Amplitude says its Wave product agent uses OpenAI models to turn product signals into improvements, reviewing analytics, session replays, customer feedback and code before preparing plans for coding agents such as Codex and tracking results. The company says work that once took about a month can now ship in one to two days; across customer launches, one in four went live the same day, half in under four days and two-thirds within a week. Wave uses GPT-6.1 Sol for deeper investigations, GPT-6 Luna for frequent scans and checks, and GPT-Image-2.5 Flare for mockups. On Amplitude’s homepage, a recommendation to personalize AI-agent suggestion cards was followed by a 160% increase in messages sent through those cards versus a control group after one day.
William Blair upgraded AMD to Outperform, forecasting revenue growth above 80% next year—more than twice AMD’s long-term model—and expecting data-center AI revenue to exceed server CPU revenue in the fourth quarter. The firm expects AMD’s Helios rack-scale system to ramp through 2027, while warning that custom chips, particularly at Meta, could challenge AMD’s Instinct accelerators.
Two reports highlight risks around AI agents and AI investing. Cequence says an online marketplace received 414,210 October visits falsely attributed to a popular consumer AI agent; all targeted sign-in and failed, indicating credential stuffing with stolen login details. That volume was about 1.5 times the genuine agent traffic it tracked across five industries that week. Cequence recommends judging sessions by behavior and adding checks for payments and account changes; in one case, behavioral analysis blocked 3.51 million unauthorized login attempts over 31 days, including attacks conventional bot tools missed. Separately, the Financial Times’ Lex column warns that the AI megatrade could leave investors with concentrated portfolios and that problems in one major investment vehicle or part of the interconnected system could spread.
Nvidia shares fell after reports of a second affair allegation involving CEO Jensen Huang and another executive, Kimberly Powell. The article says Nvidia is investigating whether to let Huang go, a prospect that has raised investor concerns. Shares closed Friday at $188.70, down from $239.24 on October 6, and traded at $173.73 after hours. At that level, Nvidia’s market value was about $3.93 trillion, roughly $300 billion below an earlier $4.2 trillion estimate. The report says the company’s future direction would be uncertain if Huang left.
Cerebras reported third-quarter revenue of $210.8 million under GAAP and $230.9 million on its core metric. Core cloud revenue rose 272% year over year to $144.9 million, and CEO Andrew Feldman expects revenue to triple in 2027. Its high-speed inference sales pipeline grew more than tenfold; Oracle plans to add Cerebras inference to Oracle Cloud Infrastructure in 2027, and Cerebras announced a deal with a top-three hyperscaler. Bilt plans to use its technology for services reaching eight million consumers and 50,000 businesses. Cerebras also reported a GAAP net loss of $161.2 million, a 20% gross margin and a negative 106% operating margin.
Other articles focus on applying AI with human expertise. PRS IN VIVO’s AI Pack Sprint pairs predictive AI with senior packaging specialists to guide packaging decisions and prepare retail-ready packs, with recommendations aimed at visual appeal, sales and brand awareness. A home-building commentary argues that AI can organize buyer feedback and competitor information but cannot replace local judgment: at High Point Development, an AI report understated a competing community’s available homes, while a decision to use light, whitewashed flooring in a New Jersey townhome model became the best-selling upgrade. The author urges builders to verify inventory, visit competitors and consult people who know the market.
For investors seeking Canadian exposure to AI infrastructure, one article points to hardware, power, engineering and software rather than a domestic chip giant. It identifies Celestica as the most direct hardware play, utilities Fortis, Hydro One, Emera and Capital Power as potential beneficiaries of data-center electricity demand, and WSP Global, Stantec and Atkinsrealis as possible participants in data-center and grid projects. It cautions that AI-related revenue is unclear for the engineering firms, many AI-linked stocks have fallen, and the figures are snapshots from November 11, 2026—not measures of each company’s AI revenue. In a separate investing discussion, Anthropic’s Claude recommends goals, diversification, low costs, broad index funds, discipline, tax awareness and periodic rebalancing. The author says AI can help analyze a portfolio but is not a licensed adviser or able to act for users, and recommends treating its output as a starting point for human review.
Key Takeaways
- Amplitude says its OpenAI-powered Wave agent can help teams ship improvements in one to two days; one homepage recommendation was followed by a 160% increase in messages through personalized suggestion cards versus a control group.
- Wave uses GPT-6.1 Sol, GPT-6 Luna and GPT-Image-2.5 Flare, and prepares plans for coding agents such as Codex.
