7 Oct - Cheaper Claude Haiku 5.5 - Anthropic launched Claude Haiku 5.5, its cheapest and fastest small model, at around 75% lower typical cost than Haiku 4.5. It is built for high-volume work like summaries and customer support, and runs on AWS, Google Cloud and Azure. Small AI models are getting cheap enough to run almost everywhere. More

8 Oct - Gemini Agents At Work - Google Cloud unveiled a Gemini agent for businesses that can work on tasks for hours or days, with its own Gmail, Calendar and Drive. It is in private preview. AI is being pitched as a coworker, not a tool. More

12 Oct - Microsoft's Own AI Chip - Microsoft announced Maia 300, the successor to Maia 200, with deployment slated for 2027. It plans to make its Maia family the primary system for running AI across its own products, and says Maia 300 can deliver up to 40% more tokens per megawatt than leading GPU-based systems. Maia 200 is already in production. More

8 Oct - TCS Earns From AI - TCS's annualised AI revenue reached $3.1 billion, over 10% of its total and up from $2.6 billion a quarter earlier. India's IT majors are now earning real money from AI. More

13 Oct - UPI Fee Delayed - NPCI pushed the 0.4% merchant fee on UPI payments above Rs 2,000 from 15 Oct to after Diwali, with a date to be notified. Merchants pay it, not the person paying. Person-to-person payments and payments under Rs 2,000 stay free of it. More

13 Oct - Firms Pull Back On AI - Meta scaled back plans to cut staff and use more AI agents after employees spent growing time fixing errors, Reuters reported. Microsoft and McKinsey are limiting employee use too. Rushed rollouts left workers cleaning up mistakes, so some firms are slowing down. More

5 Oct - India's AI Safety Rules - India is working on its own AI safety framework, built around its languages and scale. Experts disagree on whether voluntary commitments are enough or binding rules on transparency and incident reporting are needed. India is deciding how tightly to regulate AI. More

13 Oct - Scientists Want AI Oversight - About 400 scientists, including Geoffrey Hinton and Yoshua Bengio, signed an open letter calling for an international body to set AI containment standards and track containment failures. They want every serious security incident reported and eventually made public. Their case is that transparency is needed for learning and accountability. More

14 Oct - India's Quantum Computer - IBM plans to install what it calls India's largest quantum computer in Amaravati in early 2027, with TCS as partner. The site will also help Indian companies prepare for post-quantum cryptography. IBM says real-world quantum uses are still 3-5 years away. More

Project KnowHow

Ideas, decisions and lessons from building with AI, with practical resources for non-technical users

Latest post: One Set of Rules