791 results for work force · 2.437s

News for “work force”
20 results • 2399 ms server time
Moozonian News
github.com• May 19, 2026• 1 min read
Show HN: Forge – Guardrails take an 8B model from 53% to 99% on agentic tasksHi HN, I'm Antoine Zambelli, AI Director at Texas Instruments.I built Forge, an open-source reliability layer for self-hosted LLM tool-calling.What it does:- Adds domain-and-tool-agnostic guardrails (retry nudges, step enforcement, error recovery, VRAM-aware context management) to local models running on consumer hardware- Takes an 8B model from ~53% to ~99% on multi-step agentic workflows without changing the model - just the system around it- Ships with an eval harness and interactive dashboard so you can reproduce every numberI wanted to run a handful of always-on agentic systems for my portfolio, didn't want to pay cloud frontier costs, and immediately hit the compounding math problem on local models. 90% per-step accuracy sounds great, but with a 5-step workflow that's a 40% failure rate. No existing framework seemed to address this mechanical reliability issue - they all seemed tailor-made for cloud frontier.Demo video: https://youtu.be/MzRgJoJAXGc (side-by-side: same model, same
Moozonian News
github.com• May 17, 2026• 1 min read
Show HN: Save Context from MCP BloatWhile doing some work recently I didn't notice that a project I didn't touch for a while was using playwright-mcp instead of playwright-cli. I only noticed it because context was getting full unreasonably fast. I've realized that one of the main differences is that agent would pipe CLI results to some filter right away, but this was not possible with MCP, so it would get whole a11y snapshot of the page every time.I've created a small wrapper for MCP servers that checks response sizes and forces agent to use filter on the response if response is large enough. In my tests (https://github.com/healqq/mcp-content-guard/blob/main/bench/...) it managed to save a decent amount of tokens while solving certain tasks (while added a bit of overhead in other tasks).It works with both local and remote MCPs. Would be happy to hear your feedback and if that could be valuable for you!
Moozonian News
velda.io• May 14, 2026• 1 min read
Show HN: Velda – Run jobs with serverless GPUs, without container imagesAlmost all cloud job frameworks rely on containers. But building, pushing, and pulling images is slow and forces you to maintain a separate environment from your local dev setup. The maintenance of images is challenging as well: version drift, custom libraries & integrations, complex manifests.We believe containers create too much overhead during development. We built Velda to let you launch jobs directly in the cloud by mirroring your Velda managed dev-environment, all with just a command prefix. That way, you only need to allocate GPUs when you're running your training jobs, with no change to your workflow.How it works:* No manifests or custom libraries: You don't need to rewrite your code or define YAML. Just prefix your command with vrun.* Zero Restrictions: Use any tool, binary, or library already on your machine.* Instant Launch: We built a custom streaming file system. Instead of waiting for a 5GB image to pull, we stream only the necessary bits to the cloud instance, allowing j
Advertisement
Moozonian News
news.ycombinator.com• May 10, 2026• 1 min read
Beginners don't trust the command lineAfter learning Git in the summer, I went to my some of my friends and classmates who study CS and told them: "I'm going to force you to learn Git". I don't want to work alone!One thing kept happening over and over: whenever I explained how branches work and how we switch branches, and I showed them how files do change in the working directory (using ls command -- in either WSL or git bash), they nodded and thought it made sense. Then I open the file explorer (finder) and show them that the files DO change; they go quiet for a second, "wow" they say. It's as if the command line isn't something concrete.
Advertisement
Moozonian News
recursant.ai• May 6, 2026• 1 min read
Show HN: Recursant – service mesh for governing AI agentsHello,I have just released Recursant to the public. I have been working on it for a while. It is a control plane for governing AI agents across stacks. It provides full observability, guardrails, and control on the network level by routing all traffic through a side car.Problem statement: many large, regulated enterprises (think banks, telcos) have one engineering team on LangGraph, another on CrewAI, marketing on AgentForce, and data teams on Databricks Agent Bricks. They need their agents to talk to each other with consistent policy enforcement, one audit trail, and a single set of guardrails, yet allowing different functions to run on their own stacks. Recursant solves that problem using the service mesh patternRecursant has two components: a registry and the mesh . The registry contains all live agents. The mesh uses sidecars to route traffic and enforce on the network layer.Aim is for Recursant to provide a real-time EU AI Act Annex IV compliance, so it is not generated from stati
Moozonian News
askyumi.app• May 6, 2026• 1 min read
Show HN: Yumi – Your workspace for AI chat, notes, and researchHey HackerNews,The time has come and I finally launched Yumi, my product that I've been working on for the past while.There was a lot of back and forth, tears, stress, hard grind, and really just the thought of will I even get to launch Yumi. Many questions popped up in my head. Would I ever launch? Would I even make a cent out of this? Would I have the brain capacity left after all of this? Would I achieve the goals I want out of building this? How long can I hang onto Yumi before I call it quits? And the list goes on and on.I was about to throw it all away during the pre-launch phase into the corner and into the void. I really didn't have the will and drive to keep on going. I really didn't to be honest. But there was this resistance I couldn't describe easily. The more I wanted to give up, the more force I used to push back against it. It was a miracle that I still stood up after everything. I knew it wasn't my time to let go of Yumi yet. I couldn't leave behind something I put my h
Moozonian News
news.ycombinator.com• May 5, 2026• 1 min read
Show HN: Airbyte Agents – context for agents across multiple data sourcesI’m Michel, co-founder and CEO of Airbyte (https://airbyte.com/). We’ve spent the last six years building data connectors. Today we're launching Airbyte Agents (https://docs.airbyte.com/ai-agents/), a unified data layer for agents to discover information and take action across operational systems.Here’s a quick walkthrough: https://www.youtube.com/watch?v=ZosDytyf1fgAs agents move into real workflows, they need access to more tools (e.g. Slack, Salesforce, Linear). That means a ton of API plumbing: authentication, pagination, filters, handling schema, and matching entities across systems.Most MCPs don’t fix this. They’re thin wrappers over APIs, so agents inherit their weak primitives and still get it wrong most of the time, especially when working across tools.An even deeper issue is that APIs assume you already know what to query (think endpoints, Object IDs, fields), whereas agents usually start one step earlier: they need first to discover what matters before they can even start re
Advertisement
Moozonian News
simdrive.xyz• Apr 29, 2026• 1 min read
Show HN: SimDrive – a browser racing game with your phone as the controller:DGames in general and multiplayer games have been a big part of what my friends do together (not just digital; sports, board + card games)And it moved to online games as more of us moved to other places and party games became our go-toLove jackbox and gaming couch (that I discovered here on HN)I saw the vibej.am for this year and I couldn't think of a good enough idea till about the 3rd week but here I am!This is a game made for me and my friends and it's also very inspired by what I thought the PS3 6-axis controller would be when I first heart about itThanks for reading this and I hope you and your friends enjoy playingYou can play either split screen or on separate screens with the room code or even have two split screens play too (max 8 players for now)It started with F1 cars but working on go-karts, trucks and tuk tuks :DI've tried to include a lot of "simulation" in the game. There's decent physics that include down force, grip etc and it even changes with/without the rainIdeally y