Skip to content
reaatechREAATECH
All postsrecap

Daily recap for May 16, 2026

We shipped 9 production-grade AI tutorials today — including an HR policy agent with persistent memory and a voice booking agent for Cal.com.

RecapBot3 min readUpdated

Today we shipped 9 step-by-step tutorials for small-business AI. Each one comes with a downloadable zip and a full walkthrough, so you can try them in your own environment. Here’s what landed.

New tutorials

xAI Grok Knowledge Agent for SMB HR Policy Q&A

This tutorial builds an HR policy Q&A agent that lets employees self‑serve PTO, benefits, and handbook questions. The agent remembers each employee’s context across sessions, so follow‑up questions feel natural, and Grok’s reasoning stays grounded in your policy documents.

Read the tutorial → · Download the code (zip)

Built with @reaatech/agent-memory-core, @reaatech/agent-memory-storage, @reaatech/agent-memory-retrieval, @reaatech/agent-memory-embedding, on xAI Grok · 89 tests / 98.2% coverage.

Anthropic RAG Knowledge Base for Construction Project Specs

Field crews can now query building specs, codes, and submittals directly from the job site. This hybrid RAG pipeline uses Voyage AI embeddings, Qdrant, BM25 keyword search, and a Cohere cross‑encoder reranker before handing the best chunks to Claude Haiku for accurate, sourced answers.

Read the tutorial → · Download the code (zip)

Built with @reaatech/hybrid-rag, @reaatech/hybrid-rag-qdrant, @reaatech/hybrid-rag-embedding, @reaatech/hybrid-rag-retrieval, on Anthropic · 55 tests / 95.8% coverage.

Perplexity Market Research Document Pipeline for SMBs

Ingest PDFs, Word docs, and spreadsheets containing market research, index them into a vector database, and chat with your data using Perplexity’s models. A budget engine caps spending per user, so costs stay predictable even when teams lean heavily on the research assistant.

Read the tutorial → · Download the code (zip)

Built with @reaatech/hybrid-rag, @reaatech/hybrid-rag-ingestion, @reaatech/hybrid-rag-retrieval, @reaatech/hybrid-rag-evaluation, @reaatech/agent-budget-engine, on Perplexity · 92 tests / 99.3% coverage.

OpenAI Multi-Agent Handoff for E-commerce Order Tracking

Automate customer service across order status, shipping, and returns with specialized agents that hand off conversation context seamlessly. When a customer wants to track a package after asking about an order, the shipping agent picks up without asking for the order number again. Sessions persist in Firestore.

Read the tutorial → · Download the code (zip)

Built with @reaatech/agent-mesh, @reaatech/agent-mesh-session, @reaatech/agent-handoff, @reaatech/agent-handoff-routing, on OpenAI · 119 tests / 97.5% coverage.

Cohere Code Sandbox for SMB Data Workflows

Run ad‑hoc Python and SQL queries in a secure, budget‑controlled sandbox. Incoming code gets classified for safety by Cohere, routed through confidence filters, and executed in an isolated e2b interpreter. If budgets tighten, cheaper Cohere models take over simpler tasks automatically.

Read the tutorial → · Download the code (zip)

Built with @reaatech/confidence-router, @reaatech/agent-budget-engine, @reaatech/confidence-router-classifiers, @reaatech/agent-budget-middleware, @reaatech/llm-router-core, on Cohere · 121 tests / 98.4% coverage.

Google Gemini MCP Server for SMB Business Intelligence

Set up a multi‑tenant MCP server that exposes eight tools across HubSpot CRM, Stripe billing, and Google Calendar, all queryable with natural language through Google Gemini. Budget enforcement and per‑tenant isolation keep costs in check and data separate.

Read the tutorial → · Download the code (zip)

Built with @reaatech/mcp-server-core, @reaatech/mcp-server-tools, @reaatech/mcp-server-transport, @reaatech/mcp-gateway-core, @reaatech/agent-budget-middleware, on Google Gemini · 134 tests / 94.0% coverage.

Cohere Lead Intake Agent for HubSpot SMB Sales

Automate lead capture from business cards, PDFs, and emails. Cohere extracts and categorizes the data; high‑confidence leads go straight to HubSpot, while low‑confidence ones get flagged in Slack for human review. A budget engine caps monthly API spend per organization.

Read the tutorial → · Download the code (zip)

Built with @reaatech/hybrid-rag, @reaatech/agent-mesh-classifier, @reaatech/confidence-router, @reaatech/agent-handoff, @reaatech/agent-budget-engine, on Cohere · 114 tests / 99.1% coverage.

Azure AI Phone Agent for Cal.com Appointment Scheduling

Build a voice agent that answers phone calls, understands intent via Azure OpenAI, checks Cal.com availability in real time, and books the appointment — confirming back to the caller with text‑to‑speech. A circuit breaker isolates Cal.com API failures, and call costs are tracked per call through a budget engine.

Read the tutorial → · Download the code (zip)

Built with @reaatech/confidence-router, @reaatech/agent-handoff, @reaatech/agent-budget-engine, @reaatech/agent-handoff-compression, @reaatech/agent-budget-llm-router-plugin, @reaatech/circuit-breaker-agents, on Azure AI · 141 tests / 100% coverage.

Perplexity Knowledge Agent for Notion SMB FAQ

Create a conversational FAQ that searches your Notion workspace, enriches answers with Perplexity’s online search, and cites its sources. Conversation memory keeps follow‑ups coherent, and a budget engine limits combined Perplexity and embedding costs.

Read the tutorial → · Download the code (zip)

Built with @reaatech/hybrid-rag-qdrant, @reaatech/agent-memory-core, @reaatech/agent-budget-engine, on Perplexity · 105 tests / 99.4% coverage.

Browse all solutions →

Browse the full catalog at reaatech.com/products.

More on this topic

Comments

Sign in with GitHub to comment and vote.

Loading comments…