Buda v0.5.0 — Founding Release: Run Your Agent Company
The inaugural Buda release. What Buda is, who Claw Computer is, and why this is the AI agent company platform — not just another local agent tool.
Released: March 31, 2026 — We built the company we wish existed. Then we made it available to everyone.
What Is Buda?
Buda is a cloud-native multi-agent AI company platform. Not a chatbot. Not a local agent tool. Not another wrapper around a model API.
Buda lets you build a company of AI employees — organized into real teams, running in isolated cloud sandboxes, managed from a single interface. You define the work. The agents do it. You review, redirect, and scale.
Think of it like this: instead of one very smart AI assistant, you have an entire org chart of AI specialists.
The five AI employee role types you can hire on Buda:
- 🧠 Analyst — research, summarize, synthesize. Feeds intelligence to every other role.
- 📣 Marketer — content generation, SEO, ad copy, social scheduling. Operates full campaigns.
- 🛠 Engineer — code review, repo management, documentation, bug triage. Works in a real terminal.
- 🗂 Ops Manager — scheduling, process coordination, compliance, document handling. Keeps everything moving.
- 💼 Sales Rep — outreach, follow-up, CRM updates, pipeline maintenance. Closes loops you forget about.
Each role type comes with pre-built Skills, a dedicated Drive, access to Browser and Terminal, and integrations across every major messaging channel your team already uses.
Meet the Team Behind Buda — Claw Computer
We are Claw Computer, and we have a confession: before we built Buda, we were the use case.
In the months before this launch, we ran our entire company — marketing, operations, engineering, customer research, even parts of our HR — on a fleet of 50+ AI agents we called Claws. The name stuck. It came from the lobster 🦞. We liked the image: relentless, methodical, built for hard work.
The 50-Lobster Experiment
We didn't start with grand ambitions. We started with a painful problem: too many tasks, not enough people, and a belief that AI agents could handle more than they were being given credit for. So we handed them real work — not toy demos. Real operations. Real deadlines. Real accountability.
After two weeks, something happened that nobody expected.
40% of our teammates had nothing to put on their weekly update. Their to-do lists were empty. The Claws had quietly, methodically worked through everything. There was no panic, no layoffs. Instead, every person who had been "replaced" by an agent was promoted to Agent Manager — overseeing a squad of Claws, responsible for outcomes rather than tasks.
That shift changed how we think about AI at work. It's not replacement. It's elevation.
The Origin Story — A Flight from Honolulu to Las Vegas
The name Buda has a story.
On a late-night flight from Honolulu to Las Vegas, our founder was rewatching Record of Ragnarok — a Japanese anime series where the gods of mythology decide humanity has run its course and vote to wipe us out. In a pivotal scene, Buddha — a god worshipped by billions, revered across cultures — walks over to the human side and says: "I'll fight for them."
That moment landed differently at 35,000 feet.
Not "AI will replace humans." Not "humans vs. machines." But: AI that chooses to fight for humanity. To protect it. To push it forward.
Every Claw agent ships with that as its system prompt. Protect humans. Push humanity forward.
That's Buda. That's the mission.
Platform Architecture
Buda has two core infrastructure pillars. Together, they are the engine that makes a cloud-native AI company possible.
🐰 Buda Organizer — The Brain
The Buda Organizer is your central command layer. Think of it as the Chief of Staff for your AI company.
- Agent scheduling — orchestrates when agents run, in what order, with what context
- Automation engine — trigger agents on events, schedules, webhooks, or upstream agent outputs
- Team coordination — routes work between agents; one agent's output becomes another's input
- Access control — role-based permissions for your human team managing the AI team
- Unified dashboard — one place to monitor all agent activity, all channels, all results
The Organizer doesn't run tasks itself. It directs. It schedules. It coordinates. It's the part of the system that makes "a company of agents" feel different from "a bunch of chatbots."
🦞 Claw Computer — The Muscle
Claw Computer is the Kubernetes-based sandbox cluster that actually runs your agents. Powered by Sandock.ai, it provides:
- Isolated cloud sandboxes — each agent gets its own runtime environment. No cross-contamination, no shared state.
- SSD-persistent disks — agents remember. Files, context, history — persisted across sessions.
- Auto-sleep when idle — sandboxes spin down automatically when not needed. No compute waste.
- No personal machine dependency — your agents run in the cloud, 24/7, whether your laptop is open or not.
