HackerNews

What Is an AI Agent Management Platform (AMP)?

Article URL: https://www.execlave.com/blog/what-is-an-ai-agent-management-platform Comments URL: https://news.ycombinator.com/item?id=48657599 Points: 1 # Comments: 0
HackerNews

AI is learning to read the room

Article URL: https://spectrum.ieee.org/emotion-ai-context Comments URL: https://news.ycombinator.com/item?id=48657600 Points: 1 # Comments: 0
Reddit

DeepSeek doesn't like Pokemon [Showdown] for whatever reason.

I've been feeding stuff from a fan fic I am writing. [Crossover of PBD's The Academy and Left Behind: The Kids]. I don't care what you think of it. I will post the material for reference. It mentions a lot of inflammatory material [e.g. El Mozote, imperialism, reeducation camp], but the prompt was not accepted as is. When I got rid of Pokemon , it went through. Swapping Pokemon with Pokémon gets it accepted. [link] [comments...
Reddit

THE EX-GOOGLE CHARACTER AI ERA IS EVOLVING

https://reddit.com/link/1ue5le1/video/ujvu26e6c69h1/player Text-based AI characters were the first wave. Now Mel AI is showing what the next wave could look like: AI characters you can video chat with in real time. The demo shows characters that can talk, lip sync, react with facial expressions, and respond to camera context instead of just replying in a chat box. Character AI proved the demand for AI characters. Mel AI is pushing the format from text-native to video-native. ...
Reddit

South Korean AI app went viral for AI characters that can talk, react, and respond to camera context

https://reddit.com/link/1ue5o4i/video/ozv6sohrc69h1/player Instead of only texting AI characters, the app shows characters that can talk through voice, lip sync, react with facial expressions, and respond to camera context during the conversation. The demo suggests a shift from text-based character AI toward video-native AI characters, where the interaction feels closer to a live call than a chatbot. For ML developers, the interesting part is the underlying stack: vision, speech, memory, ...
Reddit

¿Estamos pasando de "herramientas de IA" a "compañeros de trabajo de IA"?

I’m starting to think the real shift won’t be just better chatbots or coding agents, but workflows where humans supervise many AI agents at once. The interesting question is: what does "work" look like when one human can coordinate 5, 10, or 50 specialized agents? Maybe the future of work is not "AI replaces humans", but "humans become the loop that verifies, redirects, and gives judgment to AI systems." Curious how people here see it: will human-in-the-loop w...
Reddit

IONS: A reasoning graph that stores claims, evidence, and reasoning paths outside the LLM

I’ve been experimenting with an open source alternative approach to AI memory and reasoning called IONS. The basic idea is that instead of storing all knowledge inside model weights, knowledge is represented as a graph of evidence backed claims called Cognitive Building Blocks (CBBs). Each CBB contains: \-A claim \-Supporting evidence \-Confidence metadata \-Provenance \-Relationships to other claims Relationships are typed: \-supports \-causes \-contradicts \-depends\_on \-deriv...
Reddit

A significant portion of the remaining training data for AI is located on magnetic tapes stored in warehouses.

I have been learning about the shortage of AI training data and one aspect that nobody considers is that much of the potential training data that can be used is not stored in any database system but rather on the old magnetic tapes that have been stored in climate controlled lockers for decades now. The 80s through the 2000s saw all major businesses, government offices, hospitals, television stations, and laboratories include backup of everything on tapes. Most of this data has neither been digi...
HackerNews

Is working in AI training data a waste of time?

Article URL: https://joinhandshake.com/ai/opportunities/generalist-bachelor-s2/ Comments URL: https://news.ycombinator.com/item?id=48654916 Points: 1 # Comments: 1
HackerNews

Ask HN: How do you test AI-generated code?

When AI generates code, I first instruct the model to find, fix, and verify any issues. After that, I start the server and test whether it actually works from the user’s perspective. What I’m looking for is a workflow where issues are received, fixed, tested, and deployed—but it seems that current AI agents aren’t very good at performing browser tests from the user’s perspective. I’ve tried using the built-in browsers in Codex and Cursor, but they often only checked whether the page loaded. In t...
HackerNews

New AI tools in darktable 5.6

Article URL: https://www.darktable.org/2026/06/meet-darktable-5.6-ai-tools/ Comments URL: https://news.ycombinator.com/item?id=48655147 Points: 1 # Comments: 0
HackerNews

AI Steps Off the Screen

Article URL: https://epics.tech/posts/2026-06-23-ai-steps-off-the-screen/ Comments URL: https://news.ycombinator.com/item?id=48655738 Points: 2 # Comments: 0
HackerNews

Cisco AI Defense Skill Scanner

Article URL: https://github.com/cisco-ai-defense/skill-scanner Comments URL: https://news.ycombinator.com/item?id=48656076 Points: 2 # Comments: 0