Article URL: https://life-after-ssri.bearblog.dev/dear-people-who-work-at-the-airport/ Comments URL: https://news.ycombinator.com/item?id=49297801 Points: 81 # Comments: 50
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
A developer is publicly building Vaxyro, an AI-powered WhatsApp and CRM platform designed specifically for real estate sales teams. The tool aims to address lead management failures — not lead generation — by automatically capturing and structuring prospect details such as budget, location, and property type from WhatsApp conversations. The AI handles initial qualification and repetitive inquiries, while flagging complex discussions, like payment schedules, for human salespeople to take over. One of the core technical challenges is converting unstructured chat messages into structured CRM data while maintaining conversation context across multiple messages. The builder is developing Vaxyro openly and is actively consulting real estate professionals to ensure the product solves actual workflow problems before adding new features.
A solo developer has built NexusMem, an open-source tool designed to give coding agents richer context by indexing git logs, shell history, and the reasoning behind decisions that never made it into commits. The tool uses a hybrid retrieval approach combining BM25 over SQLite FTS5 and vector search via sqlite-vec, with scores adjusted for signal strength and recency. The developer identified a ranking flaw through real-world testing on the tool's own repository, where the most relevant commit ranked lower than unrelated ones, and corrected the scoring formula based on empirical results. NexusMem requires only Node 22 and git to run, with Ollama as an optional dependency that enables semantic search. Within weeks of sharing the project, two independent contributors submitted fixes unprompted, which the developer described as a sign the project had become legible enough for others to navigate without guidance.
A developer has documented the first stage of building a customizable AI-powered PDF reader in Python, beginning with a Jupyter prototype rather than a full desktop application. The initial build uses PyMuPDF for rendering and text extraction, ipywidgets for the interactive interface, and Pillow for image handling, with JSON storing bookmarks and notes locally. Key features implemented in this milestone include PDF navigation, zoom, text search with highlight positioning, and persistent bookmarks and notes. The developer used a centralized application state object to keep UI controls consistent and avoid fragmented logic across buttons. The long-term roadmap includes a PySide6 desktop UI, LLM-powered summaries and question answering, text-to-speech, and voice commands.
A developer has migrated a PDF reader project from a Jupyter notebook prototype to a fully functional desktop application built with PySide6. The app supports features including page navigation, zoom controls, text search with highlights, and dockable panels for bookmarks and notes. A key design decision keeps all annotations, bookmarks, and reading progress stored in local JSON files rather than modifying the original PDF. The codebase is structured into separate modules for the UI, PDF rendering, reader state, and storage, making it easier to swap components independently. The project is not yet AI-powered, but these core reading features are being established before a large language model layer is added.
Discussion (0)
Log in to join the discussion and vote.
Log in