CEO @Tembo • prev founder @Astronomerio

Joined May 2008
ry retweeted
defensive ai is going to be big in 2027 needs to be big now wrote these last Dec: rywalker.com/agentic-defense… and rywalker.com/let-agents-figh… tldr; start spending as many tokens on defense as you do on offense (where offense = code gen)
2
2
21
829
the manipulation couldn’t withstand reality $TSLA up 6.2% $GM down 8.1% Mr. Market is 😎
$GM up 1% $TSLA down 1.6% Mr. Market is 🤡
1
5
432
(since the roadster demo event)
44
Cloud & Async Agents ▫️ Codegen @codegen ▫️ Cognition @cognition ▫️ Cosine @CosineAI ▫️ Factory @FactoryAI ▫️ Jules @GoogleLabs ▫️ OpenHands @allhands_ai ▫️ Tembo @tembo
1
1
3
118
$GM up 1% $TSLA down 1.6% Mr. Market is 🤡
1
1
5
1,016
0-60mph in 30ft @Tesla
1
11
568
yeah... we're interested @Tesla
7
485
one thing you don't realize until you try cloud agents is how much battery life you get back i run pretty much everything through tembo now, and my macbook lasts so much longer because all the heavy lifting happens in the cloud such an underrated benefit
2
1
11
989
anyone building a vulnerability scanning product for startups w simple, beautiful UX?
8
1
18
1,446
ry retweeted
it can be true that a group of early adopter “regular people” are rolling their own personal software…. and there is still opportunity to build and sell great software mid software would seem to be at risk, but i can’t imagine a u-shaped distribution of software quality… too many ppl building, for there not to be a preponderance of mid
3
1
7
909
All the code harnesses, all the subscriptions, same stuff you use locally, extended into the cloud.
Bring any major LLM subscription to Tembo! Claude, Codex, SuperGrok When connected, you only pay for VM compute!
3
1
10
924
it can be true that a group of early adopter “regular people” are rolling their own personal software…. and there is still opportunity to build and sell great software mid software would seem to be at risk, but i can’t imagine a u-shaped distribution of software quality… too many ppl building, for there not to be a preponderance of mid
3
1
7
909
ry retweeted
New in Tembo ✨ - Agent instruction version history: see past versions and roll back to a setup that worked - Browser push notifications for approvals, new PRs, and review requests - Claude Haiku 5.5 - GitLab MR follow-ups now stay in the same conversation Read the full changelog ↓ tembo.io/changelog/agent-ins…
1
2
1
18
1,524
Replying to @thomaspark_gg
@tembo works with claude + codex + supergrok subs runs any harness you want! claude code, codex, opencode, cursor etc bring a sub and only pay for some of the cheapest vm compute out there
1
2
14
360
Developers ship faster than the market can absorb. By the time there's demand, 100 people have vibe-coded the same thing. The edge isn’t the build anymore. It’s GTM.
5
1
10
1,698
Code review is collapsing under agent-generated volume. The fix isn't reading faster. It's adversarial agents: spawn agents to attack the work. No human should see a PR until it's been beat tf up.
5
1
12
879
the @cursor_ai sdk is getting good
2
387
Human-only interfaces are the new legacy. Every report, dashboard, and API you create should assume a machine is the primary user.
2
5
629
if a PR costs $1 in tokens to write... spend $3 in tokens attacking it with another agent... much cheaper than humans reviewing code the agent wrote... with the same agent that wrote it... is __not__ enough verification
4
6
862