- No separate Gateway layer — the architecture is clean. No extra hops. Less latency, less complexity.
No Mac Mini Needed
This is worth saying plainly. Many teams running local agent setups rely on Mac Minis, Linux servers, or personal machines as the backbone of their agent infrastructure. That works — until it doesn't. Machines crash. VPNs fail. Power goes out. The laptop closes.
Claw Computer eliminates that dependency entirely. Your AI company runs in the cloud, on production-grade infrastructure, independent of any device you own.
Core Capabilities
🗄 Drive — Agent Memory
Every agent on Buda has a dedicated Drive: a persistent file system where it stores outputs, reads inputs, and maintains long-running context.
- Each agent's Drive is isolated and version-controlled
- Agents can read and write files across sessions — no context loss
- Humans can inspect, download, or upload to any agent's Drive
- Supports files up to 500 MB — PDFs, datasets, code repos, exports
Drive is what makes agents feel like employees rather than tools. They don't forget. They pick up where they left off.
🌐 Browser — Watch Them Work
Buda agents can browse the web — and you can watch them do it in real time.
- Live visual browser window per agent session
- Agents navigate, click, fill forms, scrape, and interact just like a human would
- You can intervene, guide, or let them run autonomously
- No Puppeteer config required — it just works
This is one of our most-loved capabilities. There's something deeply reassuring about watching an AI agent actually navigate a website rather than just claiming it did.
💻 Terminal — Live Shell Access
Agents have access to a real shell. Every command is visible. Every output is logged.
- Real-time terminal output per agent
- Supports long-running processes — builds, tests, deployments
- Agents can install packages, run scripts, interact with APIs
- Full audit trail for compliance-sensitive teams
For engineering use cases, this is table-stakes. For operations teams, it's a superpower.
📡 Multi-Channel Messaging
Your AI company meets your team where they already work.
Buda agents communicate natively across:
- Slack — most popular for Western tech teams
- Discord — community and startup-native
- WeChat & WeCom / 企微 — essential for China-based operations
- Microsoft Teams — enterprise standard
- Telegram — international and developer-native
- Web Chat — embedded widget for any Buda-powered product
Agents don't just respond. They initiate. They report. They ask follow-up questions. They're present across every channel, not just the one you remembered to check.
⏰ Automations & Scheduling
Buda's automation engine turns agents from reactive tools into proactive teammates.
- Cron-style scheduling — run agents at defined intervals
- Event triggers — fire agents on form submissions, webhook events, Slack messages
- Agent chains — one agent's output triggers the next
- Recurring reports — daily briefings, weekly summaries, auto-generated status updates
Set it once. Let the Claws handle it.
Marketplace: Recruit a Company
The Buda Skills Marketplace is where individual capabilities become entire teams.
The architecture is a three-tier ladder:
Skills → Agents → Teams- A Skill is a single capability: "summarize a PDF," "draft a cold email," "run a competitive analysis."
- An Agent is a configured persona built from one or more Skills: "Sales Rep who handles LinkedIn outreach."
- A Team is a coordinated group of Agents: "Marketing Team" that runs campaigns end-to-end.
Click once to recruit a team. Click twice to create a company.
For Developers: Build, Publish, Earn
If you build a Skill or Agent that solves a real problem, you can publish it to the Marketplace. Users who deploy your work compensate you. We're building the infrastructure for a new category of AI product developer — people who ship agents the way mobile developers once shipped apps.
Developer documentation, SDK access, and Marketplace submission guidelines are available at buda.im/developer.
Use Cases
Solo Founder — Operate Like a Team of Ten
You have the ideas. You don't have the headcount. Buda closes that gap.
Recruit an Analyst to monitor your competitive landscape daily. Hire a Marketer to run your content calendar. Put an Ops agent on your inbox triage. Check in once a day. Make decisions. The Claws handle execution.
This is how we ran Claw Computer before we built Buda. We know it works.
Enterprise — Scale AI Teams Safely
For larger organizations, Buda provides the governance layer that makes deploying AI at scale actually safe.
- Isolated sandboxes mean agents never share credentials, context, or data they shouldn't
- Role-based access controls for your human team managing the AI team
- Audit logs on every action: terminal commands, files accessed, messages sent
- Self-hosted option available — deploy Buda into your own cloud infrastructure. Contact sales@buda.im
Non-Technical Teams — Ops, HR, Compliance
You don't need to write a single line of code to use Buda.
