The AI experiment is over. 2027 wants the receipts

The receipts are getting specific

Kuken Industries, a Japanese cooling tower manufacturer, reduced the time required to search its 70-year archive of engineering drawings from nearly an hour to 4.5 minutes using a tool built on the OpenAI API, developed with partner Recursive, a Japanese AI startup. The company expects the new search tool to reduce annual hours spent searching for drawings by approximately 90%, and engineers can search by entering a drawing number or part number, or by uploading an image or PDF. Newer engineers can now handle a wider range of jobs independently due to the search tool, the company says. m:…6536

Kuken is now piloting an additional AI tool to automate the review of partner-submitted drawings for discrepancies, and in prototype testing approximately 90% of the issues flagged by the tool were genuine discrepancies. Snowflake's internal audit team tells a similar story at a different scale: it reduced fieldwork time for evidence collection and report production by 50%, saved over 3,500 hours in 2026 through AI-driven audit processes, and says control owners across the business saved over 1,500 hours in 2026 due to audit automation. The team shifted from random statistical sampling to 100% full population assurance for specific use cases, using a human-in-the-loop model where auditors retain final accountability for all conclusions. m:…6536 m:…2776

Shipping twenty times more, with the same team

CodeRabbit, a code review platform used by over 20,000 organizations and 150,000 open-source projects, says it has shipped 20 times more software over the last six months without adding engineers, using OpenAI's Codex and reasoning models to automate code reviews and security. Its CEO Harjot Gill estimated that GPT-6 Astra caught 50% of the issues in the company's internal benchmark, and Gill pushed 2,500 to 3,000 pull requests in the 30 days before the interview. The company uses GPT-6 Luna for subagents and GPT-6 Astra for planning and orchestration, and has expanded agents beyond engineering, managing marketing tasks including Google Ads through pull requests. m:…6535

The security numbers are worth reading carefully. CodeRabbit's latest offering, CodeRabbit Secure, uses Daybreak with Codex models to find and fix vulnerabilities, and the company says it found over 43,000 security vulnerabilities across 13,000 deep scans using Codex models. These are company-reported figures, and they describe a promotional scan as well as production use, so a buyer should treat them as a claim to verify rather than a fact. But the direction is clear: the company is using agents to multiply output without multiplying headcount, and it is selling that same multiplication to its customers. m:…6535

The spending is now visible

IBM's Apptio, an IBM company, has introduced new capabilities to help enterprises track and manage AI spending alongside broader technology costs, including native AI cost and usage connectors and new AI Value and ROI tools. The update responds to a gap IBM measured directly: 85% of surveyed technology executives reported that they lack full visibility into real-time AI spending. IBM Cloudability now supports AI application costing, token usage and commitment recommendations for hyperscaler AI services, and IBM is introducing Model Context Protocol servers to connect financial, operational and portfolio data with external AI tools. m:…1733

The Apptio AI Value and ROI capabilities are designed to connect technology investments with measurable business outcomes, IBM says, and enhanced Kubernetes network cost visibility in IBM Kubecost covers AWS NAT Gateway ingress and egress, cross-zone and cross-region traffic. The subtext is that AI spending has grown large enough to need the same cost discipline as cloud spending, and that most enterprises do not yet have it. A budget line that nobody can see in real time is a budget line that cannot be governed, and governance is exactly what the next phase demands. m:…1733

2027, according to SAS

SAS executives forecast that 2027 will mark a transition for enterprise AI, moving from unchecked experimentation to a period defined by accountability and measurable business value. In 2027, enterprises will demand proof that AI delivers business value, SAS says, and CIOs must reimagine enterprise infrastructure to govern autonomous digital agents rather than human users. Companies will need clear rules for what an AI agent can decide independently versus what requires human involvement, and AI agents will begin making consequential financial decisions, such as transferring money on behalf of human principals. m:…4185

SAS also predicts a productivity paradox: organizations will struggle to demonstrate business performance improvements despite producing higher volumes of AI-generated content, and enterprises will shift focus from AI consumption costs to the cost of delivering trusted business decisions. The economics of the AI boom will face a reset due to high valuations and the cost of data center expansion, SAS says, and hyperscalers will emerge stronger from the reset due to diversified businesses and financial flexibility. The forecast is self-interested, since SAS sells analytics and governance, but it matches the receipts in this story: the teams with numbers are the ones that will keep their budgets. m:…4185