- William Blair expects AMD’s data-center AI revenue to pass its server CPU business in the fourth quarter and forecasts revenue growth above 80% next year; custom-chip competition, especially at Meta, is a key risk.
- Cequence identified 414,210 fake AI-agent visits targeting an online marketplace’s sign-in page in October; behavioral analysis blocked 3.51 million unauthorized login attempts in one separate 31-day case.
- Nvidia shares fell amid reports of a second allegation involving Jensen Huang and another executive; the stock traded at $173.73 after hours, implying a market value of about $3.93 trillion.
- Cerebras reported $210.8 million in third-quarter GAAP revenue and a $161.2 million GAAP net loss; Oracle plans to add its inference technology to OCI in 2027, and CEO Andrew Feldman expects revenue to triple in 2027.
- PRS IN VIVO’s AI Pack Sprint combines predictive AI with senior packaging expertise; a home-building article likewise stresses verifying AI-generated market data with on-the-ground judgment.
- Canadian AI investment options discussed include Celestica, utilities Fortis, Hydro One, Emera and Capital Power, and engineering firms WSP Global, Stantec and Atkinsrealis; the article’s figures are dated November 11, 2026.
- The Financial Times’ Lex column warns that concentrated AI investments could make trouble in one part of the interconnected trade spread to others.
- Anthropic’s Claude recommends diversified, low-cost investing and periodic rebalancing, but the author cautions that AI is not a licensed adviser and its guidance needs human review.
Amplitude says its AI agent can help teams ship fixes in days
Amplitude says its product agent, Wave, scans session replays, analytics, customer feedback, and code to find and rank product improvements. It creates mockups and technical plans with acceptance criteria for coding agents such as Codex, then tracks results. Amplitude uses GPT-6.1 Sol for deep investigations, GPT-6 Luna for frequent scans and checks, and GPT-Image-2.5 Flare for mockups. The company says one in four customer launches went live the same day, half in under four days, and two in three within a week.
Amplitude uses OpenAI models to speed up product improvements
Amplitude says its Wave product agent can help teams ship product improvements in 1 to 2 days, down from about a month. Built mainly on OpenAI models, Wave reviews product data, session replays, customer feedback, and code, then prepares plans for coding agents such as Codex and measures the results. On Amplitude’s homepage, Wave suggested personalizing AI agent suggestion cards, and messages sent through those cards rose 160% compared with a control group after one day. Wave uses GPT-6.1 Sol for deep investigations, GPT-6 Luna for frequent lower-cost tasks, and GPT-Image-2.5 Flare to create mockups.
Amplitude uses OpenAI models to speed up product improvements
Amplitude says its Wave product agent can help teams ship product improvements in 1 to 2 days, compared with roughly a month before. Wave uses product data, session replays, customer feedback, and code to find opportunities, prepare implementation plans for coding agents such as Codex, and measure their impact. On Amplitude’s homepage, Wave recommended personalizing suggestion cards for users starting conversations with an AI agent. Messages sent through those cards increased 160% compared with a control group after one day. Wave uses GPT-6.1 Sol for deep investigations, GPT-6 Luna for frequent scans and checks, and GPT-Image-2.5 Flare for mockups.
William Blair sees AMD AI revenue surpassing server CPUs
William Blair upgraded AMD to Outperform and expects its revenue to grow more than 80% next year, over twice the company’s long-term model. The firm expects AMD’s data center AI revenue to pass its server CPU business in the fourth quarter. It also expects AMD’s Helios rack-scale system to ramp up through 2027. William Blair cited competition from custom chips, especially at Meta Platforms, as a key risk to AMD’s Instinct accelerators.
William Blair expects AMD AI revenue to pass server CPUs
William Blair expects AMD’s data center AI revenue to overtake its server CPU business in the fourth quarter, while its Helios rack-scale system ramps through 2027. The firm sees AMD’s revenue growing more than 80% next year, which is more than twice the company’s long-term model. It identified custom chips competing with AMD’s Instinct accelerators, especially at Meta Platforms, as a main risk.
Fake AI agent traffic hides a credential stuffing attack
In October, an online marketplace received 414,210 visits claiming to come from a popular consumer AI agent. Every visit targeted the sign-in page and failed, revealing a credential stuffing attack using stolen usernames and passwords. The fake agent generated about one and a half times as many visits as all genuine agents Cequence tracked across five industries that week. Cequence advises companies to judge sessions by their behavior and add extra checks around payments and account changes. In one case, behavioral analysis blocked 3.51 million unauthorized login attempts in 31 days, including traffic that conventional bot tools missed.
