taker identity. So trade direction inferred from the feed only hits 59% sign agreement vs on-chain ground truth.. that breaks six standard microstructure measures (effective spread, Kyle's lambda, Amihud, etc.) on most top-100 markets.
I believe any Polymarket research that depends on direction should to source it from on-chain OrderFilled events, not the feed.
Don't necessarily agree with all of the writing in this article. But I just spent the past weeks reading into the whole AI valuation topic again (did dis in March already (1)). So I'll try to add some perspective:
Form my research (1) Four widely cited 2025 enterprise AI revenue figures span 40x ($37B Menlo, $1.478T Gartner). The audit-grade floor you could defend on a 10-Q against $690B of 2026 hyperscaler capex is just $63.2B, or 9.2% coverage. Telecom 1999 peaked at 28%. Even after netting only AI-incremental capex you only get to 12-15%, which is still below the closest analogue we have.
I like the idea. As others suggested it might be a good idea to drop the branding. Had the same considerations when I built a “Tinder” (1) for RSS Feeds. In the end it worked fine, if not better.