https://www.jamieede.com/tags/deepseek/ Recent content in DeepSeek on Jamie Ede Hugo en-us Fri, 22 May 2026 14:00:00 +0000 https://www.jamieede.com/posts/deepseek-instead-of-openai-chat-langflow-rag/ Fri, 22 May 2026 14:00:00 +0000 https://www.jamieede.com/posts/deepseek-instead-of-openai-chat-langflow-rag/ <p>The Langflow DataStax RAG template ships with an OpenAI chat node. For <a href="https://www.jamieede.com/astra-chat" >Astra Docs Chat</a> , I replaced it with <strong>DeepSeek</strong> for streaming answers: cheaper at volume, acceptable on technical Q&amp;A over retrieved docs.</p> <p>Context: <a href="https://www.jamieede.com/posts/building-astra-docs-chat-rag-over-datastax-on-langflow-and-cloudflare-pages/" >Building Astra Docs Chat</a> · <a href="https://www.jamieede.com/posts/langflow-rag-astra-db-ingest-and-chat-flows/" >Langflow chat flow</a> · <a href="https://www.jamieede.com/posts/astra-db-vector-store-datastax-docs-rag/" >Astra DB vector store</a> </p> <p>Embeddings stayed on OpenAI <code>text-embedding-3-small</code>: only the <strong>generation</strong> step changed. See <a href="https://www.jamieede.com/posts/chunking-embedding-technical-docs-rag/" >chunking and embedding post</a> for why those stay paired.</p> <p>Try it: <a href="https://www.jamieede.com/astra-chat" >Astra Docs Chat</a> </p> https://www.jamieede.com/posts/langflow-rag-astra-db-ingest-and-chat-flows/ Wed, 20 May 2026 16:00:00 +0000 https://www.jamieede.com/posts/langflow-rag-astra-db-ingest-and-chat-flows/ <p>The overview of <a href="https://www.jamieede.com/astra-chat" >Astra Docs Chat</a> names Langflow once and moves on: ingest graph, chat graph, done. This post opens the hood: component chains, two published endpoints, where API keys live, and why orchestration stayed in Langflow instead of a Cloudflare Worker.</p> <p>Start here if you missed the big picture: <a href="https://www.jamieede.com/posts/building-astra-docs-chat-rag-over-datastax-on-langflow-and-cloudflare-pages/" >Building Astra Docs Chat</a> </p> <p>Related: <a href="https://www.jamieede.com/posts/batch-ingest-markdown-langflow-api/" >Batch ingest</a> · <a href="https://www.jamieede.com/posts/proxying-langflow-cloudflare-pages-functions/" >Proxy</a> </p> <hr> <h2 id="two-endpoints-one-product" class="anchor-link"><a href="#two-endpoints-one-product">Two endpoints, one product<span class="pilcrow">&nbsp;¶</span></a></h2> <table> <thead> <tr> <th>Endpoint</th> <th>Purpose</th> <th>Called by</th> </tr> </thead> <tbody> <tr> <td><code>datastax-astra-ingest</code></td> <td>File → vectors in Astra DB</td> <td>Local batch ingest script (<a href="https://www.jamieede.com/posts/batch-ingest-markdown-langflow-api/" >batch ingest post</a> )</td> </tr> <tr> <td><code>datastax-astra-chat</code></td> <td>Question → retrieve → LLM answer</td> <td>Pages Function <code>/api/astra-chat</code> (<a href="https://www.jamieede.com/posts/proxying-langflow-cloudflare-pages-functions/" >proxy post</a> )</td> </tr> </tbody> </table> <p>Splitting ingest and chat keeps the public surface minimal. Visitors never trigger file upload or embedding: only the retrieval + generation path.</p>