Kilo launches KiloClaw, allowing anyone to deploy hosted OpenClaw agents into production in 60 seconds
Source: VentureBeat
KiloClaw: Deploy a Production‑Ready OpenClaw Agent in Under 60 Seconds
In the rapidly evolving landscape of artificial intelligence, the distance between a developer’s idea and a functioning agent has historically been measured in hours of configuration, dependency conflicts, and terminal‑induced headaches.
That friction point changed today. Kilo, the AI‑infrastructure startup backed by GitLab co‑founder Sid Sijbrandij, has announced the general availability of KiloClaw, a fully managed service designed to deploy a production‑ready OpenClaw agent in under 60 seconds.
By eliminating the “SSH, Docker, and YAML” barriers that have gatekept high‑end AI agents, Kilo is betting that the next phase of software development—often called “vibe coding”—will be defined not just by the quality of a model, but by the reliability of the infrastructure that hosts it.
Technology: Re‑engineering the Agentic Sandbox
OpenClaw has emerged as a viral phenomenon, amassing over 161 k GitHub stars by offering a capability many proprietary tools lack: the ability to actually perform tasks—controlling browsers, managing files, and connecting to 50+ chat platforms (WhatsApp, Signal, etc.).
“OpenClaw itself isn’t the hard part… getting it running is.” – Scott Breitenother, Kilo co‑founder & CEO (VentureBeat interview)
Architectural Shift
| Traditional “Mac Mini on a desk” | KiloClaw’s Multi‑Tenant VM |
|---|---|
| Users provision their own hardware or VPS. | Runs on a multi‑tenant VM architecture powered by Fly.io (Chicago‑based, developer‑focused public cloud). |
| Isolation & security depend on the individual developer’s setup. | Provides enterprise‑grade isolation, sandboxing, and network security out‑of‑the‑box. |
| High risk of misconfiguration (exposed API keys, open internet). | Two external proxies protect the VM, preventing accidental exposure. |
“We have a virtual machine that is a hosted OpenClaw instance, and we’re handling all that network security, sandboxing, and proxies that an enterprise company would require.” – Breitenother
Product: The “Mech Suit” & the 3 am Crash
A primary pain point for OpenClaw users is the “3 am crash”—locally hosted Node.js processes dying silently overnight with no health monitoring or auto‑restart.
KiloClaw solves this with:
- Built‑in process monitoring and a cloud‑native always‑on state.
- Persistent listening for inbound messages (WhatsApp, Slack, etc.).
“KiloClaw is just running and listening… It’s always on, waiting for your WhatsApp message or your Slack message.” – Breitenother
Agentic Affordances (the “exoskeleton for the mind”)
| Feature | Description |
|---|---|
| Scheduled automations | Set cron jobs for research, repo monitoring, report generation while you’re offline. |
| Persistent memory | “Memory Bank” stores context in structured Markdown files within the repo, preserving state across model swaps. |
| Cross‑platform command | Trigger the agent from Slack, Telegram, or a terminal, maintaining a unified execution state. |
“We’ve moved our engineers to be product owners. The time they freed up from writing code, they’re actually doing much more thinking—setting strategy for the product.” – Breitenother
The “Gateway” Advantage: 500+ Models, No Lock‑In
KiloClaw integrates natively with the Kilo Gateway. While OpenClaw originally leaned on Anthropic’s models, KiloClaw lets you toggle among 500+ models from providers such as OpenAI, Google, MiniMax, and open‑weight models like Qwen or GLM.
“Your preferred model today may not be the same—and honestly shouldn’t be the same—a month and a half from now.” – Breitenother
Pricing & Credits
- Zero markup on AI tokens – you pay the exact API rates from the model vendors.
- Kilo Pass (subscription tier) provides bonus credits (e.g., $199 / month for $278.60 in credits) for high‑volume agentic work.
How to Get Started with KiloClaw Right Now
-
Sign in or register
- Navigate to the Kilo Code web app: app.kilo.ai
- Sign in using your existing account (GitHub, Google OAuth, etc.).
-
Create your instance
- Select the “Claw” tab from the side navigation menu to open the KiloClaw dashboard.
- Click “Create Instance” to begin provisioning your agent. (See image above for button location.)
-
Choose your model
- Pick a default AI model from the dropdown menu.
