Загрузка видео...

Не удалось загрузить видео

На главную

run agent harnesses 100% private & offline. (no token costs, no API keys, 100% open-source) your agent runs locally. the model doesn't. every prompt, every file, and every secret still leaves your machine before the agent does anything with it. Magnitude fixes that. it's an open source inference server...

56,499 просмотров • 2 месяцев назад •via X (Twitter)

Комментарии: 16

Фото профиля Hussain Hashim | Building SundayBack
Hussain Hashim | Building SundayBack2 месяцев назад

@akshay_pachaar local processing is a game-changer. tackled similar issues building my own thing and it was a real headache. good move!

Фото профиля nomad.carpenter
nomad.carpenter2 месяцев назад

Nice - the part everyone skips is that bandwidth sets your token rate, not your GPU name. Most people buy silicon and then wonder why the tokens still crawl.

Фото профиля Greg Glazewski
Greg Glazewski2 месяцев назад

local model, remote prompt. thats the whole privacy pitch falling apart right there

Фото профиля Gregor
Gregor2 месяцев назад

yeah and with tool-use loops it compounds. each tool call result gets appended and re-sent, so by step 5 you've pushed the same files to the API 3-4 times over without realizing it.

Фото профиля 724
7242 месяцев назад

The memory bandwidth part is the bit most people miss. Everyone shops by VRAM then wonders why tokens crawl. Handing back a real config instead of a model list is the useful move here.

Фото профиля Chaminda
Chaminda2 месяцев назад

local model solves the data-leaving-the-machine problem. it doesn't touch the other one: the agent still has your shell and your files. a private model with unbounded write access is still an agent with unbounded write access.

Фото профиля blackbox
blackbox2 месяцев назад

Naming a config as model + compression + context size + expected speed range is the right unit — a bare model list says nothing about whether a long agent loop will finish. Local models still lose on big multi-file rewrites, so harness choice matters more than model name.

Фото профиля AI Mastery Guide
AI Mastery Guide2 месяцев назад

Fully offline is a big deal

Фото профиля Aleksandar Janca
Aleksandar Janca2 месяцев назад

half the pain with local models is getting the config right for the hardware

Фото профиля 安叫兽|Bird🕊️ 🔶 BNB
安叫兽|Bird🕊️ 🔶 BNB2 месяцев назад

离线是舒服了,显存压力也得自己扛了。

Фото профиля shipsmall
shipsmall2 месяцев назад

the secrets half is the one people underrate. even with a cloud model, my agents never see .env: a script hands out one key at a time, touch id for the important ones, every access logged. local inference fixes the model half. the tooling half needs the same discipline.

Фото профиля David Arnal
David Arnal2 месяцев назад

The interesting part is making “offline” operationally verifiable, not just local inference. A useful next step would be a built-in network-egress audit that proves the harness stays air-gapped in practice.

Фото профиля Ryan Edkins
Ryan Edkins2 месяцев назад

The model has to run locally too, not just the harness, or your prompts and files never really stay private.

Фото профиля George Rowan
George Rowan2 месяцев назад

Real privacy starts when you prompt your own model

Фото профиля Bally_AgenticAI
Bally_AgenticAI2 месяцев назад

Running an agent harness fully local keeps every prompt, file, and secret off the wire before a model ever sees it - most 'private AI' setups still leak that boundary. #AgenticAI #AITools @AnthropicAI

