žygimantas (@zhygis) header
žygimantas (@zhygis) profile picture

fixer @tempo

ams, nl
Joined October 2014
Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)Photo by žygimantas (@zhygis)
žygimantas retweeted
Today we're releasing the Tempo Facilitator -- the fastest way to start, and scale, services that accept machine payments.
21
15
6
132
26,522
I unfortunately can't take seriously the big AI devrels on Twitter. Popular software factory repos just don't compute for me as cred for "How to use AI". The people I trust the most are people focused on correctness whose bottom is on the line. Kernel/Inference folks, TigerBeetle, our own Tempo team etc. Basically systems folks. I believe that 90% of application layer advice out there is garbage and actively distracting you.
43
48
6
665
32,491
žygimantas retweeted
🇳🇱 Live in the Netherlands? Track the rain without getting tracked + stay dry right from your Lock Screen :) ⛅→ droog.is/lekker Help je me even? Droog is my love letter to the Netherlands. Als je hem gewoon in je groepsapps kan delen, dat zou echt enorm helpen! 🙏☀️
96
38
23
409
57,115
++! they proactively monitor our usage and flag mistakes that we might have in our setup
Full disclosure: I used to work at PostHog I know most of these posts are clickgagement but for the genuine folk who end up believing that building in-house or self-hosting is the better option... I can guarantee you PostHog is the most economically viable option - Their free tiers are insanely generous, not to mention their startup plan - Believe it or not there are still economies of scale in SaaS, not to mention increased reliability of a managed solution - The have one of the best billing control options (second only to AWS that I've seen) - They literally offer to refund accidents: posthog.com/handbook/growth/… (only co. I've ever seen do that) For anyone who legitimately burns through $$$ of PostHog credits I would goes as far as to say that you should refocus your efforts on internal spend, I'd guess there's serious improvements that could be made to your own bottom line
1
7
499
žygimantas retweeted
I'm open-sourcing Dasher, a lightweight headless browser that's faster and cheaper than Lightpanda and Chromium! (1/4)
20
1
64
3,983
žygimantas retweeted
We don't use anything instead of poteto mode. And it works fine without :) I also personally think that invoking skills with slash customs is probably an anti-pattern: - you want the skill to be discoverable enough without a slash command - if it's a highly specific thing that's deterministically invoked, it makes me wonder why one uses the model as a workflow executor vs giving it a high-level goal and figuring it out - codebase should be self-explanatory and legible to the agent without deterministic invocation by human. If human has to tell agent when and how to review with highly specific skills, something probably went wrong somewhere else before
16
7
4
160
25,294
žygimantas retweeted
increasingly think that there are only two successful paths for building software post 2026 1) big orgs with a semi-durable compounding core business which can be tuned and operated by a shrinking team. Improvements can easily be underwritten by a cost curve of token spend. 2) small teams (<5) of people who have overlapping skillsets and trust each other's judgement. Low-trust small orgs or existing orgs which hold onto existing ways of operating will be pulled apart by the entropy of their own slop
5
7
3
205
14,362
žygimantas retweeted
i recreated THE ENTIRE WORLD in minecraft at 1:1 scale. every building on earth, including your house! to do this, i trained a SOTA world-to-minecraft generator on maps data, complete with furnished home interiors! play and check out your neighborhood: worldcraft.gg
1,717
6,459
2,646
123,739
20,802,952
are there startups that started in the last 3 months and have specifically separated all of their confidential, work and personal devices/apps for agents? feels like if we would've started Tempo now, this would be the biggest change I'd make from the getgo
1
13
865
having to deal with RBAC, privacy, etc. is such a pain
2
119
žygimantas retweeted
When you die God plays Clams Casino
1
3
33
2,310
we're able to save half of our compute (and thus double it for things that actually need it) by just not using sandboxes for most things! (data from internal harness)
Legitimately can't believe people are still talking about managing fleets of sandboxes for their cloud agents infra. It's H2 2026. Harness out of sandbox. Virtualized bash & file system. Code mode. Sandbox only when you need to do something heavy. Not to edit files & simple data.
4
1
65
9,126
it's so nice to be able to give agents arbitrary compute and let them run with it
3
341
žygimantas retweeted
I have a major result to share. I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China. This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model. Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves. I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems. I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used. It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20. These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs. I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo…. There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.
1,823
6,424
2,210
41,251
6,798,000
we did this using nanocode, and its magical - your mac is just one of the machines the agent can use
3 more days till Conductor kicks us out their cloud whittle it down to 4 1. ❌ Amp 2. ❌ Boxes 3. 🔎 Capy 4. ❌ Cursor native agents 5. ❌ Delta 6. ❌ Emdash 7. ❌ Factory 8. ❌ Hoplite 9. ❌ Nuphos 10. ❌ Open Session 11. ❌ Orca 12. ❌ Orchestrator 13. 🔎 Paradev 14. ❌ Paseo 15. ❌ Reason Machines 16. 🔎 Replicas 17. ❌ Tembo 18. ❌ Traycer 19. 🔎 Vespin 20. ❌ Vorflux 21. ❌ unreleased app
4
414
žygimantas retweeted
Agents assigning tasks and collaborating between machines/harnesses: Claude <> Codex <> Opencode Works behind NAT (private networks) via QUIC Check it out! github.com/peerspan/peerspan
1
5
422
i'm back to being convinced that subagents >= subthreads
I’m experimenting with doing all my work through a single session. I told it to delegate everything to a roster of subagents who can message eachother and also delegate. One of them is Fable running every so often to improve the session based on a note shared between all agents. It decides which members of the roster are permanent (ScreenshotBot), which members are temporary (Investigator of Incident #22), and when to hire, promote, or retire others. Super promising so far, it's way less fatiguing to chat with the coach than tab between sessions, and new tasks are hitting less papercuts by learning from their predecessors
1
9
603
also, skills have 3 months left in them
58