We’re open sourcing the first benchmark for enterprise AI search: EnterpriseRAG-Bench. Retrieval is the foundation of every AI agent that works with company data. 1M token context windows and extended thinking are wasted if the agent can't find the right information across your tools. EnterpriseRAG-Bench evaluates how well an AI agent can navigate hundreds of thousands of Slack messages, contradictory sources, and “official docs” that haven’t been updated in years. The dataset is 500k synthetic documents and 500 questions across 9 sources like Slack, GitHub, and Google Drive. The corpus was created with thematic clusters, jargon, and ambiguities to reflect how real companies operate. We benchmarked major AI platforms, popular open source projects, and enterprise search products. Here are some of our findings: 💡 Agent harness matters just as much as retrieval technique - OpenClaw with a BM25 tool cleared nearly all platforms 💡 Recall tracks quite closely with answer correctness - if you can surface the correct information, today’s LLMs can reliably generate the right answer 💡 Onyx was the only open source product at the top of the benchmark - most RAG projects are built for personal use and could not keep pace at 500k docs This is the most technical work we’ve ever published, and the data has so many insights on how enterprise AI systems perform on company docs. The results, dataset, and our white paper are all available on GitHub: github.com/onyx-dot-app/Ente…

We matched ChatGPTs deep research, but with open weight models. In a blind head-to-head, win-rate was exactly 50%.

One thing teams don’t talk about enough is the importance of good docs. We put 2 weeks into revamping our docs from the ground up and immediately saw a 3x on engagement. Shoutout to @mintlify If you want to see how we did it, check it out here: docs.onyx.app/

We spoke to hundreds of users last month and their message is undeniable - so we’re taking a big bet We’re relaunching @onyx_dot_app as THE entry point to GenAI at work. As part of this big bet, we’ve revamped our website which explains it more clearly: onyx.app/

@tryramp rolled out @onyx_dot_app to keep up with challenges in onboarding and customer support and used GenAI to answer over 200,000 questions over Slack and 1 million queries across all channels.

Cookbook for using Onyx to automatically resolve customer success tickets 🔥 Steps: - Connect up CS tools like Zendesk, Freshdesk, Slack, etc. - Connect up your runbooks and product documentation - Turn on Agent Mode ✨ - Let Onyx cook-up answers to all your customer questions

Just shipped Deep Research across all your team’s applications - powered by LangGraph! We combined @onyx_dot_app’s search capabilities with @langchain agent framework to create a research agent that can explore the knowledge of connected apps (Google Drive, Slack, Jira, etc.)

An especially personal team update: my old manager/mentor has joined @onyx_dot_app ! When Joachim and I met, I was an NLP novice. He took a chance on me and taught me everything I know. I'm so excited that Joachim is once again taking a chance on me - this time as a founder.

OpenAI’s most recent release of o1 is disappointing despite the benchmarks. It’s going to be great for users but it also reveals that the rate of AI progress is slowing down dramatically. It’s just a combination of known techniques from past years.

I'll be speaking at FYSK on Aug 28 and sharing Danswer's progress and why now is a great time to join. Look forward to meeting you in person! Apply to attend here: foundersysk.com Thanks to @nifferkin , @dksf , and @foundersysk for the opportunity!

Just graduated @ycombinator and been swamped with demand! We're back with a 🧨: We built the first Enterprise Agent-Search to find you exactly what you need. Our AI reasons against itself and all of the documents to give you the most accurate view of its findings