Cloudflare is expanding its Agent Cloud with new tools aimed at developers interested in building and running agentic AI in development environments. The company says the collected tools will make agents easier to deploy and manage using the type of development model that involves several individual agentic instances deployed simultaneously to work on a project.
Cloudflare states that agentic, multi-agent systems require different infrastructure from conventional cloud development environments and that existing options, including virtual servers and container environments can be too costly or inefficient to deploy widely.
The company’s updated platform will combine compute, storage, security, and deployment tools under a single, umbrella platform.
Matthew Prince, Cloudflare’s co-founder and chief executive, claims agents will be used increasingly to write code, and that AI-assisted and partially-autonomous software development will become common enough to justify an infrastructure designed specifically for such workflows. The platform expansion falls under Cloudflare’s Workers serverless computing offering, and is intended for enterprise use.
Systems involving multi-agent, semi-autonomous development bots made mainstream news recently in the form of OpenClaw, a service that received swift endorsement from Nvidia, among others. Where the original purpose for OpenClaw development found its home in the hobbyist market, its publicised effects have sparked the imagination of enterprise IT decision-makers. Cloudflare is among the companies that see the same model as having potential to be effective for business-class, enterprise software.
Cloudflare’s announcement includes the addition of Dynamic Workers, isolated runtimes for AI-generated code, which can perform discrete, smaller tasks such as to call an API or perform simple data transformations. Such processes often don’t warrant a dedicated container environment or virtual machine, and the company claims that deploying Dynamic Workers is faster and cheaper than the existing provisioning methods for VMs and containers. Cloudflare’s cloud infrastructure means that Dynamic Worker instances can scale to high numbers of executing routines with no knowledge of technology such as Kubernetes. Its lightweight execution model is a better fit for agentic workloads that start, execute and stop frequently, according to a company press release.
Part of the same announcement, Artifacts is described as a Git-compatible storage primitive for AI-generated code and files. It’s designed as a version control system for autonomous systems that produce high volumes of code. Developers can create large numbers of repositories, code forks, and replicated data sets, which will remain addressable from standard Git clients.
On a larger scale, Sandboxes are persistent Linux environments with shell, background daemons, and a filesystem, acting as full virtual machines. The Think framework, part of the Agents SDK, gives developers tools that can persist over multiple sessions and work on longer-lasting tasks. In the back end, Cloudflare can now offer an expanded catalogue of AI models, thanks in part to its acquisition of Replicate. Developers will be able to choose from several proprietary or open-source models via one interface.
Cloudflare’s positioning as an intermediary between model vendors and application developers gives its users the deployment and agentic development workflow flexibility for which Cloudflare sees an increasing demand in the enterprise space.
The platform’s consolidation is built on the expectation that agents will become a meaningful class of software development tools that has meaningful use in business-critical software development environments. Cloudflare’s bet is that developer workflows will evolve from vibe-coded multiple agents spun up in order to address hobbyists’ goals, to become a viable alternative to traditional, learned developer skills.
If nothing else, the Cloudflare offerings will allow development teams to experiment on enterprise-class infrastructure using the same type of workflows that have led many thousands of hobbyist developers to invest in dedicated Mac Minis to run their workloads.
(Image source: “Cat Meets Lobster” by eviltomthai is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0)
Want to dive deeper into the tools and frameworks shaping modern development? Check out the AI & Big Data Expo, taking place in Amsterdam, California, and London. Explore cutting-edge sessions on machine learning, data pipelines, and next-gen AI applications. The event is part of TechEx and co-located with other leading technology events. Click here for more information.
DeveloperTech News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.