Thoughtworks Technology Podcast Inside AI/works™: An agentic development platform
71 snips
Feb 19, 2026 Shodhan Sheth, head of modernization who turns legacy systems into modern designs using AI tools, and Bharani Subramaniam, CTO focused on platform engineering and AI governance, discuss AI/works™. They explore building a repeatable AI platform, bridging code-to-spec and spec-to-code, legacy-system discoverability, guardrails and observability for enterprise AI, and agentic workflows and incremental modernization.
AI Snips
Chapters
Transcript
Episode notes
Platform As Control Plane For Safety
- AI/works centers on an AI gateway for token control, safety guardrails and centralized observability.
- Governance, lineage and spend controls form a core part of making agentic engineering enterprise-safe.
Engineering Rigor Around Generative AI
- ThoughtWorks frames AI/works as engineering around generative AI, not a pure LLM product.
- Rigor from decades of engineering is paired with AI benefits to reduce hallucination and increase fidelity.
Context Library Powers Forward Engineering
- The platform's unique value is a curated context library and a 'hive mind' of past solutions to enrich client-specific context.
- Relevance curation and matchmaking of historical knowledge to current problems are the hard problems the runtime must solve.