The latest articles on DEV Community by Haichi Ops (@haichi_ops_861a9ec315b1e7). https://dev.to/haichi_ops_861a9ec315b1e7 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4023033%2F08536751-2e03-4e22-85d8-76f95cd8578f.png https://dev.to/haichi_ops_861a9ec315b1e7 en Haichi Ops Mon, 13 Jul 2026 18:55:41 +0000 https://dev.to/haichi_ops_861a9ec315b1e7/i-got-tired-of-juggling-ollama-tabs-terminals-and-agent-chats-glj https://dev.to/haichi_ops_861a9ec315b1e7/i-got-tired-of-juggling-ollama-tabs-terminals-and-agent-chats-glj <p>Running models locally is the easy part. Keeping multiple agents, prompts, plans, and project context organized is where my setup kept falling apart.</p> <p>I would have Ollama running in one terminal, separate chats in different windows, notes in another file, and no clear answer to a basic question: <strong>which agent is doing what, with which model, inside which workspace?</strong></p> <p>So I built HAICHI, a self-hosted control room for local AI agent workflows on Windows and Linux.</p> <h2> The workflow I wanted </h2> <p>My minimum useful loop was:</p> <ol> <li>Detect the Ollama models already installed on the machine.</li> <li>Give two agents different roles and model assignments.</li> <li>Run a plan, discussion, or sequential pipeline without rebuilding the context by hand.</li> <li>Keep plans, messages, and results in a local SQLite database.</li> <li>Allow file access only for explicitly selected models and workspace folders.</li> </ol> <p>HAICHI v1.1 now does that from one local dashboard.</p> <h2> What the free build actually includes </h2> <p>The Personal build is free and does not require activation. It includes:</p> <ul> <li>Windows x64 and Linux x64 packages</li> <li>local Ollama model discovery</li> <li>up to two local agents</li> <li>one active workflow at a time</li> <li>plans, discussions, and pipelines</li> <li>persistent local state</li> <li>optional, workspace-scoped file permissions</li> </ul> <p>The limits are deliberate. I want the first run to answer one question: <strong>is coordinating two local agents in one workspace more useful than juggling separate chats?</strong></p> <p>Developer Pro is a separate licensed build for cloud providers, unlimited agents and workflows, and private Tailscale access. You do not need Pro to test the local workflow.</p> <h2> Honest limitations </h2> <p>This is an early desktop product, not a hosted team SaaS.</p> <ul> <li>You need Ollama installed for local models.</li> <li>Tool reliability depends on the selected model following structured tool instructions.</li> <li>Smaller models can be useful for chat and planning but may ignore file-tool formats.</li> <li>File and command permissions are off by default because local automation should not receive broad access silently.</li> </ul> <h2> I am looking for three real first workflows </h2> <p>I do not need more generic launch impressions. I need three people who already use Ollama to try one real task and tell me exactly where they stop.</p> <p>Try HAICHI Personal free: <a href="https://haichi.app/?utm_source=devto&amp;utm_medium=organic&amp;utm_campaign=first_three_users" rel="noopener noreferrer">haichi.app</a></p> <p>If download, installation, model discovery, or the first workflow blocks you, use the <strong>Tell us what blocked you</strong> link on the download section. Short and critical feedback is more useful than a polite review.</p> <p>What is the first two-agent local workflow you would try: parallel research, code review, planning, or something else?</p> ai productivity opensource