Agentic AI, LLM engineering, and the models behind modern automation — multi-agent systems, LLM evaluation and comparisons, RAG, fine-tuning, AI infrastructure, security, and production AI engineering.
From the blog
Live calls record to S3, then EventBridge triggers AWS Batch + Whisper-large-v3 (or Parakeet) for high-quality transcription. We show the full event-driven pipeline with diarization and PII redaction stitched in.
PCI DSS 4.0.1 future-dated requirements went mandatory March 31, 2025. AI voice agents that take card payments on behalf of healthcare providers — copays, deductibles, payment plans — must meet 12 requirements with DTMF masking and scope reduction.
Bloomscape and The Sill ship live plants by mail with care guarantees. Chat agents that match plant to light, room, and skill — and coach care when plants struggle — cut replacements 30%+. Here is the 2026 playbook.
Together.ai's voice infrastructure delivers Kokoro TTS at 97ms baseline TTFB, sub-200ms TTS under load, and Llama 70B at 95 tok/s with 220ms TTFT. Build a voice agent on open weights.
HubSpot's Breeze Customer Agent resolves 65%+ of conversations automatically with 39% faster ticket resolution in 2026. Here is how a chat agent on top of Service Hub extends to voice, SMS, and WhatsApp without paying $1.00 per conversation.
Self-hosting Llama 3.1 70B vs paying OpenAI: break-even falls between 10M and 30M tokens/day. We model 12M, 50M, 200M tokens/day across 24 months — including the $3–6K/mo hidden engineering cost.
Google's Agent2Agent protocol — now Linux Foundation governed with 150+ orgs supporting — is the missing primitive for cross-vendor agent collaboration. We cover the JSON-RPC mechanics, MCP overlap, and real interop scenarios.
Prisma owns the schema model, Atlas owns the migration plan, lint, and CI/CD. Together you get declarative schema, automatic diff plans, and zero-downtime production deploys for AI-heavy Postgres.
Tool calls eat 35-61% of agent task time. Speculative execution predicts the next tool from the agent's typical control flow and runs it in parallel. PASTE shows 48.5% task-time reduction in 2026.