Nvidia shares fall as Jensen Huang faces a new allegation
Nvidia shares fell amid reports of a second affair allegation involving CEO Jensen Huang and another Nvidia executive, Kimberly Powell. The article says Nvidia is investigating whether to let Huang go, raising concerns among some investors. The stock closed Friday at $188.70, down from a high of $239.24 on October 6, and was at $173.73 in after-hours trading. At that price, Nvidia’s market value was about $3.93 trillion, roughly $300 billion below an earlier estimate of $4.2 trillion. Nvidia remains a major AI hardware company, but the article says its future direction is uncertain if Huang leaves.
AI Pack Sprint pairs fast analysis with packaging experts
PRS IN VIVO’s AI Pack Sprint combines predictive AI with guidance from senior packaging experts. The service aims to help businesses make packaging decisions and prepare packs for retail. Its team offers customized recommendations intended to improve visual appeal, sales and brand awareness. The company invites businesses to contact it to learn more.
Home builders need judgment alongside AI data
AI can organize buyer feedback and information about competing homes, but developers still need experienced judgment to check the facts and make pricing and product choices. At High Point Development, AI reported that a competing community had only six homes available, but substantially more inventory existed. The author also describes choosing light, whitewashed flooring for a New Jersey townhome model despite concerns from ownership, and says it became the best-selling upgrade. Builders should verify inventory, visit competing communities and speak with people who know the market before making decisions. The article says technology can show what buyers consider, while judgment helps determine what they will buy and pay for.
Cerebras expects revenue to triple as demand for fast AI grows
Cerebras reported third-quarter revenue of $210.8 million under GAAP and $230.9 million on its core metric. Core cloud revenue reached $144.9 million, up 272% from a year earlier, and CEO Andrew Feldman expects revenue to triple in 2027. The company said its high-speed inference sales pipeline grew more than ten times, and Oracle plans to add Cerebras inference to Oracle Cloud Infrastructure in 2027. Cerebras also announced a deal with a top-three hyperscaler and said Bilt will use its technology for services reaching eight million consumers and 50,000 businesses. The company reported a GAAP net loss of $161.2 million, with a GAAP gross margin of 20% and operating margin of negative 106%.
Canadian stocks offer several ways to invest in the AI buildout
Canada’s AI investment options include hardware, power, engineering and software companies, rather than a homegrown chip giant. Celestica is the most direct hardware play, while utilities such as Fortis, Hydro One, Emera and Capital Power could benefit from data centers’ demand for electricity. Engineering firms WSP Global, Stantec and Atkinsrealis may work on data center and grid projects, though the article says their AI-related revenue is unclear. The article notes that many AI-linked stocks have fallen, and that its figures are snapshots from Nov. 11, 2026, not a measure of each company’s AI revenue.
AI investing boom raises concerns about concentrated portfolios
The Financial Times’ Lex column warns that the AI megatrade may expose investors to too much portfolio concentration. The article describes AI as an interconnected system involving investors, developers and users. If one major investment vehicle or part of the system fails, the effects could spread and threaten the broader trade. It warns that too many assets may be held in too few baskets.
Claude’s investing advice echoes Boglehead principles, but needs a human check
The author asked Claude, Anthropic’s large language model, to share investing principles. Claude recommended clear goals, diversification, low costs, broad index funds, discipline, tax awareness and periodic rebalancing. It also said AI can help analyze a portfolio and explain choices, but it is not a licensed advisor and cannot act for a user. The author found similar answers from other AI models but warns readers not to follow AI advice blindly, suggesting they treat it as a starting point and discuss it with people, such as members of the Bogleheads forum.
Sources
- Amplitude says AI agents ship fixes in days
- How Amplitude ships product improvements in 1–2 days with OpenAI
- How Amplitude ships product improvements in 1–2 days with OpenAI
- AMD's AI Sales Set to Overtake Server CPUs, William Blair Says
- AMD's AI Sales Set to Overtake Server CPUs, William Blair Says
- Bots pose as AI agents, AI agents pose as humans, and defenses believe both
- Nvidia stock price plunges amid second Jensen Huang affair allegation — the AI company could be set to lose hundreds of billions in value
- AI Pack Sprint: combining AI speed with expert insight for more confident pack decisions
- AI can sort the data, but builders still need judgment to price and sell homes
- Cerebras bets speed can triple sales
- Canadian AI-buildout stocks: hardware, power, engineering and software
- The AI megatrade’s mounting portfolio concentration risk problem
- Is Claude a Boglehead? A Conversation With AI Revealed Surprisingly Familiar Investing Principles
Comments
Please log in to post a comment.