Skip to the yard
Generative AI
-
000029
Nine LLM Inference Metrics That Reveal Latency, Throughput, and Service Quality
Separate first-token delay, streaming speed, system throughput, and SLO success to see where inference performance helps or hurts users.
-
000132
Lazarus Backdoor Reprograms Dormant Neurons to Trigger AI Agent Commands
Lazarus can alter fewer than 10 dormant LLM neurons without training, creating triggered commands that may evade standard evaluations.
-
000112
Deep Agents Uses Five Techniques to Keep Long-Running AI Work in Context
Deep Agents manages growing agent histories with file offloading, summaries, on-demand skills, subagents and portable memory.
-
000111
Agent Retrieval Design: Keep Search Results Relevant and Context Small
Deep Agents combines targeted search, file offloading and subagent review to limit wasted context and reduce stale or irrelevant results.
-
000030
AI in Enterprise BI: Faster Answers Depend on Trusted Data and Controls
AI can make enterprise analytics easier to use, but reliable answers depend on current data, shared metrics, access controls and human review.
-
000113
Workday’s Agent Platform Cuts Agent Prototyping from Months to Under a Day
A four person team built an internal platform around LangGraph and LangSmith, reducing agent prototyping time while standardizing access to Workday se...
-
000096
Pinecone Adds OpenAI Text Embedding 3 Models to Integrated Search
Pinecone can now embed documents and queries with OpenAI’s text embedding 3 models inside Documents API indexes, with usage billed through Pinecone.
-
000031
Scikit-LLM Combines Translation and Zero-Shot Classification for Multilingual Text
A scikit-learn pipeline translates multilingual text into English, then classifies it with an LLM using category labels instead of a large training se...
-
000114
Stripe Built Its Kai Productivity Agent in One Week Using Deep Agents
Kai connects Stripe employees to internal data, tools and workflows; adoption reached 83% of employees each week, while skill selection remains a scal...
-
000116
Evaluation Driven Development Helps Dosu Track LLM Reliability at Scale
Dosu uses production traces and failure examples to improve its software support assistant while checking that prompt changes don’t cause regressions.
-
000117
Promptim Automates Single-Prompt Optimization with Evaluation Loops
Promptim tests prompt revisions against datasets and evaluators, retaining changes that improve development scores while keeping human review in the l...
-
000115
LangSmith BYOC Reaches General Availability Across 15 AWS Regions
LangSmith BYOC runs agent observability and deployment in a customer’s AWS account, while LangChain manages the infrastructure and upgrades.
-
000091
Why Ill-Fitting Business Software Creates Hidden Work and When to Build Custom Tools
Workarounds can make imperfect software costly in staff time. AI-assisted development adds custom apps to the options for fixing persistent gaps.
-
000032
Claude Models Detect Four Data Leakage Patterns in Benchmark Tests
Four Claude models identified planted leakage in real pipelines, while their ability to catch adjacent problems varied by model.
-
000038
Cloudflare Allocates First $301,300 of Vite Ecosystem Fund to 44 Recipients
The fund’s first round backs 44 projects and maintainers, with quarterly distributions and monthly stipends for eligible Vite stack contributors.
-
000033
5 Practical Python Automation Scripts for Images, Files, Text and Logs
Five small Python scripts automate common image, file, text, task-list and log workflows, with practical safeguards and clear limits.
-
000118
Managed Deep Agents v0.10 Adds Browser Control and Background Subagents
Version 0.10 adds browser tasks with human login handoff, asynchronous subagents, and updates to Slack threads, run control, and project configuration...
-
000009
NVIDIA’s Open Wireless Tools Put 6G Research to the Test
NVIDIA reports growing use of its wireless tools, with projects testing faster robot responses, neural receivers, lower radio access power and quicker...
-
000039
Post-Quantum Anonymous Credentials Could Bring Privacy-Preserving Web Access
MOLE explores issuer-hidden web credentials built with multivariate cryptography, with early designs measured in kilobytes and milliseconds.
-
000120
Stagehand Cuts Browser Agent Costs with Code Mode and Compact Snapshots
Across three live website benchmarks, Stagehand paired higher task success with lower cost by reducing browser calls and keeping page context compact.
-
000119
Async Subagents Let AI Agents Handle Long Tasks While Staying Responsive
Deep Agents’ async subagents run long jobs in the background, keeping the main agent available and able to coordinate parallel work.