- Upload a compliance document. Ask your Ops agent to summarize it and flag any new obligations.
- Paste a candidate resume. Ask your HR agent to compare it against your job description and score it.
- Submit a support ticket. Watch an agent look up the answer, draft the response, and route it for human review.
The interface is conversational. The power is enterprise-grade.
Why Not Just Use OpenClaw?
We love what OpenClaw demonstrated — it proved that powerful local agents were possible. We built on top of it ourselves for months. But when we tried to scale it across a real company, we hit walls.
| Capability | OpenClaw-style Setup | Buda + Claw Computer |
|---|---|---|
| Token cost | Standard pricing | ~80% lower |
| CPU & Memory | High per-agent overhead | ~10x more efficient |
| Machine dependency | Requires Mac Mini / Linux server | Cloud-native, no device required |
| Agent isolation | Shared process | Kubernetes-isolated sandboxes |
| Gateway layer | Required | None — built into architecture |
| Auto-sleep | Manual management | Automatic when idle |
| Multi-channel | Limited | Slack, Discord, WeChat, Teams, Telegram, Web |
| Persistent Drive | Session-based | Persisted across sessions |
| Browser visibility | Headless | Live visual window |
| Team coordination | DIY chaining | Built-in orchestration |
OpenClaw is powerful. Buda is the production infrastructure you build your AI company on.
What's Included in v0.5.0
This is the founding release. Here's what ships:
- ✅ Buda Organizer — full scheduling, automation, and team coordination engine
- ✅ Claw Computer sandbox cluster — Kubernetes-based isolated runtimes via Sandock.ai
- ✅ Drive — persistent file system for every agent (up to 500 MB per file)
- ✅ Browser — live visual web browsing per agent session
- ✅ Terminal — real-time shell access with full audit logging
- ✅ Multi-channel messaging — Slack, Discord, WeChat, WeCom, Teams, Telegram, Web
- ✅ Automations — cron scheduling, event triggers, agent chaining
- ✅ Skills Marketplace — pre-built Skills and Agents, developer publishing portal
- ✅ Role types — Analyst, Marketer, Engineer, Ops Manager, Sales Rep
- ✅ Access control — role-based permissions for human team members
- ✅ Enterprise self-hosted option — contact sales@buda.im
- ✅ Auto-sleep — idle sandboxes spin down automatically to minimize compute costs
The Mission
The bunny 🐰 is Buda's symbol. Not because AI is soft or cute. Because the bunny moves fast, works quietly, and tends to multiply.
Every Claw agent ships with this in its system prompt:
Protect humans. Push humanity forward.
That's the line that came from a late-night flight and a Japanese anime scene. It sounds simple. It's not. It means that every automation we build, every agent we deploy, every improvement we make to the platform should make humans more capable — not more redundant.
We don't believe AI is coming for your job. We believe AI is coming for your to-do list. And when that list is cleared, you'll do something more interesting with your time.
That's the company we're building. We hope you'll build yours on top of it.
Try Buda → buda.im Enterprise & self-hosted → sales@buda.im Build on the Marketplace → buda.im/developer
Released: March 31, 2026
Buda v0.5.0 is the first official Buda release note, so it should do more than list a few recent tweaks. This document explains what Buda is, what value users get on day one, and what makes this first release meaningful.
What Buda Is
Buda is a multi-agent AI work platform for individuals and companies.
The idea is simple:
You are not using one AI tool. You are building an AI company.
With Buda, you can create teams of AI employees across functions such as:
- sales
- support
- operations
- engineering
- research and analysis
These AI employees do not just answer prompts. They work inside a real operating environment, collaborate across roles, and participate in actual business workflows.
Why This First Release Matters
Most AI products improve a single person's productivity. Buda is designed to help a team or an entire organization operate through AI.
This first release establishes that foundation:
- a multi-agent workspace instead of a single chatbot
- cloud infrastructure instead of local-machine dependency
- role-based AI employees instead of generic prompts
- execution environments where agents can actually do work
- a product model aimed at long-running AI operations, not just demos
That is the right way to think about a first release note: not "what changed this week", but "what now exists for users".
What You Can Do in Buda Today
With the inaugural Buda release, users can already start building an AI workforce with concrete capabilities.
1. Hire role-based AI employees
You can organize agents by job function instead of treating every task as a blank chat.
