Newsletter archive
Applied Signals Weekly: 2026-08-30
Aug 24, 2026 to Aug 30, 2026
In this week’s edition
This week’s edition surfaces where agentic systems face the sharpest pressure points: reliability under coordination, evaluation discipline, and the practical mechanics that keep multi-agent behavior stable over time.
- FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference — Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet their real-world deployment remains bottlenecked by high inference latency and unstable asynchronous execution. This challenge…
- How Signal found two hidden retry loops in our production agent Alyx — We ran Signal on Alyx, the AI engineering agent built into Arize AX. It surfaced a duplicate task-state loop and a 43-call dataset retry that appeared as valid tool activity or an OK root span.
- UrbanGround: From Local Perception to Spatial Agency in a Real-Scale City — Multimodal large language models (MLLMs) can interpret a street view, but urban agency depends on whether such local evidence remains useful after the agent starts to move. In this paper, we investigate how far current…
- What Makes Good Agentic Data? An ACE Lens on Data Generation for LLM Agents — LLM agents increasingly rely on generated interaction data to learn how to interact with external environments. Agentic data generation must maintain consistency among environments, tasks, interactions, and success…
- WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill… — Agent skills package specialized knowledge and workflows into reusable resources that extend AI agent capabilities. Recent work automatically discovers such skills from agent experience, which enables agents to…
- A skill is just an agent. So measure your changes. — A skill is just another AI agent: a prompt plus a harness that runs it. That means you can trace it, eval it, and prove a change made it better instead of shipping on vibes. Here's how we quantify skill changes with…
- Arize Phoenix has a built-in MCP server that lets your agents query traces with… — Read-only SQL and code mode let coding agents answer questions across your traces without paging thousands of spans through model context.
Worth watching
- evaluation + planning: hype or repeatable pattern?
- evaluation + frameworks: hype or repeatable pattern?
- evaluation + tool-use: hype or repeatable pattern?
Read the full analysis
- ASIL, KOPE, and StepGuard quantify agent gains via interfaces and audits (weekly trend)
- Aug 30, 2026 ASIL, Signal, FlashVLA: structured tool use, trace loops, streaming decoding
- Aug 29, 2026 ACE framework findings: UrbanGround errors, CARL Lenia control, PICasso PIC
- Aug 28, 2026 GVS5H ledger orchestration, SKILL.state explicit state, MemToC arbitration benchmark
- Aug 27, 2026 Praxist lineage traces, KOPE experience graphs, VoiceMem 134 ms memory
- Aug 26, 2026 AgentLance market orchestration and Station open-world math agents
- Aug 25, 2026 Prime Agent persistent REPL and PropUQ-MAS uncertainty propagation
- Aug 24, 2026 PrimeAgentOrchestrator, XPerf, and Nexus KV splicing updates
Subscribe
Get Applied Signals weekly newsletter in your inbox