Show HN: Agent Historic Philosophical Persona Routing and PromptsI've been building this for a while. The core of it is I've assigned different tasks in a software engineering job to different philosophers. The prompts are written toward their thinking by "them" but the keywords are generic. I absolutely love using this prompt system. I share it every chance I get. I wrote a whitepaper about it. Maybe ya'll would be interested in it. I find my experience is better with my prompts than without.The only thing you might not like is my heavy handed treatment of logging output. I was seriously fed up with LLMs trying to run huge tests or scripts and then looking at just the last 20 characters. Everything is logged out to files and then the files are queried for the answers it wants.
Anthropic's pricing wall is routing enterprise revenue to OpenAII've spent the last week trying to use Claude Opus in an IDE via a localhost proxy that wraps the Claude Code CLI.The proxy exists because there's no way to get programmatic API access to Claude on a Pro ($20/mo) or Max ($200/mo) subscription.This isn't an oversight. Anthropic has actively taken steps to prevent subscription-based API access. They've shut down third-party bridges, tightened terms of service, and made it clear this is corporate policy, not neglect, not a missing feature, but a deliberate business decision.Meanwhile, OpenAI opened API access to ChatGPT Pro/Plus subscribers. The result is predictable.A CTO friend is prototyping an electronic warfare detection system. His existing customers include a major European MNO that operates across dozens of countries. He's currently splitting ~$80/month across four AI subscriptions, constantly rationing tokens, juggling providers. He'd switch to Claude Max tomorrow if it included IDE API access. Instead, he's moving his entire pro
What Is A Routing Number? Where To Find YoursTaylor Tepper covered banking, investing and pretty much everything else in personal finance for more than a decade, with his work appearing in the New York Times, Fortune and MONEY magazine, as well ...
IMAGEAsk HN: HotPin – lossless 120B MoE inference on 24GB RAM (CPU, 50 loc)I'm a mechatronics designer with a background in control systems, robotics, PCB design, and embedded hardware. I design physical systems: motors, sensors, microcontrollers, and real-time control loops.I applied this design thinking to LLM memory management – and it worked.HotPin is a set of patches for llama.cpp that runs 30B–120B Mixture of Experts (MoE) models on far less RAM than their disk footprint, with bit-identical (lossless) output.Tested on an AMD Ryzen AI 9 HX 370 (Zen5, AVX512), 23.6GB LPDDR5X, NVMe >1GB/s, CPU-only.Results: | Model | Disk | Min RAM | Savings | tok/s | |-------|------|---------|---------|-------| | gpt-oss:120b | 58.5GB | 19.1GB | -67% | 3.84 | | qwen3:30b-a3b | 18.0GB | 10.4GB | -42% | 19.7 | | gemma4:26b-a4b | 16.2GB | 10.6GB | -35% | 11.5 | | GLM-4.7-Flash | 19.0GB | 13.3GB | -30% | 12.4 |Output is SHA-256 bit-identical to full-RAM runs. Verified.How it works (~50 lines of C++ in llama.cpp): 1. Profile MoE expert routing frequencies. 2. mmap the entire m
IMAGEAalyria hits $1.3 billion valuation after raising funds for satellite mesh networkAalyria announced a $100 million funding round Feb. 23 that values the Californian venture at $1.3 billion, supporting deployment of laser terminals and software for dynamically routing data across space, air and ground networks. The post Aalyria hits $1.3 billion valuation after raising funds for satellite mesh network appeared first on SpaceNews.
IMAGEShow HN: I built managed OpenClaw hosting with 60s provisioning in 6 daysHey HN,I'm Daniel, solo dev from Germany. I built ClawHosters (https://clawhosters.com), a managed hosting platform for OpenClaw, the open-source AI agent framework.Quick timeline: domain registered February 5th. First paying customer six days later. I probably should have spent more time on it, but it works.If you haven't seen OpenClaw, it lets you run a personal AI assistant that connects to Telegram, Discord, Slack, and WhatsApp. Self-hosting it is absolutely possible, but it's a pain. You're dealing with Docker setup, SSL certs, port forwarding, security hardening, keeping the image updated. Most people don't want to deal with any of that. They just want the thing running.That's what ClawHosters does. You pick a tier (EUR 19-59/mo), click create, and you've got a running instance with a subdomain. About 60 seconds if we have prewarmed capacity, maybe 90 seconds from a cold snapshot.Some technical details that might interest this crowd:*Subdomain routing chain.* Every instance gets
IMAGEShow HN: Synapse – Multi-model AI combining LLMs and humans for marketing outputHey HN,I’m Zack, CEO at Averi AI, and we just released Synapse, a modular AI architecture we built to solve a problem we kept running into within the marketing ecosystem:“How do you get domain-specific intelligence without trying to recreate GPT-4 from scratch?”The ProblemMost domain-specific AI tools (marketing, legal, ops, etc.) tend to fall into one of three camps: Use GPT-4/Claude as-is and rely on prompt engineeringTrain a small model from scratch but lose general reasoningGo full frontier model… and burn millions tryingWe’ve considered all three. None hit the mark.Our Approach: Multi-Model + Human RoutingSynapse is our attempt at something better: A routing architecture that matches tasks with the best resource whether that’s an LLM, a smaller domain model, or a vetted human expertA way to balance specialization and scale, instead of choosing oneIt powers our own domain-specific foundation model (AGM-2), and integrates GPT-4, Claude, and others alongside it. Tasks get routed base
