Pomerado launches MCP that turns websites into agent APIs
The YC F26 startup says hosted, self-healing integrations beat browser agents on speed and cost, with sign-in and 2FA handled for you.
Stories per day · 30 days
17 RSS
Relatedbenchmarks16benchmark11open-weights9multimodal8reinforcement-learning8mixture-of-experts7openai7tool-use6mcp5meta5nvidia5qwen5evaluation4fine-tuning4long-horizon4nemotron4on-device4open-source4
341 stories, newest first · one card per story
Stories tagged #agents
Meta releases Muse Ember, a lightweight model for agent work
Meta prices Ember at 10 cents per million input tokens and pairs it with Muse Wonder for volume.
1
Cogent uses OpenAI models to automate enterprise vulnerability fixes
Cogent reports a Fortune 1000 automaker cut mean remediation time 75% with its OpenAI-powered agents.
Veritas harness auto-generates agent-task verifiers that beat seven alternatives
With the model fixed, generated verifiers matched official verdicts better than seven harnesses on all four execution-scored benchmarks.
GOPR-CAD builds CAD models through progressive operation graphs
On a 300-part MHCAD benchmark, the training-free pipeline reaches 82.7% Structural Yield.
SurgKnowBench and SurgAgentGym test generalist models on surgery scenes
Agentic tool use mainly helps perception-heavy tasks; localization and expert-style assessment stay weak.
Rivet Launches Agent Scaling and Orchestration on S3
YC W23 startup Rivet enables scaling and orchestrating AI agents using S3 storage.
Signed Manifests Protect Requirements in LLM Agent Handoffs
Signed HMAC-SHA256 manifests reduce unsafe commits in multi-agent handoffs from 69.4% to 9.0% across 1440 trajectories.
ServerBench benchmarks agent-designed LLM serving systems
Claude Sonnet 5.5 achieves 4.87 times vLLM reference on Gemma E4B via agentic design.
Hacker News discusses AI agentic setups and tools
Users share current subscriptions to Claude and ChatGPT, models via OpenRouter, and tools like Pi and custom orchestrators for daily agent workflows.
ArcRAG enables construction-agnostic graph retrieval
Query-local graph interface induction allows fixed retriever to work on differently built knowledge graphs.
Coca-Cola: Latent context compression for LLM agents
Coca-Cola compresses tool observations into latent tokens and embeds them into the input embedding space of LLM agents.
Apple CEO says the company does not want to ship AI for its own sake
Ternus, CEO since early September, said a delayed, more powerful Siri launched this summer and good AI should feel like ordinary software.
Hugging Face paper introduces NorthCUA for scaling computer-use agents
NorthCUA turbocharges computer-use agents through an experience flywheel and staged learning achieving state-of-the-art results among open-source models.
Robust memory manager rejects poisoned LLM agent inputs
One-shot fine-tuning on synthesized episodes with changes and attacks more than doubles evidence-conditioned accuracy against poisoning and outdated facts.
RoboArchitect automates robotic agent harness with persistent memory
Coding model iteratively evolves harness around frozen policies raising success rates 3-4x on unseen scenes and tasks.
Spawn game world is modified live by Fable-class agents
Willison says a phone prompt created a castle and a rideable dragon; agent spend is put above $3 million.
Ben Hylak jokes every product should be a headless MCP agent
He carves out his own product, saying customers love its interface and the company.
KARMA updates advertising-agent knowledge without changing model weights
Authors report average hold-out score rising from 30.19 to 39.15 across 18 domain–model settings.
EditAgent expands tools and verification for agentic image editing
Experts preferred parallel listwise judging over sequential refinement in 78.5% of decided comparisons.
Supabase founders discuss agent-friendly backend features on podcast
The episode covers a Docker-free local runtime, Git-based schemas, private-alpha compute, and scale-out work including Turso.
Bellini Capital CISO says faster patching cannot close AI exposure
Brian Blakley argues attackers turn flaws into exploits faster than defenders can test, approve, and schedule patches.
WeVisCut trains a 9B agent for narration-to-video editing
Trained on more than 130,000 examples, the 9B policy matches Claude Opus 5 while using nearly 33 times fewer tokens.
GEPA-Curator filters agent experiences before prompt optimization
On four agent benchmarks it gains up to 20.7% over GEPA and 21.2% over ACE with Gemini 3.1 Pro and Claude Opus 5.