Shahram

Building @cleric_io. An AI SRE for on-call engineers that diagnoses production issues before you have to.

[Shahram @shrumm]

This’ll be fun - condensing years of lessons learnt on agent self learning into this talk. See you there!

[Cleric @cleric_io]

Our CEO, @shrumm, is speaking at AI DevSummit on May 27 about how we build agents that capture engineer intent (implicit SLOs, acceptable degradations, tribal knowledge) and what happens when that context goes stale.

Free ticket → https://t.co/I4iDfpmqgD

[Shahram @shrumm]

Is anyone using Gemini CLI for coding or agent related tasks? Seems only Codex is getting mindshare vs CC

[Shahram @shrumm]

what do you call your personal assistant agent? I instantly gravitated to Jarvis (I know, how original). I set one up for my wife and she spent a good while going back and forth and settled on Dobby.

[Shahram @shrumm]

Highly recommend if you’re in SF and building in AI. There’s no shortage of events here but 1) @diptanu is awesome and 2) the closed dinner format is my favorite.

[Diptanu Choudhury @diptanu]

We are hosting a small dinner in San Francisco at Spruce on the 24th of March for builders in AI Native software companies. If you working on harnesses for coding agents or any kind of back-office or background agents, would love for you to join us!

We have a few spots left.

Link - https://t.co/1hKTOlmvt8

[Cleric @cleric_io]

AI SRE doesn't replace engineers. Instead, it gives engineers their time back.

In our webinar, we talked about Cleric's approach to creating a self-learning AI SRE, and how it actually works in production. → https://t.co/5g2efNztIt

[Shahram @shrumm]

X has been incredible for good downtime reading material.

I spend all my time in the AI world so it was good to catch up on crypto and speculation.

Two great reads
A sobering take:
https://t.co/KxIKprGLpN
A more positive take:

[Neil @neilhar]

https://t.co/DAdYUdETiB

[Shahram @shrumm]

@kwharrison13 Great advice!

[Cleric @cleric_io]

We’re excited to announce Cleric, the first self-learning AI SRE. It continuously learns from every incident and helps software engineers move more quickly to resolve issues.

We’ve proven that our approach works. Now, Cleric is available for all teams: https://t.co/1BU09FVpVo

[Shahram @shrumm]

Even if you haven’t done quite as well financially - there’s nothing better than family. Nothing comes close.

[Mitchell Hashimoto @mitchellh]

I was someone who through my 20s wasn't even sure if I wanted kids. Work was my passion and I enjoyed it deeply. I filled up two passports. I did well financially. And yet, it's incomparable to the joy and purpose having children has given me. Like, not even close. Its crazy.

[Shahram @shrumm]

@GergelyOrosz A use case that will become prominent soon is how agent friendly these platforms are.
- Grafana’s self hosted option helps a lot here. We use it extensively in our simulation bench.
- LLMs are very good with PromQL + Datadog
- Rate limits matter (Grafana is great)

[Shahram @shrumm]

@GergelyOrosz @GergelyOrosz sounds like you had fun too - did you enjoy it?

[Shahram @shrumm]

@GergelyOrosz This is a false choice. Programming in English means you can say
“Make it #7DB096”.

[Shahram @shrumm]

Evals, evals, evals…
- Steve Balmer 2.0

[Willem Pienaar @willpienaar]

There's a huge gulf between pristine offline evals and the chaos of production environments.

In this chat with @Dpbrinkm, @sh_reya and I share a lot of the techniques we've learnt in bridging that gap, from online task-based metrics to new UX paradigms for collecting feedback.

[Shahram @shrumm]

I’ll be at the AI Agent World Tour for @cleric_io - we’ll get into all your AI SRE questions.

[Michelle Fang 🌁 @michelleefang]

Wednesday 5/28

‣ ai + health demo night #005 https://t.co/aKSwD4ZDEd

@michaelraspuzzi

‣ SF DEMO NIGHT 🚀 (w/ The GenAI Collective) https://t.co/VUEE56wADo

@genaicollective @roam @shadeformai @chappyasel

‣ AI Agent World Tour in SF https://t.co/4ZhxTBOfNp

@mlopscommunity @adamboazbecker @RahulParundekar @sramanan @LeoWalker9 @onlinematters @theaievangelist

‣ All Things Web at Vapi https://t.co/cTn0AUaTHZ

@andrelandgraf @allthingswebdev @esthor

‣ AI Dev Tools Night https://t.co/Dc6zFopV6Z

@jdorfman @ykdojo @BrandonWaselnuk @mintlify @rootlyhq @animosf

‣ Founders' Olympics: Tech Chess Tournament @ The 1097 https://t.co/Mb2K15ghXO

@remy_mess @AltalibKhalid @jvheaney @lottsnomad

[Shahram @shrumm]

@GergelyOrosz Peeps in SF this Wednesday, both @willpienaar and I will be at https://t.co/8oujKefYKD. It’s engineers only and we’ll talk about @cleric_io - it’s in production handling on-call diagnosis, remediation is left to engineers. Head over if you’re around!

[Shahram @shrumm]

@simonw Agents are also a buzzword so incentives aren’t aligned to have a clear definition - the push is to make the umbrella as wide as possible.
My cofounder was in a panel discussion at IA, the first question was “what is an agent”, took up 90% of the allotted time with no consensus.

[Shahram @shrumm]

AI SRE is a category where move fast and break things doesn’t work. Hear from @willpienaar on how we’re staying focused on earning engineer trust.

[MLOps Community @mlopscommunity]

In this episode of the MLOps Community Podcast, we chat with @willpienaar about how his team at Cleric is building an AI SRE that helps engineering teams autonomously root cause production applications.

[Shahram @shrumm]

@diptanu @tensorlake 💪🏽

[Shahram @shrumm]

This was a labor of love to get all the technical details. Enjoy with a good cup of ☕️.

[Willem Pienaar @willpienaar]

So what exactly *is* an AI SRE?

We wrote a deep dive on how AI agents can help with production operations - investigating issues, diagnosing problems, and driving them to resolution.

https://t.co/KqHupD1K66

cc @jeremylewi @niallm @Dan_Jeffries1 @rachelchalmers @hwchase17

[Shahram @shrumm]

The tricky thing with root causes is that there’s no clear definition of where to stop.
I mean… the root cause of any production issue is the Big Bang 🤷🏽‍♂️.
What most of us mean is - I just need enough detail so I know what to fix.
#sre #reliability