IMAGEDIP Spotlights Issue #2SFNP-OPS DEMO 1 DIP Hosts Workshop with Collins Aerospace at NASA/FAA North Texas Research Station (NTX), Fort Worth November 29-30, 2023 The DIP team hosted Collins Aerospace at NASA’s NTX facility to discuss potential collaboration opportunities on the SFNP-Ops demo series. Collins Aerospace were briefed on the current SFNP-Ops 1 Collaborative Digital Departure Rerouting (CDDR) […]
IMAGEDIP Spotlights #4CURRENT DIP ACTIVITIES DIP Team Presented to the New NATCA Rep and Houston District Manager in Houston Texas May 14, 2024 The Digital Information Platform (DIP) team conducted a demonstration of the Sustainable Flight National Partnership Operations (SFNP-Ops) predeparture rerouting capability for the new National Air Traffic Controllers Association (NATCA) Representative and the district manager […]
IMAGEDIP Spotlights #5DIP ACTIVITIES Operational Requirements Finalized for the SFNP-Ops Evaluation in Houston, Texas August 21, 2024 In a major milestone, the DIP Team and field demo partners agreed to a plan of required capability, use cases, schedule, support and logistics to conduct the Sustainable Flight National Partnership Operation (SFNP-Ops) predeparture rerouting operational evaluation scheduled to start […]
IMAGECognitive Space claims two SDA awardsCognitive Space will work with the Space Development Agency to optimize missile tracking and network routing. The post Cognitive Space claims two SDA awards appeared first on SpaceNews.
IMAGEShow HN: Laravel 11 with Servbay – Quick Start GuideIntroduction Laravel 11 has been released, bringing many improvements. It introduces a streamlined application structure, per-second rate limiting, health routing, and more. In addition, Laravel Reverb, a first-party, scalable WebSocket server has been introduced to provide robust real-time capabilities to your applications. In today's guide, I will setup an example laravel application, and install ServBay for local development on Mac. NOTE: ServBay is your one-stop-shop for local PHP development on Mac. It bundles together everything you need like web servers, databases, programming languages, mail servers and queue services. Create Example Application Let's get started generating an example applicaiton using composer. composer create-project laravel/laravel example-app This will create an example-app directory, cd to it. cd example-app Now you can start the server:php artisan serve Go to http://127.0.0.1:8000 and you'll see the default Laravel welcome page.Install ServBay We'll use S
IMAGERussian spacewalkers begin outfitting new lab moduleTwo Russian spacewalkers connected a new laboratory module to the International Space Station’s power grid Friday, routing and plugging in eight cables to tap into electricity generated by NASA’s solar power system.
IMAGEWatch: Spacewalk to complete outfitting of European experiment platformTwo astronauts ventured outside the International Space Station Wednesday to complete work to outfit an external European experiment platform, routing cables and installing a high-speed data relay antenna.
IMAGEBalky connectors prevent full activation of European experiment platformTwo astronauts ventured outside the International Space Station Wednesday to complete work to outfit an external European experiment platform, routing cables and installing a high-speed data relay antenna.
Show HN: We built open OpenRouter that turns usage into a better modelHi HN, we built an open source model gateway. It's a single place to manage our own self hosted, frontier, and open source models in one place.It’s is rust native, built for concurrency, and implements all the config quirks across models and providers (streaming formats, tool calls, model parameters, rate limits, and different error behavior).The gateway adds under 1 ms for BYOK requests and under 2 ms when Experiential supplies the provider key. It has every major inference provider, and 1000+ models refreshed daily via a codex agent that opens a PR.Compared to other similar projects we’re open source, take no markup, allow you to mix local models with a marketplace, and use your traffic to (opt in) train you a model. Simple routing doesn’t warrant a 10% token markup.The way we do this is given standardized OTel traces, we mine representative real tasks, use text world models to simulate rollouts for various models, apply an LLM judge, and fit a nearest neighbor classifier on top of a
Show HN: Use GLM-5.3 in Cursor today via tokengo APIHi HN,Hearing a lot of buzz around GLM-5.3, which I expect to be the best open-source coding model with the weights dropping soon, I wanted to test it where I actually do my work.I just mapped GLM-5.3 to TokenGo so I could swap out the base URL and use it directly in my IDE.The backend handles the edge routing via Cloudflare to keep latency low, critical when you're waiting for inline code completions.If you want to try GLM-5.3 (or glm-5.3-flash) in your own workflow, here is the setup.For Cursor:Go to Settings > Models.Add your API key in the OpenAI API Key field.Toggle "Override OpenAI Base URL" and set it to: https://api.tokengo.com/v1Under "Model Names", add z-ai/glm-5.3.Select it in your chat panel dropdown.We're one of the first inference providers to get the GLM 5.3 family online and fully accessible for IDEs. If you'd like to try it in a high volume production workload I'd be happy to provide test keys for yall