Sauce Labs lets enterprises bring their own model to AURA

Sauce Labs has added bring-your-own-model options to its AURA software testing platform, allowing enterprise customers to use different large language models within the same release assurance system.

The update is available to enterprise customers through their account teams and supports open-source, open-weight and proprietary models as AURA's underlying model layer.

The move addresses a growing problem for software teams as AI coding tools increase output faster than testing and verification can keep up. Developers now produce 741% more code, while release velocity has risen by less than 20%, according to Sauce Labs.

AURA is designed to write, run, analyse and repair tests with human oversight. In the latest update, Sauce Labs separated the model layer from the rest of the platform's testing workflows and execution infrastructure, allowing customers to switch models without changing the system they use to verify software before release.

That matters for larger organisations seeking to avoid dependence on a single AI provider while maintaining governance, security and internal standards. Customers can choose models that fit privacy, architecture and policy requirements, and change them as regulations or internal rules evolve, the company said.

Dr Prince Kohli, Chief Executive Officer of Sauce Labs, said the move responds to concerns about dependence on a single supplier.

"Enterprise AI should expand choice, not create a new layer of lock-in. AURA lets customers choose the intelligence layer that fits their business while keeping one consistent, governed system for release assurance. Models can change; the foundation for production confidence should not," said Dr Prince Kohli, Chief Executive Officer of Sauce Labs.

Testing gap

The announcement comes as software companies try to manage the consequences of AI-generated code entering production systems at greater scale. As coding output rises, testing teams are under pressure to keep up with both the volume and complexity of changes.

Sauce Labs positions AURA as the layer between code generation and software release. Rather than tying testing to one model provider, it argues that enterprises should be able to use whichever models best meet current needs for cost, accuracy, latency and governance, while keeping a single controlled validation process.

The company also stressed the role of execution data. AURA draws on evidence from more than 8.7 billion test executions to assess how software behaves in real conditions and feeds production errors back into later testing cycles, according to Sauce Labs.

That focus on execution evidence is part of a broader argument about ownership of data and context in enterprise AI systems. Some software suppliers are trying to package context and workflow information as a proprietary platform feature, but Sauce Labs argued that customers should retain control over the models, systems and data they already use.

Enterprise demand

Early demand for the model-choice feature has been strong, Sauce Labs said, though it did not disclose customer numbers or revenue tied to the rollout. The platform is used by more than 300,000 enterprise users, including eight of the world's 10 largest financial institutions, according to the company.

Sauce Labs also cited independently validated customer results for AURA, including more than 90% fewer production incidents, 47% faster release cycles and 38% of engineering capacity reclaimed. It did not provide further detail on the methodology behind those figures.

The update fits with Sauce Labs' longstanding ties to open software ecosystems. The company was founded by the creators of Selenium and Appium, two widely used test automation frameworks, and said AURA is framework-agnostic and designed to work with existing CI/CD environments.

That means organisations can change the AI model layer without replacing their development pipelines or fragmenting release controls across separate tools. Sauce Labs also said AURA can run tests across more than 10,000 devices as part of its broader execution setup.

For enterprises, the question is becoming less about whether to use AI in software development and more about how to govern it once code generation becomes routine. Sauce Labs' latest move reflects that shift by treating model choice as a procurement and control issue as much as a technical one.

AURA connects to the models and systems customers have already selected, tuned and governed, rather than replacing them, while adding verification based on how software performs under real conditions, the company said.