Witbe’s new hybrid automation framework allows teams to combine deterministic testing with agentic AI on real devices, building and scaling agentic automation across video workflows.
Witbe’s Agentic SDK, a new test automation framework, is designed to help video service providers build, operate and scale agentic automation across video workflows – from testing and validation to monitoring – while keeping automation grounded in deterministic logic and the behaviour of real consumer devices.
The framework reduces on-boarding time and minimises ongoing test maintenance. It enables automation to handle unplanned UI changes and human-like content assertions autonomously. Typically, such assertions are clear but debatable claims that can be supported with evidence.
Rather than replacing existing automation with fully autonomous AI, Agentic SDK introduces a hybrid model that allows teams to apply agentic AI selectively, where it adds value. Meanwhile, the reliability, transparency and repeatability required in production environments are preserved.
Scaling Automation across Devices, Apps and Markets
When automation is applied to the complexity of the real world, teams are forced to rethink how testing and monitoring are designed and scaled. As video services expand across hundreds of applications, devices, operating systems and markets, automation must handle constant variability while still delivering precise, repeatable results.
AI adds the flexibility needed to navigate changing interfaces and unpredictable conditions, while deterministic logic anchors automation in known validation procedures and measurable KPIs. The main value of a deterministic system is its predictability and ability to consistently reproduce the same output from the same set of inputs. Not at all random, the outcome depends only on the initial conditions and the logic rules.
Combining these characteristics with AI’s flexibility, within a single governed framework, helps to scale automation across complex workflows while limiting manual intervention. The long-term cost of maintaining test scripts is reduced, as well as the cost of running regression tests to prevent updates from interfering with existing functionality.
Witbe's Agentic SDK was built to address these challenges and serves as a structured way to introduce agentic AI into automation workflows without sacrificing determinism or observability.
One Platform, Many Ways to Automate
Teams can use this single automation platform to run simple, one-off goal-based tests, and then introduce more complex measurements through preset scenarios with hybrid continuous execution. Ultimately, they can customise workflows in detail using a python-based, AI-native SDK – all within the same framework. This approach enables teams with varying skill levels, maturity and expertise to work from a common foundation – without being forced into a one-size-fits-all automation model.
Goal-based Automation for Fast Results
For teams getting started with automation or occasional validation processes, the Agentic SDK supports goal-based execution, driven by agentic AI. Users can describe a test objective in plain language, allowing the system to generate, execute and analyse tests automatically on real devices.
This mode is useful for fast exploratory testing, release validation and non-continuous QA workflows, delivering immediate results without scripting, lengthy set-ups or ongoing framework maintenance. Teams can move from onboarding to first results in a matter of days rather than weeks or months. They can edit or refine generated steps as needed, while agentic execution handles navigation and adaptation during test runs.
Hybrid Execution with Pre-Built Scenarios
In contrast, for continuous testing and operational workflows, Agentic SDK presents pre-built scenarios based on Witbe’s expertise, specialised in video QA and monitoring. These scenarios deliver recognised KPIs for common use cases such as QA validation, live TV monitoring and VOD asset verification
Execution combines deterministic logic with agentic assistance only when necessary, enabling reliable 24/7 automation without the maintenance burden typically associated with traditional scripting. This hybrid model ensures consistent results while allowing automation to adapt smoothly between UI changes, unexpected states and unplanned events across devices without breaking scenarios or requiring manual rework.
Full Customization with the Agentic SDK
For advanced use cases, Agentic SDK gives full access to an open, extensible automation environment. The SDK serves as an AI-native, Python-based environment designed for secure execution and scale, where agentic reasoning, deterministic validation and governance are engineered together rather than bolted on.
Teams can create, edit and extend workflows by mixing AI-driven and algorithmic steps. They can define custom KPIs and build reusable automation components. The SDK enables organisations to move beyond testing and apply agentic automation across broader video workflows. This flexibility allows teams to innovate without losing governance, reuse or control.
Real-Device Testing and Monitoring
Agentic SDK integrates natively with Witbe’s real-device testing and monitoring infrastructure, enabling agentic automation to operate on the same Smart TVs, set-top boxes, mobile devices, and browsers used by real viewers.
This hybrid approach reflects Witbe’s long-standing expertise in testing and monitoring video services on real devices. With more than two decades of experience supporting broadcasters, streaming platforms, and operators worldwide, Witbe has built its technology around deterministic measurement, repeatable validation, and real-world playback behaviour; principles that now form the foundation of the Agentic SDK.
“AI is moving fast, and in many cases, it’s being layered onto products without changing how they truly work,” said Mathieu Planche, CEO of Witbe. “At Witbe, intelligence has always been part of the product: from our early algorithms to today’s agentic systems. With the Agentic SDK, we are redefining how testing and monitoring are designed, combining agentic AI and deterministic logic in a single governed framework. This marks a structural shift for our industry, and for how video services will be tested and operated going forward.”
Witbe Agentic SDK is available immediately as part of Witbe’s automation and monitoring platform. It is delivered as part of Witbe Suite’s latest release, version 41, compatible with all existing Witbox deployments, ready for use with no additional hardware or infrastructure to deploy. www.witbe.net
