Glean, Salesforce, Snowflake, and Microsoft are shipping features that let AI agents invoke other agents as ordinary tool calls, creating what security researcher Niyikiza calls 'shadow delegation' wh…
An engineer argues that the Model Context Protocol (MCP) moves context but does not create it, highlighting that while MCP simplifies tool integration for AI agents, it fails to resolve cross-system f…
OpenAI and Anthropic both launched features in June and July 2025 that let their AI systems learn tasks by watching a user demonstrate them, rather than relying on prompts alone. OpenAI's Record & Rep…
OpenAI's Private Safety Processing preview, announced August 19, 2026, extends safety checks for Zero Data Retention customers to detect patterns across related interactions without giving OpenAI pers…
Glean CEO Arvind Jain said model routing is driven by cost concerns as frontier models like Opus and GPT become more expensive per token, with Glean claiming to be 4x more cost-effective than Claude C…
Cline, an AI coding assistant, is open-sourcing its evaluation framework for open-weight coding agents, revealing that its requests run 20–30% heavier on tokens than the most efficient harnesses. The …
Sequoia Capital's thesis that AI companies will sell work rather than software is flawed because using AI for a task trains the model to do it, commoditizing the service, according to a rebuttal by An…
Microsoft and Glean are racing to build an 'enterprise brain' that gives AI agents persistent memory, addressing the problem that agents forget everything between sessions. The technology aims to make…
Cloudflare Inc. released Cloudflare OS, an open-source agent workspace under the Apache 2.0 license on August 5, which the company's employees have used since May, allowing organizations to deploy a b…
Glean, the Work AI platform, is hiring a Machine Learning Engineer for Assistant Quality in San Francisco, CA, with a salary of $180k–205k/yr. The role focuses on improving the quality of Glean's AI A…
A recent study by Glean found employees save about 11 hours a week using AI but spend more than six hours checking outputs, fixing mistakes, and adding context, a cost the author argues is the biggest…
Developers building chatbots on top of large language models can improve user experience by using a 'clipboard' approach for citations, where the model outputs short keys that a postprocessor replaces…
Developers building LLM-powered chatbots can improve efficiency and user experience by using a 'clipboard' approach for citations, where the model outputs short placeholder keys that a postprocessor r…
Greenhouse CEO Daniel Chait says job seekers are paying $20 for AI tools to auto-apply to every job on the platform, creating an "AI doom loop" where both sides are unhappy. Recruiters face 254 applic…
Amazon is routing most Alexa+ queries away from Anthropic's expensive frontier models to cheaper alternatives, and Stanford's 2026 AI Index confirms a broader industry shift toward intelligence per do…
A friend of the author failed a senior AI engineer interview at an unnamed company after being asked what happens when a retriever returns a document contradicting the user's intent, a question that n…
NVIDIA Nemotron open models enable enterprises and nations to build customizable, trustworthy AI tailored to specific needs, with companies like Abridge, Glean, H Company, and Harvey already specializ…
Multiple AI startups, including Mercor, Anthropic, Sierra, Glean, and Gusto, are reporting accelerating revenue growth, reaching billion-dollar milestones in increasingly shorter timeframes. The compa…
AI companies are building context graphs to give agents structured memory across sessions, but schemas vary widely by use case—Perplexity's Brain uses an episodic model, Capital One's is evidence-cent…
AI strategist Sol Rashidi found that AI agents often demand more supervision than they save in time, spending up to 6.4 hours per week 'botsitting' them. She replaced two of her four AI agents with hu…