Industry Adoption

As enterprises expand use of generative AI and agents, unexpected cost overruns from token-based billing have become a growing headache. @IT's five-article roundup examines why token waste happens and what practical measures teams can take on the ground to keep AI spending within budget.

Specific Labs released Real-SWE, which tests coding agents on licensed private production codebases with business-critical tasks such as billing and tax logic, using each model's native harness. Claude Fable 5.1 via Claude Code leads at 38.8%, ahead of GPT-6 Astra (33.8%) and Gemini 3.8 Flash (31.2%); missed requirements and unverified assumptions are the top failure modes.

Simon Willison quotes Anthropic's Boris Cherny arguing that production code written by Claude should be held to a higher bar than human-written code. Anthropic layers on guardrails: extensive lint rules and tests, Claude-driven end-to-end tests, Claude-powered fuzzers run daily, automated code and security reviews, and automated refactoring. Without them, he warns, the result is hard to maintain.

Orbital has partnered with Reflex Aerospace to build a constellation of data centers in space. The deal gives the experienced German developer its first US customer, and Reflex has promised to supply its technology to Orbital exclusively.

NEC launched the Corporate AI Workforce Division on 1 August, an autonomous unit where AI fills every role from department head to frontline staff. The AI staff hold 1-on-1s and take engagement and stress surveys, with the results feeding back into workplace improvements.