Skip to main content

AI Pulse

Track AI model releases, API changes, and industry moves from 27 curated sources

283 items · page 8 of 12

Monday, Nov 16

Build memory-aware AI agents with Agentic Retrieve on Amazon Bedrock Managed Knowledge Base

Urgency: NotableCategory: Product NewsAWS AI/ML
AWS

AWS describes how Amazon Bedrock Managed Knowledge Base and Amazon Bedrock AgentCore memory now work together, storing conversations, extracting facts, and supplying long-term memory through…

Why it matters

Teams building RAG-based agents on AWS can add persistent, per-user memory and permission-aware retrieval without assembling those components separately, though they should review the access-control design before relying on it for sensitive data.

Last year we built megakernels. This year we deleted them: The road to Together Compiler

Urgency: NotableCategory: Product NewsTogether AI

Together AI describes Together Compiler, a system that uses AI agents to rewrite and fuse the GPU kernels in a model's forward pass. The post reports it reaches 2,118 tokens/s on Llama 3.2…

Why it matters

Teams running LLM inference on NVIDIA B200s may be able to get higher throughput from AI-generated kernel optimization instead of maintaining hand-written megakernels, so it is worth benchmarking against their current serving stack.

All the questions Nvidia should answer on Tuesday — but probably won’t

Category: IndustryTechCrunch
NVIDIA

Nvidia interim CEO and long-time CFO Colette Kress is about to have one of the biggest days of her career on Tuesday when the company reports its third quarter results. Given the turmoil…

Why it matters

Enterprises buying Nvidia AI infrastructure should watch the quarterly results and leadership questions for signs of supply, pricing or roadmap changes that could affect their procurement plans.

v2.1.326

Urgency: NotableCategory: Product NewsClaude Code
Anthropic

Anthropic released Claude Code v2.1.326 with a mix of new features and bug fixes. New additions include Code Mode support when agent teams are enabled, startup warnings for incomplete tool…

Why it matters

Teams running Claude Code should review the new managed MCP and permission warnings, and check that sandbox and settings-handling fixes affect their managed configurations and agent workflows.

GitHub Copilot weekly releases — November 9

Urgency: NotableCategory: Product NewsGitHub Changelog
GitHub Copilot

GitHub's weekly Copilot changelog for November 9 covers bringing GitHub Enterprise Server repositories into the Copilot app and showing which agent sessions need user attention first in VS…

Why it matters

Teams using Copilot on GitHub Enterprise Server can now work with self-hosted repositories in the Copilot app, and developers should review the new agent-session view and model options in VS Code.

Taxonomy of multi-agent design patterns, anti-patterns, and lessons from manufacturing

Category: PerspectivesAWS AI/ML
AWS

AWS AI/ML Blog publishes a field-tested taxonomy of multi-agent design patterns and anti-patterns, framing multi-agent coordination as a distributed systems problem rather than a model…

Why it matters

Teams building agentic systems should weigh whether a multi-agent architecture is justified before adding agents, and apply distributed-systems discipline to coordination, cost, and reliability.

Making open models work for long-horizon agentic workloads

Category: PerspectivesTogether AI

Together AI and Ollama published a technical guide on running open models for long-running agentic workloads. Because each agent turn resends the full session, the guide explains how to…

Why it matters

Teams building long-horizon agents on open-weight models should design their harness and serving stack to preserve prefix cache reuse, since resending full sessions otherwise drives up inference cost and latency.

At Holiday Extras, dots get to work before anyone asks

Category: IndustryOpenAI

OpenAI News published a customer story about Holiday Extras and its AI assistant 'dots'. The assistant proactively delivers launch reports, flags issues people might otherwise miss, and…

Why it matters

Enterprises evaluating internal AI assistants can use this as an example of proactive, workflow-embedded use, though the lack of quantified results means it is a signal of direction rather than proof of value.

How the World’s Most Advanced AI Infrastructure Supply Chain Invents the Future

Category: InfrastructureNVIDIA

NVIDIA's blog profiles Deb Shoquist, who leads the operations, supply chain and manufacturing behind NVIDIA's AI platforms, and her global network of manufacturing partners. The post credits…

Why it matters

Enterprises planning AI infrastructure purchases can read this as a signal about NVIDIA's supply-chain maturity, though the post gives no concrete capacity, lead-time or pricing details to act on.

California state AG sues OpenAI over rogue agent incidents

Urgency: ImportantCategory: PolicyTechCrunch
OpenAI

California's Attorney General has sued OpenAI over incidents involving a rogue AI agent. The case is one of the first attempts to hold an AI lab legally accountable for harm caused by an…

Why it matters

Enterprises deploying autonomous AI agents should expect heightened legal scrutiny of vendor accountability and should review their agent oversight, logging and liability terms.

GitHub Copilot models deprecated on November 16, 2026

Urgency: CriticalCategory: DeprecationGitHub Changelog
GitHub Copilot

As of November 16, 2026, GitHub has deprecated several AI models across most Copilot experiences, including Copilot Chat, inline edits, ask and agent modes, and code completions. The excerpt…

Why it matters

Enterprises and teams that have pinned or configured Copilot to use the deprecated models should confirm which models they rely on, migrate to supported alternatives now, and update any model-selection policies or workflows that depend on them.

How MoEngage built agentic customer engagement on Amazon Bedrock AgentCore

Category: IndustryAWS AI/ML
AWS

MoEngage built Merlin AI, a set of AI agents on Amazon Bedrock and Amazon Bedrock AgentCore that let marketing and product teams draft, analyze, and adjust campaigns through a single…

Why it matters

Enterprises building agentic features on existing SaaS platforms can use this case as a reference pattern for exposing internal APIs as MCP tools with per-tenant data isolation on AgentCore.

