groundy

Groundy — independent coverage of developer tools, infrastructure, and platforms

agents

Why Multi-Agent LLM Delegation Concentrates Risk

Splitting tasks across planner and worker agents concentrates risk because principals cannot observe hidden actions. Base models defect. Teams must add verifiable execution.

6min·08:25·jul 29
models

Kimi Linear Cuts KV Cache 75% but Recall Remains the Binding Constraint

Kimi Linear cuts KV cache 75% and boosts decode 6x for 1M-token contexts, but recall stays the binding constraint. Route memory-bound jobs only after benchmarking in-context.

6min·18:34·jul 29
policy

Why Vendor Model Cards Fail Clinical Ethics Procurement

PrinciplismQA exposes a gap generic leaderboards miss. Hospital compliance teams must build domain-specific ethics evals to test autonomy versus beneficence conflicts before.

agents

Harness vs Scaffold: Why Claude Code and LangGraph Are Not Interchangeable

A 2026 study shows harness choice swings coding-agent token costs by 40x while pass rates barely move. Misplacing tool gating in the scaffold layer creates integration debt.

devtools

Fine-Tuning vs RAG for Internal APIs: StarCoder2 Constraints

StarCoder2's 4,096-token sliding attention limits RAG for internal APIs. Fine-tuning encodes proprietary signatures in weights but requires a permanent dataset curation loop.

agents

MCP Tool Discovery Moves From Hardcoded Config to Runtime Agent Search

Runtime tool discovery moves the MCP trust boundary from manifest review to per-call authorization, turning a lockfile choke point into unbounded binding decisions.

infra

Calibrated LLM Monitoring: Conformal Prediction with Drift Detection

A cloud-native architecture packages conformal prediction as a sub-2ms microservice to replace raw logprob confidence with finite-sample coverage guarantees, gated by.

devtools

Mellum2 Unverified: Why MoE Active Parameters Matter More Than Total Size

Mellum2 specs lack sources. DeepSeek-Coder-V2 proves MoE economics. Local teams must demand active parameter counts before deploying coding models.

featurededitor's picks
most popularlast 6 days · by traffic
tophighest rated
feedall 644 articles · reverse chronological
  1. jul 29agentsWhy Multi-Agent LLM Delegation Concentrates Risk
  2. jul 29modelsKimi Linear Cuts KV Cache 75% but Recall Remains the Binding Constraint
  3. jul 29policyWhy Vendor Model Cards Fail Clinical Ethics Procurement
  4. jul 29agentsHarness vs Scaffold: Why Claude Code and LangGraph Are Not Interchangeable
  5. jul 28devtoolsFine-Tuning vs RAG for Internal APIs: StarCoder2 Constraints
  6. jul 28agentsMCP Tool Discovery Moves From Hardcoded Config to Runtime Agent Search
  7. jul 28infraCalibrated LLM Monitoring: Conformal Prediction with Drift Detection
  8. jul 28devtoolsMellum2 Unverified: Why MoE Active Parameters Matter More Than Total Size
  9. jul 28agentsCode-as-Action Agents Beat GAIA But Require Runtime Sandboxing
  10. jul 28policyTRIDENT Benchmark: LLM Safety Gaps in Finance, Medicine, and Law
  11. jul 28modelsWhy Chat Leaderboards Do Not Predict Image Quality
  12. jul 27devtoolsPyPI Wheel Reproducibility: 15% Byte-Identical, 79% Source-Equivalent
  13. jul 27modelsKimi K3 Procurement: Governance Review Over Phantom Government Assessments
  14. jul 27policyEU AI Act Traceability: Why ML Pipelines Fail Conformity Assessment
  15. jul 27infraCloudflare AI Crawler Controls: Block, Charge, or Allow Bots Per Route
  16. jul 27agentsWhy Agent Security Tests Must Audit Full Trajectories, Not Single Turns
  17. jul 26agentsx402 Per-Call Payments: Agent Wallet Custody and Replay Risks
  18. jul 26modelsDeepSeek Compute Leak: Why Open-Weight Routing Needs a Swap Path
  19. jul 26modelsContext Ordering Beats Window Size for Long-Context Agents
  20. jul 26devtoolsCHRONO-RESOLUTION: npm, PyPI, and crates.io lockfile drift measured at release points
  21. jul 25infraPostgres LISTEN/NOTIFY Scales: When to Drop Redis for Job Fan-Out
  22. jul 25devtoolsCLI-Tool-Bench: Why Patch Leaderboards Fail for 0-to-1 Code Generation
  23. jul 25policyImplicit Bias in LLMs Passes NYC and EU Audits
  24. jul 25agentsCodeRabbit Review Study: 56% Rejection Rate Demands Targeted Scoping
  25. jul 24modelsDiffusion LLMs: Training Cost, Not Parallel Decoding, Drives Deployment
  26. jul 24infraTailscale on Azure: Measure Direct vs DERP Routing to Control Latency and Egress
  27. jul 24infraAccelerate vs Megatron Core: The Model Size Curve for Distributed Training
  28. jul 24agentsLLM Agents Ignore Mid-Flight Halt Signals: 0 of 40 Trials Stopped
  29. jul 24modelsOpen-Weight Routers vs Fable 5: The Routing Math That Actually Matters
  30. jul 24agentsAgent-First CLIs: Why GitHub, npm, and PyPI Must Publish Machine-Readable Contracts
  31. jul 23policyEU Driver Monitoring: GDPR Compliance Without Consent
  32. jul 23infraWhy cgroups, not permission prompts, bound AI agent CPU and memory
  33. jul 23modelsDeepSeek-V4 1M Context vs RAG: Why Retrieval Stays
  34. jul 23modelsQwen-Image-3.0 Does Not Exist: Why Self-Hosting Image Models Is Premature
  35. jul 22policyEU AI Act bans emotion AI in schools, but permits it where models fail
  36. jul 22agentsMCP and AGENTS.md Standardize Context, Not Agent Coordination
  37. jul 22industryAI Search's One-Answer Rule: When Better Content Makes Search Worse
  38. jul 22agentsCursor's Swarm Math: When Cheap Agents Save Money and When They Fail
  39. jul 21agentsRuntime monitoring beats alignment for agent-to-agent coercion
  40. jul 21infravLLM Configs Shift Energy, Latency, and Accuracy: A 9,000-Run Study
  41. jul 21policyHuggingFace vs GitHub Models vs Replicate: Policy Compliance for Uploaders
  42. jul 20modelsKimi K3: 2.8T Parameters, MoE Routing, and Self-Hosting Reality
  43. jul 20modelsKimi K3 vs Qwen3.8 Max: Routing Strategy for July 2026
  44. jul 20agentsCloudflare's Agent Stack: Edge Trust, Identity, and Metering
  45. jul 20modelsQwen3.8 Max Preview: Missing Benchmarks, Weights, and Pricing
  46. jul 20policySAMark Text Watermarking: Paraphrase Robustness and the Policy Gap
  47. jul 20industryHuggingFace's $100M Series C Locks Teams Into Deployment
  48. jul 19modelsHuggingFace 100x Inference: Generalizable vs Platform-Locked Optimizations
  49. jul 19agentsLM Studio Bionic vs Claude Code: Local-First vs Cloud Agent Tradeoffs
  50. jul 19infraAWS Estimated Billing Was Off by $1.7B: Reconciling Actual Cloud Spend
load older →