Agents & Frameworks
Why Your AI Travel Agent Would Book a Bullfight
A new travel-agent benchmark finds frontier models book animal-exploitation options below chance when the welfare preference is implicit. Fix the action space, not the prompt.
The Groundy archive · Page 23 of 43
All articles
Browse Groundy's complete archive of 1027 articles on AI, developer tools and infrastructure. Page 23 of 43.
The archive
529–552 of 1027 articles · Newest first
- Developer Tools
Coding Agents Hallucinate Internal APIs: Execution Memory Beats RAG Context
MEMCoder's execution memory lifts private-library pass@1 by 18.41 points over RAG by learning from runtime feedback, shifting bottleneck from retrieval to sandboxed execution.
- Industry & Business
Meta's Layoff Admission Weakens the Case for AI Headcount Cuts
Zuckerberg said Meta's AI agent development has not accelerated as expected, weeks after 8,000 layoffs. CIOs lose their flagship case study for AI-driven headcount cuts.
- Industry & Business
Kimi K3 Confirmed for July After K2.7 Lost 11 of 12 Benchmark Cells
Kimi K3 shipped five weeks after K2.7 Code. The rapid reset shows why routing teams need rolling evaluations, price controls and version-agnostic contracts.
- Agents & Frameworks
Can Multi-Agent RAG Run Air-Gapped? A Forensics System Shows How
CHARLIE runs multi-agent RAG on-premise for forensic evidence, removing cloud LLM calls but shifting the bottleneck to GPU capacity, structured memory, and audit logging.
- Industry & Business
Meituan Open-Sources LongCat-2.0, a 1.6T Model Trained on 50,000 Chinese GPUs
Meituan says LongCat-2.0 is a 1.6-trillion-parameter MoE trained on 50,000 domestic chips. If true, export controls may not confine frontier model training to national labs.
- Models & Research
When Do Time Series Foundation Models Pay Off? The Break-Even Threshold
A July 2026 arXiv break-even study finds pretrained time series foundation models win unconditionally on half of 30 datasets, but lose to ARIMA or XGBoost on a fifth at any.
- Agents & Frameworks
Can You Prove an Agentic Trading Pipeline Has No Look-Ahead Bias?
A July 2026 preprint reframes look-ahead bias as temporal non-interference, letting a type checker prove a pipeline leak-free before it runs rather than finding leaks after.
- Agents & Frameworks
How Far Ahead Can a Coding Agent Plan? The Horizon Bottleneck
A July 2026 preprint says coding agents plan edits roughly 25 steps ahead before their latent program model collapses, so context length and pass@k may mismeasure failures.
- Infrastructure & Runtime
Cloudflare Meerkat: What Globally Distributed Consensus Costs at the Edge
Cloudflare's Meerkat runs QuePaxa consensus at the edge, so every write waits on a cross-region quorum. The write-latency tax suits control-plane state, not transactions.
- Infrastructure & Runtime
Vercel CDN Now Honors External Origin Cache-Control: Audit Your Headers
Since April 6, 2026, new Vercel projects honor Cache-Control from external origins by default, so operators must audit rewrite headers or risk stale, unintended responses.
- Industry & Business
Prompt Refinement vs Reflective Dialogue: Which Builds Better AI Coders?
A July 2026 study found Socratic tutoring outperformed prompt refinement for independent LLM use among students, suggesting engineering teams may underinvest in evaluation.
- Infrastructure & Runtime
AI Found Real Bugs in Cloudflare's Circl Crypto Library
zkSecurity's AI agent found seven real bugs in Cloudflare's CIRCL crypto library, all fixed upstream, showing AI-assisted review is now a baseline layer for crypto code.
- Infrastructure & Runtime
Pruning RAG Context: What to Cut Before the LLM Sees It
Pruning RAG context is a ranking decision. Reranking and compression keep only answer-changing chunks, shifting cost from the prompt onto retrieval and shrinking the cite set.
- Ethics, Policy & Safety
Can US Export Controls Contain AI Built Without American Chips?
Meituan's unverified LongCat-2.0 claim, a trillion-parameter model on domestic chips, would weaken US chip controls if true. Without proof, the bottleneck question is open.
- Developer Tools
The Vercel-Supabase Pairing Exposes the Distribution Tax Backend Vendors Pay
The Vercel-plus-Supabase pairing exposes the platform trade for backend vendors: distribution costs margin and the customer relationship in exchange for reach.
- Ethics, Policy & Safety
Entropy Regularization Buys RL Robustness That Certification Can't Credit
A July 2026 arXiv preprint proves entropy regularization in continuous-time RL lower-bounds robustness to joint perturbations, but ISO 26262 and IEC 61508 cannot credit it.
- Ethics, Policy & Safety
Fusion's ML Disruption Predictors Have No Shared Validation Standard
A new EAST paper makes real-time tokamak disruption predictors cheap, but ML safety systems lack shared cross-machine validation standards before commercial plants debut.
- Developer Tools
Vercel Flags Segments Reach the CLI: Feature Flags as Code, Not Dashboard Clicks
Vercel's July 3 CLI release makes flag segments scriptable code, but segment edits propagate to all referencing flags and deletes stay blocked until dependencies clear.
- Industry & Business
OpenAI's Latest Funding Round Bets Investors Will Wait for a 2027 IPO
OpenAI priced its record funding round at $852 billion while delaying its IPO to 2027, forcing private backers to bet monetization can outrun compute burn before it lists.
- Infrastructure & Runtime
Cloudflare's x402 Gateway: What Per-Request API Billing Actually Needs
Cloudflare's Monetization Gateway prices resources per request in stablecoins via HTTP 402, shifting fee friction to callers and splitting pricing into metered and flat plans.
- Ethics, Policy & Safety
AI-Generated CSAM Risks Expose Filter-First Safety Gaps
A July 2026 ICML spotlight paper argues preventing AI-generated CSAM requires upstream design controls, because auditing, red teaming, and benchmarking cannot include it.
- Ethics, Policy & Safety
Treating AI Governance as Code Moves Compliance Into the Build Pipeline
The CANONIC preprint treats AI governance as a compiler check, lowering audit costs but confirming that structural admission cannot detect slop; it only produces an audit.
- Ethics, Policy & Safety
GitHub Issues Are Now Where GDPR and CCPA Compliance Gets Decided
An arXiv study of 32,820 GitHub issues finds developers negotiating GDPR and CCPA line by line, shifting privacy liability to maintainers and limiting automated scans.