A green test run can still hide a changed answer.
In this example, the tests confirmed the retriever ran and the data was parsed, but nobody asked the agent about a copay. It started saying $45 instead of $25.
Evals from code fail the pull request when the answer shifts.
1
79
Nobody wants to be the person staring at a graph during a rollout.
Set guardrail metrics and thresholds before the release, and let your configurations decide whether to continue, pause, or roll back.
More in our guide: bit.ly/3VuMxpG
61
One bad change shouldn’t mean rolling back the whole deploy. 🙃
Our guide shows how: bit.ly/48nTlIH
One rule: real stories only.
Our CPO @jnolen sits down with engineering leaders who’ve built AI software factories to talk about where agents helped, where things broke, and what they’d do differently.
Hear it from the builders: bit.ly/4BlCkfx
1
119
Release Autopilot is now generally available!
Your agents are already writing the code. Now your releases can run themselves, with your teams setting the rules.
Meet Release Autopilot: bit.ly/4jiUZRY
1
115
We see a crowded Venetian ballroom in your future. 🥠
Our CEO, @edith_h, explains how engineers keep agents in charge so they keep shipping. Add it to your @awscloud re: Invent agenda before the room fills up.
Add it to your agenda: bit.ly/4BlCkfx
1
139
Agents can write the code...then what?
Join @dgorton and @codewithalexis at our next Runtime Labs workshop. You’ll take a feature from PR to production with Release Autopilot.
Save your spot: bit.ly/4sWvLwc
1
119
Your team isn’t the only one shipping differently these days. @heyjemiah at @newrelic and Griffin Ashe at @github see this across thousands of customers, and the pattern is real.
Here’s what a release looks like when it runs itself. Release Autopilot spots risky changes, adds flags, rolls out progressively, rolls back on degradation, and cleans up after. Your engineers set the guardrails.
→ bit.ly/4jSkSsF
*Select features in beta.
Release Autopilot is here! Flagging, release, rollback, and cleanup now run themselves while your guardrails stay in charge.
Full story: bit.ly/4jSkSsF
*Select features in beta today.
This will be much needed after a long day at @github Universe.
Join us on Wednesday this week after GitHub Universe for bottomless drinks and time to decompress with us and our friends at @LaunchDarkly 🤝
Food, drinks, and stunning water views — what more could you ask for.
RSVP before spots run out! luma.com/shipsip
3
245
Going to @github Universe? Your coffee strategy matters as much as your session schedule, and we did the fieldwork for you. ☕
Report back at the booth → bit.ly/4de26Id
1
164
We’re a launch partner in the @SpaceXSI Partner Network! 🚀
SpaceXSI is building some of the most ambitious agent systems anywhere, and our job stays the same: when agents do the work, someone still has to control what ships.
1
1
219
Ghosts? Fine.
Zombies? Whatever.
An agent loose in prod with no rollback plan? NOW we’re scared.
Conquer real horror at our Halloween Hack-a-thon! 🎃
👻 Join us if you dare: bit.ly/4hZhlHB
193
The control freaks are coming to Las Vegas! 🎰
Find us at @AWS re: Invent, booth #1645, for demos and sessions on controlling the AI software factory.
Calendars fill fast in Vegas → bit.ly/4BlCkfx
1
2
212
Nothing beats @github Universe season in San Francisco. 🌁
The city fills up with builders, and we’ll be right in the middle of it both days.
Find us there → bit.ly/4de26Id
1
184
POV: an agent just opened a PR for a new in-app assistant, and instead of panicking...you press play on this video. 😏
234
Our next Control Panel is Nov 5!
How engineering teams stay in control as agents take on more of the work. Leadership from LaunchDarkly, plus a special guest.
Save your spot: bit.ly/4iWBDT8
If you haven't set up an Adaptive Trigger yet, this is your sign.
The whole setup takes a few minutes: pick a metric, set a threshold, and you're done.
Bonus: it works on AI configs, so it can swap model providers automatically.
Full demo → bit.ly/4htKNnz
1
234










