Daily recap for May 27, 2026

Today we shipped four new step-by-step tutorials for small-business AI. These downloadable recipes cover an agent mesh that runs entirely on your own hardware, a voice agent for restaurant reservations, a lease extraction pipeline for property managers, and a code sandbox for QuickBooks analysis — each ready to deploy in your stack.

New tutorials

Ollama agent mesh for small business workflow automation

Orchestrate a fleet of local LLM specialists that handle email triage, CRM updates, and report generation without sending data to the cloud. The mesh runs entirely on your own hardware using Ollama's local models, distributing tasks to specialized agents with confidence scoring to decide whether to proceed, ask for clarification, or escalate. Context is maintained across multi-turn conversations, and no customer data ever leaves your network.

Read the tutorial → · Download the code (zip)

Built with @reaatech/agent-mesh, @reaatech/agent-mesh-router, @reaatech/confidence-router, @reaatech/session-continuity, @reaatech/agent-handoff, @reaatech/llm-router-core, on ollama · 139 tests / 97.19% coverage.

AWS Bedrock document pipeline for AppFolio lease data extraction

Property managers can automate lease data extraction from PDFs and scanned images into AppFolio. The pipeline ingests documents, extracts text using unpdf or Tesseract OCR, then sends structured prompts to AWS Bedrock (Claude) for field extraction. A structured repair core fixes any malformed JSON before writing to AppFolio’s REST API, and a tool-use firewall ensures the agent only calls allowed endpoints and requires human approval for destructive actions. Hours of manual entry are eliminated.

Read the tutorial → · Download the code (zip)

Built with @reaatech/structured-repair-core, @reaatech/llm-cost-telemetry, @reaatech/agent-budget-engine, @reaatech/tool-use-firewall-core, on bedrock · 97 tests / 100.00% coverage.

OpenAI voice agent for OpenTable reservation management

Restaurants can let diners handle reservations by phone using natural conversation. This voice agent uses Twilio for telephony, Deepgram for speech-to-text, OpenAI’s GPT model for intent classification, Cartesia for text-to-speech, and a custom OpenTable API client. Session continuity maintains context across multi-turn calls, and structured repair ensures output reliability. The entire pipeline is wired together with REAA’s voice-agent-core orchestrator.

Read the tutorial → · Download the code (zip)

Built with @reaatech/voice-agent-core, @reaatech/voice-agent-telephony, @reaatech/voice-agent-stt, @reaatech/voice-agent-tts, @reaatech/voice-agent-mcp-client, @reaatech/session-continuity, @reaatech/structured-repair-core, on openai · 127 tests / 98.96% coverage.

Vertex AI code sandbox for QuickBooks financial modeling

Small business owners can ask conversational questions about their QuickBooks data and get safe, sandboxed code execution that generates answers. Vertex AI generates Python code to query QuickBooks, which runs in a secure e2b sandbox. Spend controls and human-in-the-loop approvals via a tool-use firewall prevent runaway costs or unintended writes. Session continuity and structured repair ensure reliable, context-aware answers across multi-turn conversations.

Read the tutorial → · Download the code (zip)

Built with @reaatech/session-continuity, @reaatech/tool-use-firewall-core, @reaatech/tool-use-firewall-approvals, @reaatech/llm-cost-telemetry, @reaatech/structured-repair-core, on vertex · 159 tests / 98.91% coverage.

Browse all solutions →

Browse the full catalog at reaatech.com/products.