Carbon

Glimpse the future -- a new generation of founders in Seattle. The map below, created by YoungTech Seattle in collaboration with Peter Mueller of Breakwater Ventures, represents a look at the incredibly diverse set of Seattle companies with a founder or co-founder under 35. Despite Seattle's entrepreneurial and funding ecosystems disproportionately favoring those with decades of industry experience, these founders are beating the odds. This small but growing cohort packs a serious punch -- with an average founding year of 2021, they have collectively raised more than $225M in venture funding. With the best undoubtedly yet to come, this is certainly a group to keep your eye on. If we missed you on this list we apologize and we would love to learn more about you and what you're working on so please reach out! --- Note: This does not include the many young founders still building in stealth - can't wait for your public launches!

Over a single weekend, Mai from MindPal Labs integrated Carbon to provide more knowledge sources for their customers. 🧠 Now their "virtual taskforce" of AI agents can assist users with a wide variety of tasks by learning from the user's knowledge and data from Google Drive, SharePoint, Outlook, etc. Read the case study with MindPal here 👉 https://lnkd.in/gy3t-t5s

Carbon now natively supports PongoAI's next-gen reranking model! Excited to be collaborating with a fellow Seattle founder on this one (Caleb John)! 🤝

Carbon now supports self-hosted deployments to Amazon Web Services (AWS) GovCloud. 🇺🇸 🦅 This unlocks our ability to work with clients who need to host sensitive data and regulated workloads in the cloud. We aim to assist customers in meeting their US government compliance requirements, including ITAR and FedRAMP High baseline requirements.

I'm thrilled to announce that Carbon has launched our Least Recently Used (LRU) Cache for file embeddings! 🔥❄   This feature allows customers to reduce storage costs without compromising performance. It works by storing embeddings for frequently accessed files in hot storage and automatically moving less frequently used files to cold storage. This feature is now available for our in-house and third-party vector storage databases.

Excited to announce our connector for DigitalOcean Spaces! 🔌 Carbon will index all the files (text, images, audio, video) in your Spaces buckets and then stream them to your vector store.

628 billion characters processed to date. 📈 Carbon is on track to process 1 trillion characters by the end of the year. 🫡

Excited to announce our connector for Google Cloud Storage! 🔌 Carbon will index all the files (text, images, audio, video) in your GCS bucket and then stream them to your vector store.

If you're building in the AI application layer, these are essentially all the components you need, to produce the most contextual response, in the fastest time, while spending the least amount of $$ in computation cost. Here's a list of companies who provide these : 1. End to end Fine tuning & inference - Pipeshift AI (YC S24) (Arko C) 2. RAG - Carbon (Derek Tu) WitHub ( Rohan Gupta) CogniSwitch Inc. (Dilip Ittyera) 3. Router - Martian Portkey (Rohit Agarwal) 4. Infra Management - Portkey 5. End to end Workflow Automation - Dayzero AI (Yours truly) One thing we realised at Dayzero AI after building a few of these in-house, that it just wasn't required. It was definitely fun, and a lot of learning, but enhh. Until and unless you're selling to devs, you do not need to build the entire infra in-house. Your customers DO NOT CARE. Focus on their workflow, go deep into that and simply keep refining your prompts. While maintaining a clean data set for retraining again :P #rag #ai #startups #infra #saas

Announcing our new integration with AssemblyAI! This provides customers with more options beyond our existing Deepgram integration and enables semantic (and hybrid) search on audio files using AssemblyAI's next-gen speech-to-text models. 🔊 💫 Customers can upload files in 20+ audio formats (mp3, wav, flac, etc.), then index and search through all transcriptions directly via Carbon.