Dosu

I love the idea of "just-in-time" documentation - a how-to guide, onboarding guide, or cookbook generate for your exact use-case whenever you need it. It's the dream devX for any software library or service. This recent example of Dosu on Langfuse (YC W23) highlighted the potential of "just-in-time" docs: https://lnkd.in/gx6qyWuc A user wanted to know how to integrate Langfuse with Gradio, but there isn't any documentation for their use-case, so they created a new discussion on GitHub. Within minutes of posting the user and Dosu collaborated to generate a near-perfect cookbook for the Gradio x Langfuse use-case. The reaction of the user sums it up perfectly: "I love this tool even more!!" Overtime, I think users will expect "just-in-time" documentation to be the default documentation experience for integrations, how-to guides, and other use-case specific docs in the future. 

Here is one more alternate view of modern AI stack. There are apps like amazon.com, hotels.com etc which exists today which most likely will become transaction rails backbone in future. The new world of agents will use these transaction rails for task completion. An agent at the top will use an orchestration layer to route between planning of the tasks (using model layer) and task completion (tooling layer). Tooling layer will provide agent-to-app pathway for transaction completion. Following companies are interesting backbone of this stack > Agentic layer : Decagon, Replit, Harvey, Dosu, Perplexity > Orchestration layer: LlamaIndex, LangChain, CrewAI, phidata, Letta > Tooling layer: Plotch.ai, Composio, Browserbase > Genai layer: Copy.ai, Synthesia, Runway > Model layer: Sarvam, OpenAI > GPU Layer: NVIDIA > CPU layer: Intel Corporation How will this modern AI stack unfold in 2025? Lets see

Here is one more alternate view of modern AI stack. There are apps like amazon.com, hotels.com etc which exists today which most likely will become transaction rails backbone in future. The new world of agents will use these transaction rails for task completion. An agent at the top will use an orchestration layer to route between planning of the tasks (using model layer) and task completion (tooling layer). Tooling layer will provide agent-to-app pathway for transaction completion. Following companies are interesting backbone of this stack > Agentic layer : Decagon, Replit, Harvey, Dosu, Perplexity > Orchestration layer: LlamaIndex, LangChain, CrewAI, phidata, Letta > Tooling layer: Plotch.ai, Composio, Browserbase > Genai layer: Copy.ai, Synthesia, Runway > Model layer: Sarvam, OpenAI > GPU Layer: NVIDIA > CPU layer: Intel Corporation How will this modern AI stack unfold in 2025? Lets see

We are a proud partner of this year's virtual #AINativeDevCon. The team Tessl has put together an incredible lineup of world-leading experts in AI software development. This is not one to miss. We hope to see you there! http://tessl.co/3U7zyXa

Just got to #KubeCon! If you're here, use Dosu, or want to chat OSS maintenance, let me know

🎆 Last night we hosted our first Bay Area meetup "Composable AI Infra for Agents". We had some amazing talks from Devin Stein, Chang She and Akela Drissner-Schmid. Thank you all for coming, we had a really great time meeting both old and new community members! Big shout out to LanceDB, Continue and Dosu for collaborating with us on this. To many more such meetups! 🥂 Join our next community meetup in Paris, France: https://lu.ma/gsf3mjbz

Happening tonight! Come for technical happy hour with dltHub, LanceDB and Dosu on Nov 7th. In person only. Door opens at 5:30PM, with pizza and drink to get your started. Register for the event: https://lu.ma/y3sh3tpj

Devin Stein is the CEO and Founder of Dosu. Prior to Dosu, Devin was an early engineer and leader at various startups. Outside of work, he is an active open source contributor and maintainer. Dosu leverages LanceDB's out-of-the-box versioning and time-travel capabilities to help them reproduce, experiment, and evaluate on historic data. Join us on Nov 7th and hear from Devin on how Dosu is 𝗦𝗼𝗹𝘃𝗶𝗻𝗴 𝗔𝗜 𝗖𝘂𝘀𝘁𝗼𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝗖𝗼𝗺𝗽𝗼𝘀𝗮𝗯𝗶𝗹𝗶𝘁𝘆. Register here: https://lu.ma/y3sh3tpj

🎙️ 𝗝𝗼𝗶𝗻 𝗢𝘂𝗿 𝗘𝘅𝗰𝗹𝘂𝘀𝗶𝘃𝗲 𝗠𝗲𝗲𝘁𝘂𝗽: 𝗖𝗼𝗺𝗽𝗼𝘀𝗮𝗯𝗹𝗲 𝗔𝗜 𝗜𝗻𝗳𝗿𝗮 𝗳𝗼𝗿 𝗔𝗴𝗲𝗻𝘁𝘀! We're excited to invite you to delves into the core AI infrastructure that powers AI Agents. Join us for an insightful and interactive session that will equip you with the knowledge and tools to build robust and adaptable AI agents Register for the event: https://lu.ma/y3sh3tpj This is a technical happy hour hosted by dltHub and LanceDB. Thank you for our friends at Continue and Dosu for the support! See you Nov 7th!

"With this kind of thing, you don’t necessarily expect a rep to go above and beyond, but Jester has done that at every point." Mark Steinbrick is an engineer at Dosu, a company that recently raised $8.5M to automate engineering knowledge work. Dosu helps software teams triage complex technical issues and automatically keep documentation up to date. Dosu's GitHub app has been installed on over 25,000 repositories and is used by thousands of engineers every day. A few months ago, Mark turned to Fewertasks to help expand to a segment that has been rapidly adopting Dosu: open source software (OSS). Mark needed a rep to help his team sift through thousands of open source projects, identify the highest priority projects, find the maintainers, and conduct outreach to those individuals. Prior to hiring FewerTasks, Dosu was reaching out to ~10 contacts per WEEK. Today, that number stands at 20+ contacts per DAY. I sat down with Mark and asked him about his experience with FewerTasks including what has surprised him and advice he'd have for founders considering hiring an offshore team member.