Thoughtworks, a global technology consultancy that integrates design, engineering and AI to drive digital innovation, today released volume 35 of its Technology Radar, a biannual report informed by the organization's experiences with clients.
In developing this edition, the Technology Radar team observed that as agents are handed larger, longer-running tasks, the hardest engineering challenges increasingly lie beyond interactions with the model itself: securing the environment in which agents operate, deciding what they can access, establishing meaningful human oversight and controlling the cost of running them.
One indication of this focus is where Radar makes its strongest recommendations. The 11 practices and technologies placed in its Adopt ring, which include techniques such as agent sandboxing and threat modeling, signal growing importance of safeguards needed to run AI reliably in production and underline need to act now.
Earlier editions traced growing capabilities of AI in software engineering, and volume 34 examined friction and risks of using it at scale. Volume 35 turns to managing those risks as adoption grows. Its themes span systems that govern agent behavior, role of human judgment and infrastructure that determines cost and reliability.
"Not long ago, the question we heard from clients was how fast agents could write code. Now it's whether anyone can trust, review and afford what they produce," said Rachel Laycock, chief technology officer at Thoughtworks. "Models will keep improving, but lasting value will go to the organizations that invest in the engineering and oversight around them."
Harness engineering at scale: The environment around an AI agent, including prompts, tools, memory, guardrails and feedback loops, is becoming product in its own right, with owners, versions and dependencies. As growing ecosystem of plugins and supporting tools takes shape, organizations must treat harness as engineered system with appropriate evaluation, security, access controls and ongoing maintenance.
Putting more agents to work, without losing human judgment: Developers increasingly running multiple agent sessions in parallel and delegating longer-running tasks including unattended overnight builds. As agents produce more work, human review and software delivery pipelines can become bottlenecks. Radar cautions against "human in the loop theater," where approval becomes ritual rather than meaningful check, and encourages teams to assess review capacity before scaling agent use.
Beyond frontier models: Proprietary frontier models no longer obvious default for every task. Open-weight models narrowed gap for many workloads and offer additional deployment options including running within organization's own infrastructure. Radar cautions against defaulting to frontier model for every task.
Economics of running AI at scale: Cost extends beyond model pricing. Routing, batching, storage, networking and scheduling all influence economics of running AI workloads in production. Much innovation needed to make AI economically sustainable lies deeper in technology stack, where models, GPUs, storage and networks managed as programmable resources through common platform layer.
Together, findings point to broader shift in enterprise AI: realizing potential requires more than access to capable models. Organizations must build engineering practices, infrastructure and human oversight needed to operate agents responsibly at scale.
About Thoughtworks
Thoughtworks is a global technology consultancy that integrates design, engineering and AI to drive digital innovation. We are over 10,000 Thoughtworkers strong across 18 countries. For 30+ years, we've delivered extraordinary impact together with our clients by helping them solve complex business problems with technology as the differentiator.
