Thursday, November 5, 2026
A consultant who advises one-person firms on generative AI has dropped his own advice to rotate between free chatbots. His new pitch is a workflow audit that marks where a person has to check the machine.
Google, ChatGPT and Perplexity now answer many questions before a customer reaches a company website. An SEO agency co-founder explains his five-part method, and the platforms' own rules show where it fits.
AI researchers, forecasting platforms and lab chiefs have put dates on artificial general intelligence ranging from 2026 to the 2040s, and the estimates keep moving. Rishabh Singh, founder of Nova Reach and Nova Infinity and co-founder and CTO of JoinMeIn, argues in a September essay that the most useful answer to the AGI question is not a date but a way to build.
Meta released Muse Wonder and Muse Wonder Full on Monday, its first frontier AI models since it rebuilt its AI effort 14 months ago. Muse Wonder costs $4 per million input tokens and $20 per million output tokens on the Meta Model API; Muse Wonder Full is sold only inside Muse Code plans of $50 and $200 a month. Meta's shares closed up 3.4 per cent. The benchmark claims are Meta's, and independent testing has only just begun.
Marvell Technology raised its fiscal 2028 revenue forecast to about $20 billion at its investor day on Tuesday, above the $18.2 billion analysts expected, and introduced a first-ever fiscal 2031 revenue range of $70 billion to $90 billion. From fiscal 2026 revenue of $8.2 billion, the targets imply compound annual growth above 50 percent for five straight years, powered by custom AI chips, networking and optical interconnects for data centers.
C.H. Robinson agreed on Sunday to buy RXO, the truck broker spun out of XPO in 2022, for about $5.8 billion including debt. Its own slides value the deal at 13.2 times RXO's expected earnings before interest, tax, depreciation and amortisation, but only after $300 million of annual cost savings. Without them the multiple is about 42. C.H. Robinson's shares fell 10.8 per cent on Monday.
Shareholders in Blue Owl Technology Income Corp. asked to sell 39 per cent of the fund back in the third quarter, about $1.1 billion. The fund will buy back 5 per cent, so each request will be filled at about 13 per cent. At Blue Owl's larger credit fund, requests fell for a second quarter, to 16.8 per cent of shares. Across the two, investors asked for $4.2 billion.
The world's chipmakers sold $159.7 billion of semiconductors in August, 144 per cent more than a year earlier, the Semiconductor Industry Association said on Monday, and passed $1 trillion for the year to date. In February the industry's forecast for all of 2026 was roughly $1 trillion. It is now more than $1.6 trillion. Micron's results show how much of the jump is price.
Schneider Electric agreed on Sunday to buy PTC, the Boston maker of design and product-lifecycle software, for $205 a share in cash, valuing its equity at about $22.6 billion. That is 42.3 per cent above Friday's close, and 5 per cent below PTC's peak close in August 2025. The French group will borrow most of the money, sell up to €6 billion of new shares and pause its buyback for two years. Its shares fell about 10 per cent in Paris.
Contracts obtained through litigation describe four frontier labs each holding Defense Department agreements worth up to $200m, with bidirectional data exchange including model benchmarks and engineers embedded with the military. Procurement usually buys a product. This buys a relationship in both directions.
Corporate AI budgets are shifting out of innovation labs and into core operating lines, a change that is quietly redrawing how companies account for technology itself.
The most consequential AI deployments of the year are not chatbots. They are agents reconciling invoices, chasing documents and closing tickets nobody wanted to touch.
After years of scraping first and negotiating later, AI developers are building formal licensing pipelines, and content owners are learning what their archives are worth.
Away from enterprise pilots and keynote demos, small firms are putting generative tools to work on the least glamorous parts of operating a business.
Procurement departments, tired of negotiating novel terms for every AI vendor, are converging on shared contract language for data handling, evaluation and exit.
In the absence of comprehensive federal legislation, statehouses have produced a growing body of AI law, and companies are complying with the strictest version everywhere.
The frontier gets the headlines, but the volume is moving to compact models that run cheaply, privately and close to the work.
Voluntary vagueness about where AI touches customers is becoming a liability. A simple disclosure norm would serve companies better than the silence they are defending.