Daily recap for June 8, 2026

Today we shipped seven new tutorials that show small businesses how to automate invoice processing, qualify leads, recover from agent failures, and more—all with production-grade code you can download and run.

New tutorials

Databricks MCP Server for SMB Data Analytics

This tutorial walks you through building a multi-tenant MCP server that lets AI agents query your Databricks SQL warehouse securely. It isolates each customer's data, enforces rate limits, and logs every access—removing the heavy lifting from your team. If you’ve wanted to give AI agents safe, auditable access to your business data, start here.

Read the tutorial → Download the code (zip)

Built with @reaatech/mcp-server-core, @reaatech/mcp-server-engine, @reaatech/mcp-server-auth, @reaatech/mcp-server-observability, @reaatech/multi-tenant-mcp-middleware, @reaatech/multi-tenant-mcp-tenant-resolver, @reaatech/multi-tenant-mcp-rate-limiter, on Databricks · 85 tests · 98.75% coverage.

xAI Grok Invoice Extraction for QuickBooks SMBs

When you’re drowning in PDF invoices, this recipe extracts the numbers you need into structured data—and fixes any malformed output the LLM spits out. It uses xAI Grok for extraction, but the real reliability comes from our repair, cost-monitoring, and budget-enforcement packages. It then pushes clean invoices straight into QuickBooks Online.

Read the tutorial → Download the code (zip)

Built with @reaatech/structured-repair-core, @reaatech/llm-cost-telemetry, @reaatech/llm-cost-telemetry-calculator, @reaatech/agent-budget-engine, @reaatech/agent-budget-spend-tracker, @reaatech/idempotency-middleware, on xAI Grok · 83 tests · 98.88% coverage.

AI Runbook Automation for Agent Failure Recovery

When your AI agents go down, don’t wait for a human to notice. This tutorial wires up automatic health checks and runbooks that isolate failing components, retry critical actions safely, and notify your team. It’s provider-agnostic—your agents can use any LLM, and the recovery machinery stays the same.

Read the tutorial → Download the code (zip)

Built with @reaatech/agent-runbook, @reaatech/agent-runbook-incident, @reaatech/agent-runbook-health-checks, @reaatech/circuit-breaker-core, @reaatech/circuit-breaker-agents, @reaatech/idempotency-middleware, on agnostic · 115 tests · 94.59% coverage.

Google Gemini Lead Intake for Jotform SMB Lead Qualification

Stop manually sorting Jotform submissions. This pipeline uses Gemini to classify leads: hot ones go straight to Slack, ambiguous ones get a follow-up email, and spam is dropped—all with cost tracking. It’s a one-endpoint system you can drop into your Next.js site.

Read the tutorial → Download the code (zip)

Built with @reaatech/confidence-router, @reaatech/confidence-router-classifiers, @reaatech/structured-repair-core, @reaatech/agent-handoff-routing, @reaatech/llm-cost-telemetry, @reaatech/session-continuity, on Google Gemini · 87 tests · 100% coverage.

Anthropic Document Pipeline for NetSuite SMB Invoice Processing

If your bookkeeper still types invoices into NetSuite by hand, this one’s for you. The pipeline extracts fields with Claude, caches recurring vendor templates to save money, enforces per-tenant cost caps, and posts vendor bills via the NetSuite REST API. It’s built to handle real-world PDFs, not just clean demo files.

Read the tutorial → Download the code (zip)

Built with @reaatech/llm-cache, @reaatech/llm-cache-adapters-redis, @reaatech/structured-repair-core, @reaatech/llm-cost-telemetry, @reaatech/agent-budget-engine, @reaatech/llm-router-core, on Anthropic · 84 tests · 97.95% coverage.

AWS Bedrock Reliability Suite for Auth0 SMB Auth Operations

Auth0 is critical for login, but it can fail silently. This tutorial wraps every API call with circuit breakers, rotates client secrets on a schedule, and triggers incident runbooks via Bedrock when failures spike. The whole thing runs as a Trigger.dev workflow so you don’t need a 24/7 ops team.

Read the tutorial → Download the code (zip)

Built with @reaatech/circuit-breaker-core, @reaatech/idempotency-middleware, @reaatech/secret-rotation-core, @reaatech/secret-rotation-provider-aws, @reaatech/agent-runbook-agent, @reaatech/agent-runbook-alerts, on Bedrock · 94 tests · 95.09% coverage.

Mistral AI Voice Agent for After-Hours Customer Support

A voice AI receptionist that answers calls after hours, understands natural speech with Deepgram, responds with Mistral, and books appointments on your calendar. It uses our voice agent core for the real-time audio loop—so you don’t have to build WebSocket plumbing from scratch.

Read the tutorial → Download the code (zip)

Built with @reaatech/voice-agent-core, @reaatech/session-continuity, @reaatech/voice-agent-telephony, @reaatech/voice-agent-stt, @reaatech/voice-agent-tts, @reaatech/agent-memory, on Mistral · 110 tests · 95.76% coverage.

Browse all solutions →

Building blocks shipped

Agent Eval Harness

Two patch releases landed today for the evals-quality family. Both fixed CI failures on main (type checks and Docker builds) that had been blocking merges. The @reaatech/agent-eval-harness-observability package also upgraded its OpenTelemetry stack from v1 to v2, which improves tracing and metrics for agent evaluations. No new packages—just sturdier plumbing for your quality gates.

Browse the agent-eval-harness catalog →

Browse the full catalog at reaatech.com/products.