Why we’re betting on OpenClaw

We’re introducing Enterprise OpenClaw by RAW Labs, our offering built on OpenClaw and OpenClaw Enterprise, with business integrations, deployment, maintenance and support.

There are plenty of agent frameworks we could build on. We chose OpenClaw because it combines things we care about: an open codebase, independence from any one model provider, a large community, substantial security engineering and a team whose direction we believe in. Its beginnings as a personal agent are part of what makes that combination interesting for enterprise use.

The personal agent was the beginning

OpenClaw became famous by making personal agents tangible. People could run an agent themselves, connect it to applications and experiment with what it could do. What began as Peter Steinberger’s personal project grew into a worldwide community and an independently stewarded open-source project. The Foundation describes that history in its founding announcement.

That attention brought a demanding audience. Developers extended it, researchers examined it, and organizations began exploring how to use it internally. Our view is that this experience gives OpenClaw a considerable head start: the project has had to address a breadth of uses and questions that would be difficult for a smaller effort to encounter on its own.

For us, its popularity matters because of the engineering and community that have grown around it. We want to build where that work is happening.

The scrutiny matters as much as the adoption

OpenClaw has attracted intense security scrutiny, and its team has made security a substantial, public engineering effort.

There is concrete work behind that statement. Just in September, OpenClaw published the results of a broad Trail of Bits audit conducted through OpenAI’s Patch the Planet initiative. The team reported that every actionable issue had been repaired, with fixes shipped in stable releases. Its public security programme also covers vulnerability disclosure, skill scanning and regression checks. See the audit account and security overview.

We put weight on a team that invites examination, publishes what it learns and follows through on fixes. That record is one of our reasons for choosing OpenClaw as an enterprise foundation. It gives us something concrete to assess and contribute to as the project develops.

It gives us room to choose

OpenClaw’s breadth is another reason we’re building on it. It supports a wide range of model providers, local models, messaging applications, tools and extensions. Its integration directory currently lists 29 chat channels, 64 model and media providers, and 142 official plugins, alongside community extensions. Those are different categories of support, but together they show how much ground the ecosystem already covers. OpenClaw integrations

For a company, that means more options when deciding how an agent should fit into its environment. Different tasks may call for different models. Teams already have applications they work in. Infrastructure requirements vary.

We want a foundation that accommodates those choices. OpenClaw’s model-agnostic approach and broad ecosystem are a strong fit for that aim.

Enterprise use is already informing the project

OpenClaw Enterprise brings the project’s approach into organizational deployment, with an open control plane for managing agents across teams. It began at OpenAI and was donated to the OpenClaw Foundation, with further development in collaboration with Red Hat and NVIDIA.

It is also new. The September announcement described it as suitable for internal pilot workloads ahead of its 1.0 release. That same announcement reports internal pilots at OpenAI and Red Hat, including OpenAI agents working with codebases and plugins. OpenClaw Enterprise announcement

That is a meaningful combination for us: an early enterprise project whose development is already informed by people using agents inside large technology organizations. We want to participate in that work while bringing our own experience with business systems and customer deployments.

Can one foundation support that much different work?

Our decision to build on OpenClaw reflects a broader view of where software is going. We expect general-purpose agent systems to take on an increasing range of work that previously required separate applications, integrations and workflow projects.

Our own work over recent years — in data management, data integration and workflows — has influenced that view. The business problems vary, but many of the underlying requirements recur: get information from several systems, understand it in context, carry out a sequence of actions and involve a person when needed. Historically, much of that behaviour had to be written specifically for each application.

With agents, more of it can be expressed through instructions and tools on a common runtime. An agent can use existing applications, execute code where appropriate and combine repeatable procedures with work that requires interpretation. OpenClaw already brings together tools, memory, schedules, plugins and workflow capabilities. Its integration ecosystem shows how broadly that foundation can be extended.

That changes how we think about building for a company. A new responsibility may call for another agent, a new skill or an additional connection to an existing system. It need not require a separate application with its own interface and infrastructure.

This is why we’re comfortable concentrating our effort on one foundation while pursuing different business applications. The specialization belongs in the agent’s instructions, tools, knowledge and extensions. The runtime can be shared.

There will still be work that calls for conventional software, exact calculations or dedicated systems. We expect agents to use those systems as part of their work. Our bet is that the range of responsibilities they can handle will keep expanding, and that a growing share of software development will happen around the harness.

OpenClaw is where we want to build that experience.

An independent foundation

OpenClaw’s independence is another reason for our choice. It is stewarded by the OpenClaw Foundation, an independent American 501(c)(3) nonprofit established to keep the project open and neutral.

Peter Steinberger joined OpenAI, and OpenAI supports the Foundation as a donor. OpenClaw itself remains independent; its documentation explicitly states that donors do not own, control or direct the project. Foundation announcement, ownership FAQ

For us, that structure fits the role we believe OpenClaw can play: a common foundation that companies can build on, with participation from competing model providers, infrastructure companies and independent developers. We want to contribute to an ecosystem with room for all of them.

We’re also choosing the people

A platform decision includes a judgment about who will keep building it.

We admire the team’s vision for personal agents and the work of the maintainers turning it into a shared foundation. The community now includes more than 3,300 code contributors, with maintainers from organizations including Microsoft, NVIDIA, OpenAI, Red Hat and Tencent. Foundation people directory

We expect some agent platforms to become much more widely established than others. Nobody knows how that will play out. Our judgment is that OpenClaw is already ahead in the combination of ecosystem breadth, community participation and public security work that matters to us. We’re prepared to invest our engineering effort behind that judgment.

What you get from RAW Labs

With Enterprise OpenClaw by RAW Labs, you can buy support for the platform and services to put it to work in your company.

Our subscription covers maintenance and support for the combined system: OpenClaw, OpenClaw Enterprise and the integrations and configurations included in the supported release. That includes compatibility work, tested updates, security triage and help investigating platform problems, within an agreed support scope and long-term stability.

Our services cover the work of getting from an idea to a working deployment. We can help identify the first application, prepare the environment, deploy the platform, integrate your systems and build the initial agents. Security work is part of that engagement: assessing the deployment, configuring the relevant controls and testing them against the intended use.

We can also build the extensions your application needs, including custom tools, connectors, plugins, skills and workflows. We work with your team on evaluation, training and handover, and can provide ongoing maintenance for that custom work. Managed operation is available under a separately agreed scope.

Our work builds on the engineering of the OpenClaw and OpenClaw Enterprise teams. We contribute improvements upstream and intend to continue doing so as our work develops. Customer deployments give us practical problems to investigate; where the resulting improvements belong in the shared projects, we want them to benefit the wider community.

You can engage us to help your own team build and operate agents, or to take on more of the implementation and ongoing work. The offering combines a maintained foundation with the expertise to apply it to your business.

If your company is exploring OpenClaw, or agents, or even AI integration in general, talk to us about what you want to build.