Research

Filed research investigations.

• AI Browser Automation & Scraping Tools 2026 — agent-browser, browser-use, Stagehand deep-dive and architecture patterns
• AI-Native Dev Tools Landscape 2026 — Positioning, shipped AI features, MCP adoption, commoditization risk
• AI Social Media Reply Tools 2026 — MCP-native tools, commercial SaaS, workflow pipelines
• Automated Social Media Engagement Risks 2026 — ToS, shadow banning, EU/Dutch GDPR, EU AI Act compliance
• Dev Tool Growth Playbooks — How Linear, Vercel, Supabase, Railway grew from zero to traction
• Dev Tool Pricing Landscape 2026 Q2 — Competitor pricing, feature gating, AI trends, Kendo positioning
• Developer PM Sentiment 2026 — Jira fatigue, context-switching data, pain point ranking
• EU Data Sovereignty Demand 2026 — Regulatory wave, segment-by-segment assessment, competitive positioning
• Kendo App Plans Landscape — Feature map, architectural decisions, process evolution across 97 app plans
• The Kendo Way — Architectural Patterns — Patterns, conventions, anti-patterns from DECISIONS.md + ARCHITECTURE.md
• Claude Managed Agents — Capability Survey & Kendo Application Sketches — Capability survey of Anthropic CMA, kendo-specific implications, three application sketches, decisions
• MCP Ecosystem Landscape 2026 Q2 — 10+ PM tools now have MCP servers — window has closed
• Claude Managed Agents — Production Spike & Migration Decision — Empirical CMA-vs-Messages-API spike: 19× cheaper on synthetic prompts, 20.7× cheaper on production-shape; harness lessons. Latency claim corrected by 2026-05-09 postmortem
• Production Telemetry Contradicts the Spike's Wall-Clock Claim — Postmortem: production p95 is 28s, not 483s; spike picked one open-ended prompt as headline; cost case stands, latency case reversed
• Unified AI SDKs (TypeScript) 2026 — Vercel AI SDK, OpenAI Agents, Mastra, LiteLLM — comparison matrix