Daily recap for May 19, 2026

Today we shipped 9 new tutorialized solutions for small-business AI. They cover five complete application recipes — including phone-call handling, issue triage, and lead routing — plus four template libraries for logging, rate limiting, browser automation, and secure code execution. Each tutorial comes with a downloadable codebase you can try right now.

New tutorials

Google Gemini Voice Agent for Twilio Call Handling

Build a voice agent that answers Twilio calls, transcribes speech with Deepgram, processes intents with Gemini, and responds with ElevenLabs text-to-speech. It handles appointment booking by checking calendar availability and answers FAQs using conversation memory. All calls are traced in Langfuse, and a response cache keeps repeated phrases fast and cheap.

Read the tutorial → Download the code (zip)

Under the hood: Google Gemini, @reaatech/agent-mesh, @reaatech/agent-memory, @reaatech/confidence-router, @reaatech/structured-output-repair. 91 tests, 98.95% coverage.

Azure AI Multi-Agent Handoff for Linear Issue Triage

Automatically classify and route incoming Linear issues to specialist agents for debugging, documentation, or scheduling. A Temporal workflow triggers on new issues: the orchestrator classifies the issue via Azure OpenAI, a confidence router decides the path, and the handoff protocol transfers full context. Circuit breakers protect against API failures, and a budget guard enforces cost limits before the triage comment is posted.

Read the tutorial → Download the code (zip)

Under the hood: Azure AI, @mastra/core, @reaatech/agent-handoff-protocol, @reaatech/confidence-router, @reaatech/circuit-breaker. 124 tests, 99.65% coverage.

AWS Bedrock Multi-Agent Handoff for Slack Support Triage

Route Slack support requests to specialist AI agents for billing, technical issues, or account management — all without human intervention. The Slack Bolt app classifies messages with a keyword-based confidence router, then hands off to Bedrock-powered agents through @reaatech/agent-handoff. Designed for support teams that need to automate routine triage and keep response times low.

Read the tutorial → Download the code (zip)

Under the hood: AWS Bedrock, @reaatech/agent-mesh, @reaatech/agent-handoff, @reaatech/confidence-router, @slack/bolt. 72 tests, 96.66% coverage.

Google Gemini AI Spend Control for SMBs

Add real-time cost tracking and budget enforcement to any Next.js app that uses Gemini. Pre-flight checks block calls that would exceed a hard cap, automatic model downgrading switches to cheaper alternatives when limits approach, and a live /api/spend dashboard surfaces per-tenant status. Built with @reaatech/llm-cost-telemetry and @reaatech/agent-budget-engine.

Read the tutorial → Download the code (zip)

Under the hood: Google Gemini, @reaatech/llm-cost-telemetry, @reaatech/agent-budget-engine, @reaatech/llm-router-core. 79 tests, 96.55% coverage.

OpenAI Lead Intake with Adaptive Routing

Capture, qualify, and route inbound leads automatically. An Express server parses attached PDFs and DOCX files with LlamaIndex Cloud, classifies intent using a keyword matcher and OpenAI, and hands off high-confidence leads to a webhook while syncing contacts to HubSpot. The confidence router scores each lead, and the budget engine caps per-lead LLM cost.

Read the tutorial → Download the code (zip)

Under the hood: OpenAI, @reaatech/confidence-router, @reaatech/agent-handoff, @reaatech/agent-budget-engine, LlamaIndex Cloud. 199 tests, 100% coverage.

New template libraries

Browser Automation Toolkit for AI Agents

A set of reliable browser interaction primitives for AI agents: page navigation, form filling, content extraction, and optional human-in-the-loop supervision. Use it to build agents that operate the web for testing, data scraping, or process automation.

Read the tutorial →

Under the hood: archetype browser-automation, stack TypeScript, ESM, pnpm-workspaces, vitest.

REAA Structured Logger – Core + Pino Adapter

A typed structured-log builder with automatic redaction of sensitive fields. Includes a logger core and a Pino transport for production-grade, fast logging in Node.js applications.

Read the tutorial → Download the code (zip)

Under the hood: archetype TypeScript-library-monorepo, stack TypeScript, vitest, tsup, biome, turbo, changesets.

REAA Rate Limiter – Core + Redis Adapter

Pluggable rate limiting with token-bucket and sliding-window algorithms. Ships with a Redis backend via ioredis for distributed rate limiting across processes or servers.

Read the tutorial → Download the code (zip)

Under the hood: archetype TypeScript-library-monorepo, stack TypeScript, vitest, tsup, biome, turbo, changesets.

Code Execution Sandbox for AI Agents

Isolated, secure code execution environments for LLM-generated code. Supports pluggable backends (Docker, VMs) and integrates with Model Context Protocol for use as an agent tool.

Read the tutorial →

Under the hood: archetype code-execution, stack TypeScript, ESM, pnpm-workspaces, vitest, Node, CLI.

Browse the full catalog at reaatech.com/products.