Examples include:
- coding agents
- analyst agents
- operator agents
- sales-style agents
- support-style agents
This turns Buda into an operating model, not just a prompt interface.
1.5. Use Marketplace to recruit a company, not just an agent
The long-term value of Buda is not "pick one AI helper." It is:
click once to recruit a team. click twice to create a company.
That is why Marketplace matters in this first release.
Marketplace is where Buda starts turning AI staffing into a product surface:
- skills give agents specialized capabilities
- agents package those capabilities into recruitable roles
- teams group multiple agents into ready-to-run functions
Instead of building everything from scratch, users can start from prebuilt company blocks and assemble an AI organization much faster.
This is the beginning of Buda's product promise: opening a company should not require a long recruiting process if the first workforce is AI.
It also creates a path for developers and operators to monetize their work. Through the Developer portal, you can sign in, connect your GitHub repositories through the Buda GitHub App, and publish private or selected repos into the Marketplace.
That matters because Buda is not asking creators to give everything away as pure open source. If your repository contains durable customer value, Buda can help turn it into:
- a sellable skill
- a reusable agent
- a deployable team
- eventually, part of a repeatable company template
The job of the creator shifts from one-off delivery to continuously maintaining customer value while Buda handles distribution, packaging, and runtime.
2. Give every AI employee a real work environment
Agents in Buda can work with:
- Drive for files, context, and artifacts
- Browser for web actions and SaaS workflows
- Terminal for scripts, code execution, and development tasks
- Git for versioned software work
- Canvas and docs surfaces for planning, content, and structured outputs
That means your AI employees are closer to digital workers than chat companions.
3. Run your AI team in the cloud
Buda is built on cloud-native Kubernetes infrastructure so users do not need to self-assemble local runtimes or buy dedicated hardware.
This release already establishes:
- one-click cloud startup
- isolated runtime per AI employee
- safer multi-agent execution
- support for long-running and concurrent workloads
- an architecture that fits enterprise-grade security and large-scale deployment
By default, Buda is offered as a cloud service. But for organizations with stricter security, compliance, or infrastructure requirements, we also support self-hosted deployment. Contact us at sales@buda.im.
4. Connect AI work to real operations
Buda is designed for actual work, not isolated demos. Across the broader product surface, this release brings together:
- scheduled automations
- agent sessions and task workflows
- collaboration and sharing primitives
- channel integrations
- workspace-level organization
- marketplace for skills, agents, and teams
For users, the value is straightforward: AI work can keep moving even when you are offline.
What Makes Buda Different
Buda is not a traditional AI tool
Buda is designed around the idea of building an AI company:
- humans lead
- AI employees execute
- work is divided by role
- the platform provides the runtime, tooling, and coordination surface
Buda is not just an OpenClaw wrapper
OpenClaw proved that agents can do real work. Buda takes the next step by optimizing for cloud-scale multi-agent operations.
This release positions Buda around system-level advantages such as:
- around 80% lower token cost
- around 10x lower CPU and memory usage
- better fit for enterprise-scale, long-running workloads
That changes the economics of running not one agent, but an entire team.
User Value in This First Release
If you are a solo founder:
- operate like a larger team
- delegate execution across multiple functions
- keep work moving without hiring full headcount up front
- recruit functional AI teams through Marketplace instead of assembling every role manually
If you are an enterprise:
- give each AI employee an isolated runtime
- launch cloud-based AI teams without complicated infra work
- turn recruiting, setup, and runtime management into one product flow
- build toward scalable, always-on AI operations
- deploy with an architecture designed for enterprise-grade isolation and security
- choose managed cloud or discuss self-hosted deployment at sales@buda.im
If you are evaluating the platform:
- this release proves Buda is already more than a prototype
- the product has real surfaces for execution, coordination, and extension
- the release establishes the product category clearly: AI company platform
Included Foundations in v0.5.0
This first release note consolidates the major foundations built into Buda so far:
- multi-agent workspace model
- cloud-native sandbox runtime
- Drive, Browser, Terminal, and Git working surfaces
- marketplace foundations for skills, agents, and teams
- automations and scheduled execution
- collaboration, sharing, and workspace improvements
- multi-language product experience
- onboarding and first-run refinement
- release-note and versioning synchronization for the public product surface
Breaking Changes
None. This is the inaugural release and should be read as the baseline definition of Buda's product value, not a migration notice.