- Options include a variety of free (temporarily) models and paid alternatives.
(The original content cuts off here; continue with the model selection UI as needed.)
MiniMax
Configure messaging channels (optional)
During setup you can optionally connect your agent to Discord, Telegram, or Slack and communicate with your KiloClaw agent directly over those channels — instead of on the Kilo Code website. To move faster you may skip this step; you can always add the supported bot keys and configure these channels later in the instance settings.
Provision and start
- Click “Create and Provision” to set up your virtual machine.
- Once the instance is provisioned, click “Start” to boot the agent (typically takes only a few seconds).
Verify and access
- Click the “Open” button to enter the OpenClaw interface.
- For security, click “Access Code” to generate a one‑time verification token that validates your device for the first time.
Begin vibe coding
After verification you can start interacting with your agent directly in the chat interface. The agent runs 24/7 on a dedicated virtual machine, listening for commands across all connected platforms.
PinchBench – Benchmarking the Agentic Era
According to Brendan O’Leary, Developer Relations at Kilo Code (former Developer Evangelist at GitLab), users unsure which model to select should consult PinchBench, an open‑source benchmarking tool that evaluates models on 23 real‑world agentic tasks (e.g., email sorting, blog post generation).
What PinchBench Offers
- Agent‑focused evaluation: Unlike traditional benchmarks such as MMLU or HumanEval that test isolated chat prompts, PinchBench tests agents on multi‑step tasks like calendar management and multi‑source research.
- Broad model coverage: Helps developers navigate the choice among 500+ models.
- Scatter‑plot visualization: Shows “Cost to Intelligence,” highlighting models that deliver the highest proficiency for the lowest price. O’Leary calls this his “favorite graph for looking at models—how much you spend versus how much is the success rate.”
Origin & Inspiration
- The project was spearheaded by O’Leary, who noted that the benchmark was “kind of inspired by… other little kind of fun benches” created by developer YouTuber Theo Browne (@t3dotgg), CEO/Founder of Ping Labs.
- O’Leary ran the benchmark “hundreds and hundreds of times against OpenClaw” to ensure accuracy and launched a YouTube series titled “Will It Claw?” to test KiloClaw on various tasks.
Judging System
- A high‑end judge model—Claude 3.5 Opus—grades the output of other models.
- The judge model provides specific notes on execution quality, ensuring consistent, objective evaluation.
Transparency & Self‑Hosting
- For those who prefer to host their own infrastructure, O’Leary provides a skill file that can be downloaded to benchmark a personal OpenClaw instance independently.
“We’re doing this work anyway to know which defaults we should recommend,” O’Leary said. “We decided to open source it because the individual developer shouldn’t have to think about which model is best for the job. We want to give people more and more information.”
Philosophy
- O’Leary likens the benchmark to “the Olympics”, with tasks ranging from objectively graded to those requiring nuanced assessment.
Industry Context: Distinguishing KiloClaw from the Growing OpenClaw Family
- Nanoclaw – lightweight variant gaining traction.
- Runlayer – targets the enterprise “Virtual Private Server” niche.
KiloClaw’s Differentiators
- No fork: “It’s not a fork, and that’s what’s important,” says Breitenother. KiloClaw hosts the actual OpenClaw version on a well‑tuned, managed virtual machine, ensuring automatic updates as the core project evolves—no manual
git pullrequired. - Open‑core licensing: While KiloClaw is a paid hosted service, the underlying Kilo CLI and core extensions remain MIT‑licensed, allowing community auditing—a critical feature for security‑conscious enterprises.
Conclusion: Toward an Agentic Future
The launch of KiloClaw marks a strategic move by Kilo to expand its user base beyond “wonky” developers to enterprise managers and non‑technical professionals. By offering a one‑click path to a production agent, the company aims to democratize the “magical moments” of AI.
- In the first two weeks, 3,500+ developers joined the waitlist, pushing KiloClaw in directions ranging from Discord management to repository maintenance.
- Breitenother: “Our mission is to build the best all‑in‑one AI work platform. Whether you are a developer, a product manager, or a data engineer, we want all of these personas to experience the magic of the exoskeleton for the mind.”
KiloClaw is available now, offering 7 days of free compute for all new users, with thousands of developers already on board.
Already having cleared the waitlist, the era of the managed AI agent appears to have arrived—no Mac Mini required.