playtesting every harness again, i cannot believe how you folks don't lose it over the TUI taking over 50ms to start, and TTFT >2s - these companies are achieving AGI and here we're stuck with the slowpoke harnesses wtf
Show more
our special guest is out!
excited to have @CompleteSkeptic join us for Frontiers!
Diogo & @typesafeai have taken over the developer community with Jev, and we're thrilled to see what they have to share at Frontiers!
apply below, closes this week! oct 12-14 fort mason, sf.
Show more
5 days left to apply to Frontiers by @paradigm!
talks from @vercel @physical_int @AmpCode @pierrecomputer @tempo @NousResearch @vectorware @commonwarexyz @PrimeIntellect + all-day hacking + epic merch
oct 12โ14, fort mason, san francisco.
apply below!
Show more
one of these apps must adopt zk passport so everyone follows suit
crypto and ai are converging on the insight that TEEs/confidential compute is endgame technology bc of verifiability and programmable privacy
excited to see where this lands (and maybe read my post from 2024 on the 5 levels of secure hardware below)
Show more
we're about to release @tempo an aggregation service for routing across both tools and across other aggregators
like a search engine where you ask for outcomes instead of specific tools and it figures out what needs to be stitched together & how
partners at tempo dot xyz
Show more
Excited to share Wave 2 of Frontiers by @paradigm Speakers! Frontiers is our intense annual conference & hackathon for those beyond the edge in SF, Oct 12-14th 2026.
- @johannes_hage of Prime Intellect
- @_patrickogrady of Commonware
- Sam Herring of Nous Research
Apply to attend in the link below!
Logistics:
- Wave 1 of acceptances goes out today.
- 2 weeks left for any new applications.
- Merch and more sick videos to be announced.
- 1 more super special speaker to be announced!
Show more
@thorstenball Yep I think Amp probably only game in town that gets his things should be
Introducing Speculative Programmatic Tool Calling (sPTC)!
A general class of technique for speculating on tool calls during code generation in a harness and queuing them early to overlap with token generation + REPL execution time.
Blog:
Show more
Write up : (not up to date)
Sample Code :
paradigm is hosting a LAN party next Friday 8/28 in SF. 5v5 league, smash, halo and more. we have some spots open, DM your most impressive video game stat for an invite. glhf
all our rust OSS repos (alloy/foundry/reth/revm/tempo etc.) do not depend on arrayref / are unaffected by the supply chain compromise - stay safe folks
anyone tried hooking on apply_patch / edit tool calls and running a very fast LLM scanner on it? (i know ppl run LSPs and stuff already)
How is your day going?
++ on the s3/wal arch since first reading turbopuffer blogs
Rethinking delta sync at scale with @turbopuffer.
Engineer @traverspw on how we made Linearโs sync latency predictable as workspaces grow:
as we keep moving towards personal software & owning the tokenflow, the first thing to be tackled is the "personal harness" - implemented as "ai inside an app" vs "app inside ai" (i.e. the inversion of "codex has browser pane" - "app in your browser has codex inside of it")
fx by vercel is act 1
Show more
is 6.3mb. It starts up in 10ยตsยน. It's a Zig-compiled static ELF binary, or an even smaller ๐๐๐๐๐ก.๐ ๐๐๐ยฒ
AI will make most infrastructure natively optimized. Case in point: fx can literally finish tasks faster than other agents can boot.
Fast is a one-way street.
ยน that's microseconds
ยฒ fun fact: the wasm build is actually *smaller*. A lot of binary weight comes from the TLS/HTTP stack. In WebAssembly, fx delegates ๐๐๐๐๐() to the JS runtime instead. You can try this on
Show more
The trend of using S3 for databases is continuing, if not gaining momentum. I've seen three or four new examples in the past week with Cursor Origin being the headline yesterday. I wrote about why I thought this would happen a couple of year ago:
Show more
seeing a lot of love for WAL+S3
read how Chroma does it - open-source
