Daily recap for May 19, 2026

Today we shipped three new tutorials for small-business AI — each with a downloadable project you can run in under an hour. Triaging Slack support with Bedrock agents, controlling Gemini spend, and qualifying leads with adaptive routing.

New tutorials

AWS Bedrock Multi‑Agent Handoff for Slack Support Triage

Slack support channels get noisy fast. This tutorial gives you a triage bot that reads each message, decides whether it’s about billing, a technical bug, or an account question, and hands it to the right specialist agent — all without a person routing tickets. The agents run on AWS Bedrock LLMs, swap conversation context cleanly, and classify requests using a keyword-confidence hybrid. Download the project, add your Bedrock and Slack keys, and you have a running triage system in your workspace.

Read the tutorial → and download the project zip.

Built with @reaatech/agent-mesh, @reaatech/agent-handoff, @reaatech/agent-handoff-routing, @reaatech/confidence-router, @reaatech/agent-memory, on Bedrock · 72 tests · 96.66% coverage.

Google Gemini AI Spend Control for SMBs

Gemini is powerful, but if you’re a small business, per-token costs can add up fast with no clear picture of where the money is going. This recipe instruments every Gemini API call in your app to track spend, enforce budgets, and automatically switch to cheaper models when you’re approaching a cap. You get a real-time spend dashboard at GET /api/spend and per-tenant budget headers on every response. It uses @reaatech/llm-cost-telemetry for instrumentation and @reaatech/agent-budget-engine for pre-call checks and model downgrading, with OpenTelemetry to bridge your existing monitoring stack.

Read the tutorial → and download the project zip.

Built with @reaatech/llm-cost-telemetry, @reaatech/agent-budget-engine, @reaatech/llm-cost-telemetry-aggregation, @reaatech/agent-budget-middleware, @reaatech/llm-router-core, @reaatech/agent-budget-otel-bridge, on Google Gemini · 79 tests · 96.55% coverage.

OpenAI Lead Intake with Adaptive Routing

Inbound leads are worth money only if someone follows up fast. This solution captures lead data, including uploaded PDFs and DOCX files, qualifies the lead by reading document contents and form fields, and scores its readiness to buy. High-confidence leads get routed to the right sales rep automatically, while lower-confidence ones can be nurtured. An OpenAI LLM does the thinking, @reaatech/confidence-router scores intent, and @reaatech/agent-handoff pushes the qualified lead to your team’s webhook or HubSpot. The built-in budget engine keeps LLM costs per lead predictable.

Read the tutorial → and download the project zip.

Built with @reaatech/confidence-router, @reaatech/confidence-router-classifiers, @reaatech/agent-handoff, @reaatech/agent-budget-engine, on OpenAI · 199 tests · 100.00% coverage.

Browse all solutions →