What's new in AI/works™: Stronger governance, observability and deeper enterprise context

In January, we launched AI/works™, Thoughtworks’ agentic development platform for modernizing legacy systems, building new enterprise systems and redefining the software development lifecycle.

Its Code to Spec, Dynamic Spec and Spec to Code capabilities help organizations understand and modernize legacy systems, define requirements, generate code, test outputs while monitoring through Runtime Operations and the Control Plane to continuously improve enterprise technology.

AI/works™ new release version 2.0 builds on that foundation with stronger enterprise governance, security and observability, along with new and enhanced capabilities for modernization, evaluation and enterprise knowledge.

These updates are designed to help organizations accelerate AI-enabled software delivery while maintaining the visibility, security and control needed in complex enterprise environments.

Generating code is only one part of enterprise software delivery.

Organizations must also understand the systems they already have, preserve important business rules and verify that generated software meets its intended requirements. They need to maintain audit trails, monitor security and understand how their compute and token resources are being used.

AI-enabled development also requires trusted business, industry and regulatory context. Without that context, generated outputs may not reflect the organization’s requirements or the environment in which its technology operates.

AI/works™ new release addresses these practical enterprise needs.

Expanded Control Plane capabilities give teams a central place to manage users, workspaces, projects and access policies.

Teams can maintain complete audit trails, monitor their security posture and gain visibility into compute and token spend. This gives technology leaders and delivery teams a clearer view of how AI/works™ is being used across their projects.

As organizations move from experimenting with agentic AI to applying it at scale, this level of governance and visibility has become increasingly important.

Understanding legacy technology is the first step in modernization. Organizations must also define what should replace or extend it.

AI/works™ helps teams move beyond understanding legacy code by enabling them to visualize their future-state architecture and code base.

Robust modernization rules help preserve important business logic, identify potential conflicts and define the architectural patterns used to generate new code. This provides a clearer path from understanding an existing system to building the technology that will support the organization in the future.

AI-generated code must be evaluated against the requirements it was intended to meet.

The Evals capability traces requirements through to generated code and checks outputs against defined requirements. This helps teams understand what is ready to ship, identify potential risks and surface security gaps.

The goal is not to remove human judgment. It is to provide technologists with better evidence throughout the delivery process so they can review, refine and approve outputs with greater confidence.

Effective AI-enabled development depends on access to trusted organizational knowledge.

Knowledge Fabric ingests enterprise information into a secure, continuously evolving knowledge repository. Information is discoverable, can be organized by industry and domain, then enriched with regulations and proprietary enterprise knowledge.

This gives AI-enabled workflows the business and technical context needed to produce more relevant outputs while maintaining appropriate controls over how information is accessed and used.

AI/works™ new release can help teams:

1. Analyze existing applications and explain their behavior through code to spec
2. Visualize future-state architecture and apply modernization rules
3. Capture and refine requirements, architecture and design in the form of the SuperSpec through dynamic spec
4. Generate and refine working software through spec to code
5. Check generated outputs against requirements through Evals
6. Ground development in trusted organizational knowledge through Knowledge Fabric
7. Continuously monitor and maintain your applications with Runtime Ops by detecting change, updating the Super Spec, and regenerating impacted code to keep systems modern by default.
8. Govern the platform’s AI infrastructure - activity, security, quality and compute and token resource use through the Control Plane

AI/works™ combines our rich heritage of proven engineering practices and sensible defaults with agentic AI and our expert technologists to deliver scalable, enterprise-grade systems while reducing risk and accelerating time to value.

A leading European airline is working with Thoughtworks to apply AI-enabled engineering to a complex modernization program.

The airline is using components of AI/works™ to reverse-engineer legacy systems as it continues to evolve its digital architecture and e-commerce capabilities. This work is helping the airline understand legacy systems more quickly and build a stronger, more composable digital foundation while maintaining resilience, quality and the customer experience. It will also support the next phase of the airline’s technology evolution.

AI/works™ supports Thoughtworks’ 3/3/3 methodology, which helps organizations make focused progress over three days, three weeks and three months as they move from initial ideas toward production-ready technology at enterprise scale.

The platform is designed to work across mainframe, monolithic, cloud and distributed technology environments. It also works with leading technology ecosystems including AWS, Google Cloud, Microsoft Azure, Databricks, Anthropic, Nvidia and Snowflake. Through Thoughtworks’ collaboration with Mechanical Orchard, AI/works™ also supports mainframe modernization.

Teams interested in the release can:

• Join the LinkedIn Live session scheduled for August 19, 2026.
• Explore the AI/works™ website for more details
• Contact Thoughtworks to discuss a client project or co-innovation opportunity

We will continue to evolve AI/works™ through practical delivery experience, feedback from our technologists and collaboration with clients addressing complex modernization and product engineering challenges.

Learn more about AI/works™ new release version 2.0 and request a demonstration at AI/works™.