We've been cooking Latitude v2 and we're launching soon.
v2 surfaces real failures from your AI production traces, then auto-generates evals so the same problem doesn't repeat.
Sentry, but for AI agents.
Link to the waitlist below🔥
Do you use Claude Code and want free merch?
We’re opening a merch store to celebrate the v2 launch and giving away 5 t-shirts every week.
To participate:
---
1. Connect Claude Code to Latitude with 1 command
2. Every Friday you'll get an email with your Claude Code usage stats
3. Screenshot it and post it on X tagging @trylatitude
The top 5 posts each week win a t-shirt. We'll DM the winners directly.
---
Link to the merch store: https://t.co/Qv8mpb03x6
Link to Latitude: https://t.co/p9mznoyTBM
Claude Code observability is on Latitude🔥
Install it with 1 command and see full sessions: system prompt, every tool call, subagents, per-turn cost.
Useful if you keep hitting @claudeai limits or run Claude Code as a harness for your own agent.
Free and open source.
We’ve put together a guide on how to get started with AI evals.
It covers:
- The 3 types of evals and when to use each
- How to design a repeatable evaluation framework
- A checklist to implement this in your own workflow in < 3 days
We've been cooking Latitude v2 and we're launching soon.
v2 surfaces real failures from your AI production traces, then auto-generates evals so the same problem doesn't repeat.
Sentry, but for AI agents.
Link to the waitlist below🔥
What's inside:
→ Generic vs. targeted evals (and why the difference matters)
→ Online vs. offline regression testing
→ Human-in-the-loop evaluation + annotating traces
→ LLM-as-a-Judge: when to use it and how to write judge prompts that work
→ Programmatic rules + how to integrate all 3 eval types
→ Building a golden dataset from real issues
When we released the first track of our AI PM course, 1,000+ people registered in 3 days.
That pushed us to ship the full course quickly.
If you’re building AI products and want a practical framework, this is for you.
The full course is now on YouTube 🚀
New video on the channel 👇
We walk through how to manage prompt versions in Latitude: publishing, tracking changes, and safely rolling back when quality drops in prod.
Watch it here → https://t.co/6Z72JtLlB0
We just launched Rosetta: a translation layer for LLM provider messages.
Rosetta converts messages between OpenAI, Anthropic, Google, and custom formats into a single standardized format (GenAI).
If you're working with multiple LLMs, this will save you hours of work.
Try it out: https://t.co/Dvs3EbdcUs
Yesterday we finished our 6-week AI PM Course!
We’ve collected everything you need to level up as an AI PM into the course: 6 tracks, 30 practical videos, quizzes, and a certificate included.
It's 100% free.
We've historically done a terrible job explaining how @trylatitude works for developers, in part because I'm just very bad at the attention-based distribution game we are all playing these days.
So, let me set the record straight.
Latitude is a fully OTEL-compatible observability platform that supports basically any LLM stack in Python or Typescript. You don't need to migrate your prompts to start using Latitude, nor ever buy into our other abstractions. Simply wrap your LLM generations with our telemetry script, and you are good to.
We already handle millions of traces per month and have been growing consistently for the past 6 months.
Link to docs in reply 👇
It’s fully practical.
You’ll learn how to build AI features that don’t break in production.
You can sign up and learn at your own pace:
https://t.co/2LQSpN1kCb
We’re launching a free course 🔥
Right now, a lot of PMs are suddenly expected to “figure out AI”, but very few have actually been taught how to build AI features that behave reliably.
After working closely with hundreds of product teams, we kept seeing the same patterns: what works, what breaks, and why.
So we turned all of that into a 6-chapter course.
New chapter every week starting tomorrow (Dec 3).