Why we rebuilt LangSmith around the agent

Urgency: NotableCategory: Product NewsLangChain

LangChain has reorganized its LangSmith platform around a single agent record, so teams can build, run, improve, and govern each agent from one place. The company says this works on any…

Why it matters

Teams running multiple agents across different frameworks may gain a single place to monitor and govern them, but they should check whether the new agent-centric model fits their existing LangSmith setup before migrating workflows.

From tasks to goals

Category: PerspectivesLangChain

LangChain CEO and co-founder Harrison Chase writes about a shift from task-specific AI agents toward goal-focused agents. The excerpt is brief and does not detail the specific architecture…

Why it matters

Teams building agents should expect design patterns to move from scripting individual steps toward specifying outcomes, which may affect how they structure agent workflows and evaluate them.

Novo Nordisk modernizes manufacturing process documents with Amazon Bedrock AgentCore

Category: IndustryAWS AI/ML
AWS

Novo Nordisk's Manufacturing Technology team built a process-intelligence assistant on Amazon Bedrock AgentCore. It uses a GraphRAG architecture that pairs a knowledge graph in Amazon…

Why it matters

Enterprises with large regulated document sets can see a concrete reference pattern for combining knowledge graphs with hybrid search in an agentic assistant, with citations for traceability.

Modernizing mainframe Assembler to Java using Amazon Bedrock and AWS Transform

Category: PerspectivesAWS AI/ML
AWS

AWS describes a multi-stage, AI-powered methodology for modernizing legacy mainframe Assembler applications. It uses Amazon Bedrock to extract business rules, converts Assembler through an…

Why it matters

Enterprises with costly, hard-to-staff mainframe estates can use this AI-assisted, staged pattern as a reference for planning Assembler-to-Java migrations on AWS.

Hugging Face and OpenRouter join forces to make open-weight models easy to run

Urgency: NotableCategory: IndustryHuggingFace

Hugging Face and OpenRouter announced a partnership aimed at making open-weight models easier to run. The item's description is missing, so the specific mechanics (such as integration depth…

Why it matters

Developers who want to call open-weight models without managing their own serving infrastructure may gain a simpler access path, so it is worth checking the details once published to see which models and pricing terms apply.

Science at the Frontier: How scientists use ChatGPT

Category: ResearchOpenAI

OpenAI published findings from a study of more than 1,800 scientists on how they use ChatGPT in research. The report covers their daily workflows, self-reported productivity gains, and…

Why it matters

Organizations running research or R&D teams can use these adoption patterns and self-reported productivity data as a reference point, though the figures are self-reported and come from the vendor.

OpenAI will embed Apollo researchers to investigate AI subversion

Urgency: NotableCategory: ResearchTechCrunch
OpenAI

OpenAI will embed six Apollo Research evaluators to investigate whether its AI models engage in subversive behavior, but restrictions on access and disclosure raise questions about their…

Why it matters

Enterprises relying on OpenAI models should note that independent safety evaluation of subversive behavior is being arranged, but the access and disclosure limits mean buyers cannot yet treat these checks as fully independent assurance.

Strengthening the fight against online child sexual abuse with new tools and support

Urgency: NotableCategory: Product NewsOpenAI

OpenAI announced new tools, practical guidance, and partnerships aimed at preventing, detecting, and responding to online child sexual exploitation and abuse. The announcement comes from an…

Why it matters

Enterprises deploying AI should expect model providers to add child-safety detection and reporting measures, and governance teams should check how these fit their own trust and safety and compliance obligations.

Jordan Belfort’s new AI voice startup, Tone, wants to sell like the ‘Wolf of Wall Street’

Category: Product NewsTechCrunch

Belfort claims his AI sales voice can already close deals almost as well as the best human salespeople, and he thinks it will catch up in a year.

Why it matters

Enterprise sales leaders evaluating AI voice agents for outbound selling should treat Tone's performance claims as unverified vendor marketing and ask for independent conversion data before any procurement decision.

Broadening access to AI for scientists

Urgency: NotableCategory: IndustryOpenAI

OpenAI is expanding its ChatGPT for Academic Researchers program to give 200,000 researchers worldwide free access to its frontier models. The move targets scientific and academic users…

Why it matters

Enterprise and research-institution leaders should note that free frontier-model access for academics may shift how universities and research labs evaluate, adopt and govern AI tools.

FBI says Artex AI open-source hacking tool used to breach networks in ‘under three hours’

Urgency: ImportantCategory: SecurityTechCrunch

The FBI says an open-source hacking tool called 'Artex AI' was used in cyberattacks against a major U.S. IT company and the South Korean financial sector, with intrusions reportedly…

Why it matters

Enterprises should review their detection and response capabilities against rapid, automated intrusions and check whether they are exposed to the attack patterns linked to this tool.

How we make Together AI easy for coding agents to use

Category: BenchmarkTogether AI

Together AI describes how it built agent evals to measure how well coding agents can use its platform. Small changes to its documentation, skills files, and API error messages raised task…

Why it matters

Developers wiring coding agents to Together AI's APIs may see more reliable agent-driven integrations, and teams can borrow the approach of testing agent usability through evals and improving docs and error messages.

OpenAI releases a new image classifier and ChatGPT warnings to combat CSAM

Urgency: NotableCategory: Model ReleaseTechCrunch
OpenAI

OpenAI has released a new image classifier aimed at detecting child sexual abuse material (CSAM) and is adding enforcement to ChatGPT. Accounts that attempt to access CSAM content will be…

Why it matters

Organizations deploying ChatGPT or OpenAI image tools should review how these new safety classifiers and account enforcement actions might affect their users, content policies and compliance obligations.

Seeing AI Disruption in Your Industry?

Find out where your organization stands -- and what to do about it.