Inngest is agent infrastructure that lives in your codebase. Write your logic as functions that automate retries, control flow, and unlock full observability.
It just got much easier to evaluate agents in production—against real business outcomes.
Agent Evals are now in public beta on v4.8.0 of the TS SDK, and include:
💯 𝗦𝗰𝗼𝗿𝗶𝗻𝗴: Attach a metric to any run or step. Include outcomes that don't arrive until after the run
We got such great feedback on Sterling and Amanda's demos at our last meetup in SF, that we decided to run it back—virtually!
Join Inngest and Vapi for a hands-on workshop where you'll learn how to build a production-ready voice agent. Luma link below!
What a week! We welcomed over 1,000 of you to chat about building AI in production, across multiple events at our SF office.
And we want more.
Subscribe to our Luma for the next:
- 𝗜𝗻𝗻𝗛𝗼𝘂𝘀𝗲: Daytime DJ, demos, and catered coffee
- 𝗔𝗜 𝗶𝗻 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻: Meetup
Did you catch our car wraps last week?
Of course you did. You're an active participant in your own life—enraptured by the majesty of every moment.
step.run = build agents that can't fail
defer() = wait until real world events arrive
step.score = attach a
ICYMI we added Scoring, Experiments, and our new Defer primitive to Inngest.
Build durable agents with step-level scoring that you can defer until the real-world signal arrives.
Check out our latest pattern for exactly how to set it up.
Full house at InnHouse yesterday, and full waitlist for this morning's breakfast with @nebiustf. Come by for tunes and tacos 8-930am 🎶 🌮 No conference pass required!
Next up on First Pass: A discussion on Loop Engineering
No one is prompting anymore. It’s all about writing loops with goal functions and going to sleep
@djfarrelly from @inngest joined us to break down Loop Engineering. What it means / what are the implications
I'm at @aiDotEngineer speaking about how your agent architecture has a half-life of 6 months (and what to do about that) at 12:05 @ Expo NE - come by and say hi or catch me at the @inngest booth after
Nothing is a one-shot prompt anymore. So why are your evals still shaped like one? Inngest launched Agent Evals to score the 𝘄𝗵𝗼𝗹𝗲 run—not just the last answer—based on outcomes that might come days later.
Come check it out at @aiDotEngineer today 🎤 🍝
Next up on First Pass: A discussion on Loop Engineering
No one is prompting anymore. It’s all about writing loops with goal functions and going to sleep
@djfarrelly from @inngest joined us to break down Loop Engineering. What it means / what are the implications
Your view this week 📸
Our view this week 🦅
Join us on the show floor at @aiDotEngineer World Fair in SF, or one of our adjacent events across the street (yes literally across the street)!
Next week!
🌯 Burritos
☕️ Beans
🎶 Beats
Let's go!!! We're almost at capacity for our grab-and-go lunch with @e2b + DJ Melevate 🎶 on July 1.
And Señor Sisig waits for no man.
Register below ⬇️
