Tag
AMD has achieved a 46% revenue share in x86 server CPUs, up from nearly zero in 2017, driven by its new Venice processors that offer 2.2x performance per core over Nvidia's comparable chip. The shift highlights a growing preference for modular CPU+GPU setups in AI agentic workloads over Nvidia's vertically integrated racks.
A tutorial on building agentic workflows in Python using LangGraph, covering state, nodes, edges, tool integration, and conversation memory.
Jerry Liu announces a live webinar on parsing complex financial documents using LlamaIndex to enable cleaner structured outputs for agentic workflows.
The article argues that the true value of enterprise AI lies in agentic workflows rather than chatbots, suggesting a shift in focus.
Google introduces Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber models, offering improved token efficiency, lower latency, and better performance for agentic workflows, with 3.6 Flash reducing output token usage by 17% and showing gains in coding and knowledge tasks.
This paper introduces an agentic workflow that uses LLMs and VLMs to iteratively generate and improve high-quality mathematical diagrams for K-12 education, addressing the reliability gap in AI-generated visual aids.
Levie argues that lowering the cost of AI intelligence (tokens) is key to large-scale agentic adoption, and predicts most information work will involve agents in the future, citing innovation in both open and closed models.
ByteDance's Seed team has released the Seed2.0 model card, detailing a model designed to bridge the gap between lab benchmarks and real-world software engineering. The card highlights deployment tiers, performance comparisons, and honest acknowledgment of gaps versus frontier models.
Introduces CausalDS, a benchmark for evaluating causal reasoning in LLM-based data science agents, using synthetic structural causal models and natural language stories to test associational, interventional, and counterfactual reasoning along with tool use and abstention.
Introduces PatchOptic, an interface for shared-state LLM workflows that uses projected reads and verified structured patches to ensure valid updates. Evaluated with PatchBench across 46 cases, showing reduced token cost and leakage while maintaining quality.
Anthropic published four types of loops for Claude Code to operate autonomously: turn-based, goal-based, time-based, and proactive, allowing different levels of task handoff.
Forethought is a neurosymbolic reasoning system that treats reasoning as an explicit, verifiable program composed from symbolic and neural primitives. It improves base-model accuracy by about 30% relative and enables small models to match frontier models while being model-agnostic and auditable.
Benchmark results for Qwen3.6-35B-A3B-UD-Q8_K_XL on DGX Spark using llama.cpp script by Mia, showing fast token generation times across various context lengths.
The article discusses common causes of cost spikes in AI workflows, such as retries, repeated tool calls, long-running workflows, and growing context, and asks how teams investigate such issues.
The tweet agrees that $200/week is sufficient for engineering and research work, criticizing wasteful spending on expensive models and bloated agentic workflows.
OpenRouter posted on the Insights blog, pointing out that four open-weight models have reached a stage capable of supporting real agent workflows, and explained why the company chose these models in June.
Anthropic's computer use team is hiring product engineers and researchers, seeking candidates who are passionate about agentic workflows and comfortable with ambiguity.
This paper demonstrates that compiling agentic workflow procedures into the weights of a small fine-tuned model achieves near-frontier quality at 128–462× cost reduction compared to in-context baselines, addressing perceived barriers of quality, cost, and flexibility.
John Lindquist shares his talk 'Agentic Power User's Playbook,' covering practical workflows, shortcuts, and habits for efficiently managing AI agents.
A discussion about real-world failures of autonomous AI agents in production, such as sending unauthorized emails, modifying records, deleting data, and spending money, seeking experiences and guardrails.