https://www.amazon.science/tag/amazon-agi-lab Amazon AGI Lab en-US Wed, 27 May 2026 16:34:34 GMT https://www.amazon.science/blog/introducing-the-perception-agent-harness-annotation-and-verification-open-source Today we're announcing the open-source release of the first two primitives for our perception agent harness: annotation and verification. Wed, 27 May 2026 16:34:34 GMT https://www.amazon.science/blog/introducing-the-perception-agent-harness-annotation-and-verification-open-source https://www.amazon.science/blog/portable-reasoning-releasing-text-bound-intelligence-into-agentic-interaction Large language models today can solve algebra, pass academic benchmarks, and generate highly structured chain-of-thought explanations. In text-only settings, they often feel startlingly intelligent — methodical, articulate, even strategic. But place those models inside an interactive environment — ask them to click buttons, scroll pages, fill out forms, and submit answers — and their behavior changes. Their careful reasoning falters. They guess where they once deduced. They adhere to templates and produce limited procedural narration: stating what they see and what they will click next, without first forming a structured plan and acting in accordance with plan. It’s as if part of their intelligence has quietly gone offline the moment the cursor appears. Mon, 20 Apr 2026 20:36:51 GMT https://www.amazon.science/blog/portable-reasoning-releasing-text-bound-intelligence-into-agentic-interaction https://www.amazon.science/blog/a-practical-recipe-for-training-computer-use-agents-with-rl LLMs are getting pretty good at talking. Getting them to reliably act on a computer — clicking, typing, and navigating real websites to achieve a goal — is a different beast. Thu, 16 Apr 2026 13:00:00 GMT https://www.amazon.science/blog/a-practical-recipe-for-training-computer-use-agents-with-rl https://www.amazon.science/blog/how-agentic-ai-helps-heal-the-systems-we-cant-replace By learning the idiosyncrasies of accumulated layers of legacy systems, AI agents can preserve institutional knowledge and provide a unified interface to a range of services. Mon, 16 Mar 2026 13:00:00 GMT https://www.amazon.science/blog/how-agentic-ai-helps-heal-the-systems-we-cant-replace https://www.amazon.science/blog/designing-ai-interfaces-that-align-with-how-people-actually-work "If I were to ding the field right now on one thing, it is that there has been a massive lack of creativity on how people interface with these increasingly smart LLMs and agents." Mon, 23 Feb 2026 17:03:55 GMT https://www.amazon.science/blog/designing-ai-interfaces-that-align-with-how-people-actually-work https://www.amazon.science/blog/the-unseen-work-of-building-reliable-ai-agents "Reinforcement learning gyms" train agents on the many low-level tasks that they must chain together to execute customer requests. Wed, 07 Jan 2026 17:04:36 GMT https://www.amazon.science/blog/the-unseen-work-of-building-reliable-ai-agents https://www.amazon.science/blog/new-making-a-mind-podcast-explores-science-of-intelligence Making a Mind is hosted by Dr. Danielle Perszyk, cognitive scientist and head of the human-computer interaction (HCI) team at Amazon’s AGI Lab. The podcast explores the science of intelligence through conversations with leading AI researchers, examining the scientific breakthroughs required to realize useful general intelligence. Wed, 07 Jan 2026 17:04:25 GMT https://www.amazon.science/blog/new-making-a-mind-podcast-explores-science-of-intelligence https://www.amazon.science/blog/scaling-agent-reasoning-with-variations-coherence-in-trajectory-rollouts Web agents that help customers book travel, shop online, and manage tasks still struggle with reliability — misclicking items, misreading intent, and drifting mid-task. The gap isn't just data; it's how agents learn to reason. Part of the problem is limited high quality training data, but the deeper issue is what the data fails to represent. Agents are rarely trained on structured supervision that shows how reasoning unfolds in context over time. Mon, 22 Dec 2025 19:39:16 GMT https://www.amazon.science/blog/scaling-agent-reasoning-with-variations-coherence-in-trajectory-rollouts https://www.amazon.science/blog/the-wide-and-widening-world-of-tools-and-tool-calling Even in a world where foundation model have made the manifold applications of artificial intelligence (AI) seemingly ubiquitous, the recent rise of agentic AI and the resultant proliferation of tool-using agents represent a significant step forward. Wed, 17 Dec 2025 16:15:11 GMT https://www.amazon.science/blog/the-wide-and-widening-world-of-tools-and-tool-calling https://www.amazon.science/blog/amazon-nova-act-service For years, we've heard bold claims about how AI will transform work—creating visions of virtual assistants that magically automate away digital drudgery. Yet organizations experimenting with AI agents have encountered a sobering reality: most agents only work sometimes, failing to deliver the reliability needed for mission-critical workflows. Tue, 02 Dec 2025 16:02:11 GMT https://www.amazon.science/blog/amazon-nova-act-service