Langtail is an LLMOps platform that helps teams speed up the development of AI-powered apps and ship to production with fewer surprises.
With Langtail, you can debug prompts, run tests, and observe whatโs happening in production.
๐ Hi Product Hunters! Iโm Petr, co-founder and CEO of Langtail.
Over the past year, Iโve spoken with dozens of founders and engineers about their experiences building AI-powered apps. I heard from almost everyone that taking an app from a simple demo to a great product that can be shipped to users is HARD WORK. Some of the common challenges I heard:
- LLMs can be unpredictable, making traditional testing tools more frustrating than helpful.
- Writing a good prompt is an art form, and sometimes the best person for the job isnโt an engineer - which is a problem if all of the prompts live in git.
- The best practices for using LLMs are constantly changing as new models are released and as teams get more experience.
We built Langtail to address the pain points that a team experiences as they take their proof-of-concept and turn it into a real product.
Here are the top features in Langtail:
๐งช **Playground:** Debug and collaborate on prompts with your team.
โ๏ธ **Tests:** Create a suite of tests to evaluate how changes affect your promptโs output.
๐ **Deployments:** Publish prompts as API endpoints, allowing you to easily invoke them from your app.
๐ **Observability:** Logs provide insight into how your prompt is responding to real-world users and metrics help you monitor key metrics like latency and cost.
If you give Langtail a try today, use the promo code PH50 for 50% off for life. You can claim it in the Billing section.
If you have any thoughts about your first few minutes poking around the platform, drop them as a comment here and me or someone else on the team will respond!
Ever since I have known @petrbrzek, he and his team has always built tools you want to use - attention to detail is their brand. And current state of LLMs can really use a debugging & testing workbench :).
@tomaskafka We like to think so! We've been obsessing over every detail of Langtail in preparation for this launch. It isn't perfect, but we think it's the best UX out there today. ๐
Been using LangTail for a few months now, highly recommend. It has kept me sane.
If you want your LLM apps to behave uncontrollably all the time, don't use LangTail. On the other hand, if you are serious about the product you are building, you know what to do :P
Love the product and the team's hard work.
Keep up the great work!
@snazzyham Thank you for your trust, Sohan, in using Langtail with your clients. We are working on new features that speed up your workflow with clients even more. ๐
This is already a killer tool for many use-cases we are already using it for. Super excited for the upcoming features and good luck with the launch and further development! ๐
Just yesterday I recommended Langtail to a friend who is starting to build with LLMs. I think it's the best choice out there to start with a simple prototype and move to production.
๐ Hey, I've tried your platform a few times now, and wow, the UX is outstanding! The effort that went into the tool you launched today is incredible - great job, @petrbrzek@rychlis and the Langtail team!
Whatโs in the pipeline for the future?
Any chance youโll support hot new open-source models like Llama3?
Also, super keen to see if there's a plan for a self-hosted/on-premise gateway option.
Thanks, and good luck on your journey! ๐
@good_lly Hello, thanks for your feedback!
Actually, last week when LLama3 came out, I was pretty intrigued and started experimenting with that myself.
If you upvote the feature request here, I will reach out and invite you to an experimental version of Langtail supporting more models in the coming weeks: https://lngt.io/mlVS7
Hello @good_lly, thanks for the great comment. We are preparing:
โถ๏ธ Support for multiple LLM providers
โถ๏ธ Focus on prompt evaluations
โถ๏ธ Support for OpenAI assistants
We are already internally testing Llama3 and will launch it soon.
The team behind this is incredible and has great passion for making the UX be best-in-class. They are also extremely focused on solving real problems by providing the tools for people to ship things with confidence. Also, they are laser focused not just on the easy parts, but also on the long-term evolution of your applications.
I had the chance to see an early version and it was already incredible back then and they only improved it over time.
Congratulations on the launch of Langtail Public Beta, Petr and team! It's exciting to see a platform that tackles the challenges of developing AI-powered apps head-on. The features like Playground, Tests, Deployments, and Observability seem like game-changers for teams working on AI projects.
As the landscape of AI models constantly evolves, how do you plan to stay ahead and keep Langtail up-to-date with the latest advancements in the field?
LLM products are creating a flurry of bad experiences in their rush to hit the market quickly. But Petr and his team have been demonstrating since day one just how serious they are about doing this job with outstanding designs. I've been following them for over a year now and I highly recommend them to everyone. I'm certain they're going to reach fantastic places.
Langtail