SecurityBrief US - Technology news for CISOs & cybersecurity decision-makers
United States
Upsun launches Dispatch to govern AI software workflows

Upsun launches Dispatch to govern AI software workflows

Tue, 29th Sep 2026 (Today)
Joseph Gabriel Lagonsin
JOSEPH GABRIEL LAGONSIN News Editor

Upsun has launched Upsun Dispatch, a platform for AI and human software development workflows aimed at teams that want more oversight of AI agents in software delivery.

The system is designed to move AI-assisted development beyond individual developers using tools on local machines into a shared process with approvals, review, and cost tracking. Upsun Dispatch runs agent work in cloud-based isolated sandboxes and records prompts, code changes, approvals, and spending for each workflow run.

The launch comes as software companies and their customers try to determine how much autonomy to give AI agents in production work. Coding assistants have become more capable, but businesses still face delays in review, coordination, and risk management after code is generated.

Upsun is introducing the platform with two initial workflows: Automated Code Review and Issue-to-Code. The first is designed to automate parts of code inspection. The second links a development issue to code generation inside a managed workflow that still requires human decisions at defined stages.

Rather than positioning AI agents as standalone tools, Upsun is framing the software around workflow management. Teams can set checkpoints where people must approve or review work before it progresses, while agents handle surrounding tasks within the same process.

That structure also opens the workflow to staff outside core engineering teams. Product, design, security, and quality assurance teams can participate in the same governed process, potentially expanding AI use in software projects beyond the developers who typically operate these tools directly.

Another focus is cost visibility. Each workflow run is logged with its associated usage cost, giving organisations a clearer view of token spending and the economics of agent use as it happens. This addresses a common concern among companies adopting AI tools across teams, where usage can spread quickly without a central record of what is being spent and on which tasks.

The platform is model-agnostic and stack-agnostic. At launch, it supports GitHub and GitLab, along with Linear and Jira, allowing teams to connect it to established code repositories and issue-tracking systems without committing their software process to a single AI model provider.

Fabien Potencier, Chief Product and Technology Officer at Upsun, said the company built the product around a problem that emerges after code generation becomes easier.

"AI is making it possible to produce code much faster. But writing code is only part of the job. Rapid code creation causes bottlenecks and complications that have moved downstream. This is what delays time to release and has made costs unmanageable," said Fabien Potencier, Chief Product and Technology Officer at Upsun.

Potencier said the issue extends beyond individual productivity gains for developers.

"Teams still need to decide what should be built, review the result, and take responsibility for what reaches production. As AI becomes a larger part of development, we need to think beyond the individual developer and give the whole team, beyond engineering, a shared way of working with it," Potencier said.

Broader platform

Upsun is positioning Dispatch alongside its existing cloud platform, which it says supports the wider software development lifecycle, from planning and coding to building, testing, deploying, and running applications. The company was previously known as Platform.sh and says the underlying platform has been used for production applications for more than a decade.

The business describes Dispatch as a response to its own use of AI in software development and to conversations with design partners across software companies, agencies, and large enterprises. This suggests the initial target market is organisations already using AI tools in development and beginning to encounter problems around governance, consistency, and cost control.

There is also a practical distinction in how the platform is intended to be used. Instead of leaving agents on employee laptops with varying prompts and workflows, Upsun is trying to establish repeatable processes in a shared environment. For businesses under pressure to demonstrate auditability and maintain release standards, that may prove as important as the productivity gains from the agents themselves.

Upsun said the product will be offered with a free tier that includes 20 agent-minutes per day, while paid plans are usage-based and support bring-your-own-model keys.

Fred Plais, Chief Executive Officer at Upsun, said: "The technology grew out of Upsun's own experience adopting AI across software development, commitment to innovation and thought leadership, and our decade-long engagement with the developer community and technology leaders that need to balance time to market with quality."