Фото профиля Lee Garcia
Lee Garcia2 месяцев назад

Amazing

Похожие видео

Harness Engineering 2.0 is here! We benchmark models all the time but the harness running the model can dramatically improve the result, cost, and execution time too. And finally, devs just open-sourced Harness Arena, where you can test the same task across multiple harness and model combinations side by side. You provide one task and then run it across several model x harness configs that you want to test. Every config receives the same prompt, files, tools, skills, and evaluation checks. Harness Arena then returns: - total cost - execution time - whether the task passed - files and artifacts produced If you want to try it yourself, it is implemented within the open-source HarnessRouter project. Here's the repo: (don't forget to star it ⭐ ) HarnessRouter is a plug-and-play infrastructure layer that lets you run any harness under a single interface, like: - Codex - Hermes - Claude code - DeepSeek Harness - System One, powered by Jev - And 25 more agent harnesses In Harness Arena, the cheapest run does not automatically win. A config must complete the task correctly first. Cost and execution time are compared only among the configurations that pass. Here's the result from one Harness Arena test: - Hermes + GPT-5.2: 5/5 successful runs using 0.47 credits - Claude Code + Opus 4.8: 3/5 successful runs using 223 credits That is around 474x difference in credits, while the more expensive configuration also failed two runs. You can keep the model fixed to measure how much the harness changes the result. Or vary both to find the best harness and model pairing for that task. This helps answer practical questions such as: - Should this task run on Codex or Claude? - Can a cheaper config still pass every check? - Which setup is best to handle this workload? Once you find the best config, you can run future tasks through it using the same Unified Harness Protocol (UHP) interface for sessions, streaming, files, artifacts, cancellation, and failures. So you are no longer choosing an agent stack based only on a model leaderboard. Instead, you can benchmark the complete runtime on the task your product actually needs to solve. To dive deeper, my co-founder wrote an article explaining why model routing is different from harness routing, how UHP works, and what supporting multiple agent runtimes involves. It also covers the local setup, a working request, and how sessions and files move through the system. Read it below.

Avi Chawla

32,761 просмотров • 13 дней назад

How to Build a Local AI System for Free (Then Scale Up) 👇 Here's the fix: - Install Ollama. It's free and it's one command. - Pull a model like Llama 3.2. Any laptop with 8GB of RAM can run it. - Add Open WebUI, and you've got your own private ChatGPT where nothing ever leaves your machine. - Built on OpenAI's API? Change one line to point it at your own computer, and every request is free. Test it on your real work for one weekend. If it handles your emails and docs, cancel what you don't need, keep one plan for the really hard stuff, and you're still saving over a thousand bucks a year. One important note so nobody gets the wrong idea: the 8GB laptop setup in this video is Rung 1, the free starting point to prove the workflow, not the ceiling. An 8GB machine runs a 7B model, which is enough for everyday writing, docs, and emails. Bigger jobs need more memory, and local AI is really a memory ladder: 24GB runs 32B models, and 128GB unified-memory boxes run 70B models at conversational speed. The point of starting here is to test it for free before spending a dollar, then buy one rung up from your real needs. Frontier reasoning and cutting-edge coding still belong to the cloud, so keep one subscription for that hard 20% and run the other 80% at home. Comment "AI" and I'll send you the full blueprint. #techtok #localai#claude #automatedsystem #automation #claudeautomation #systemdesign #system #restructureai

ReStructure AI

17,400 просмотров • 1 месяц назад

Hermes agent just left the terminal. 𝗛𝗲𝗿𝗺𝗲𝘀 𝗗𝗲𝘀𝗸𝘁𝗼𝗽 dropped yesterday. native app for macOS, Windows, and Linux. for months Hermes was the agent that learned your projects, wrote its own skills, and built a model of who you are. all of it buried in terminal logs. now it has a window. the important part is that it's not a wrapper. it runs the same agent core, the same sessions, memory, and skills as the CLI. you can start a task in the terminal and finish it in the app without anything resetting. the state is shared across every interface, not copied between them. what the GUI actually adds: → streaming chat that shows live tool calls and inline reasoning instead of a spinner → a preview rail that renders pages, code, and images right beside the conversation → an artifacts panel that collects every file the agent has ever produced → remote gateway mode, so you can point the app at a VPS and run the heavy work elsewhere → skills, cron, profiles, and gateways managed point-and-click instead of through YAML → voice mode, drag-drop files, and inline image generation remote gateway mode is the one worth slowing down on. the agent runs 24/7 on a $5 server while you control it from your laptop like a local app. other agent UIs are chatboxes with a logo. this one shows the autonomy instead of hiding it, so you watch the skills load, the tools fire, and the artifacts pile up as it works. it was teased in Jensen's GTC keynote. MIT licensed, local-first, no telemetry. if you already run Hermes, download it and everything is already there. your chats, memory, and skills carry straight over. i wrote a full masterclass on Hermes Agent that walks through the SOUL. md identity layer, the three-tier memory system, the self-evolving skills loop, and how to run three specialized agents 24/7. desktop is the interface that finally does all of it justice. the article is quoted below.

Akshay 🚀

51,540 просмотров • 5 месяцев назад

THIS GUY CONNECTED HIS AI AGENTS TO HIS OBSIDIAN AND BUILT A BRAIN THAT LEARNS ON ITS OWN. HERE'S HOW TO BUILD IT Obsidian is just markdown files sitting in a folder. That turns out to be the perfect memory for an AI agent, because an agent can read and write those files directly. He wired his agents into the vault so they pull context from it, do the work, and write what they learned back. The notes aren't the point. The loop is, and it gets sharper every cycle How to build it: 1. Point an agent at your vault. The fastest way, no plugins, no API keys: open a terminal and run npx obsidian-mcp /path/to/your/vault. That exposes your Obsidian folder to Claude as a tool it can read, search, and write to. Add it to your Claude Code or Cowork config and restart 2. Confirm it can see the brain. Ask it: "list the notes in my vault and summarize what's in them." If it reads them back, the connection is live. Now it starts every task with everything the vault already holds instead of from zero 3. Give each agent one job and a write-back rule. Tell it: "research this, then save what you found as a new note in /brain with links to related notes." One agent researches, one summarizes, one plans. Each writes its output back into the vault 4. Close the loop. Add one line to every agent's instructions: "read /brain before starting, write your result back when done." Now each task leaves the vault richer, and the next run reads that before it works. It compounds instead of resetting 5. You only steer. Review what the brain produces, point it at the next thing. The agents handle the reading, writing, and connecting The edge isn't better notes. It's a brain that feeds itself, so the work gets sharper every cycle instead of starting over Bookmark this

Yarchi

58,730 просмотров • 5 месяцев назад

How to create AI employees — start to finish, everything included The complete build of an AI employee, start to finish — the whole path in under four minutes. The agent underneath — every AI employee starts as an agent: a brain (the AI model), tools (real connections into your email, calendar, spreadsheets so it acts instead of only talking), and a loop (it works the job, checks itself, and keeps going until it's done). Feed the brain context and it works on YOUR business. The four upgrades that make it an employee — memory (corrections get written down and read back before every job — explain a thing once and it stays explained), permanent context (your world lives with it and changes as your business changes), skills (your way of doing a job, taught once, followed every time), and a schedule (an agent runs when you call it — an employee clocks in on its own, seven in the morning, inbox already sorted). The build tool — an AI coding assistant: Claude Code or Codex. That class of tool IS the agent machinery — it navigates your context, picks its next move, loops until the job is done, and carries memory between runs. You set it up in plain English, and when a piece you need exists nowhere, it writes that piece itself. The full assembly — pick one job, give it context, wire the tools, teach the skill, add memory, set the schedule. One rule built in: high-stakes moves wait for your yes; everything else runs around the clock. The ceiling — one employee becomes a pattern, patterns become a team, and the team becomes a department passing work to each other. This video is the skeleton of my entire curriculum — every step is a full course, taught from zero, step by step, about four months front to back. It's all inside my community. #aiemployee #aiagent #aiautomation

Professor Glitch

12,600 просмотров • 2 месяцев назад

Finally, an open-source runtime security layer for your agent harnesses. AI agents can call tools, run shell commands, modify files, and interact with production systems. But when something goes wrong, teams often have to reconstruct what the agent actually did from scattered logs after the fact. Agent Beacon is built to change that. It runs locally and records agent activity as it happens, including: → Tool calls → Shell commands → File changes → Approval decisions → Session and harness context The useful part is that it normalizes all of this into the same event format across 23+ agent harnesses. So instead of writing different detection logic for Claude Code, Codex, or another harness, security teams can reason about the underlying action itself. A command executed is a command executed, regardless of which harness produced it. Beacon also records how confidently an event was captured. An action can be directly observed from the runtime or inferred from indirect evidence. That matters when you start writing security rules against this data. For example, you can detect when an agent reads an environment variable, modifies a sensitive file, executes a suspicious command, or performs an action without the expected approval. And the detection happens while the session is still unfolding, not days later during incident reconstruction. Everything runs locally by default. You can inspect sessions through the dashboard, write your own detection rules, or forward the same normalized events to tools like Splunk, Datadog, Elastic, Sentinel, or CrowdStrike. Beacon GitHub: (don't forget to star 🌟) I also wrote a detailed article on the same, covering how important runtime security is and how to build it for your agent harnesses. The article is quoted below.

Akshay 🚀

27,859 просмотров • 2 месяцев назад