Daily recap for June 17, 2026

New tutorials

xAI Grok Tax Form Extraction for SMB Accounting

This tutorial builds a document pipeline for small-business accounting teams: upload a tax form PDF, and it extracts structured, schema-validated JSON ready for QuickBooks or Xero. It uses xAI Grok's reasoning to parse the form, then repairs any malformed output with REAA's structured repair engine. Per-form cost is tracked, and identical pages are cached to avoid reprocessing. The built-in budget engine prevents surprise bills.

Read the tutorial → · Download the code (zip)

Built with @reaatech/structured-repair-core, @reaatech/llm-cost-telemetry, @reaatech/llm-cache, @reaatech/agent-budget-engine, on xAI Grok · 92 tests / 97.20% coverage.

Perplexity RAG Eval Suite for SMB Knowledge Bases

This CLI tool gives SMBs a way to automate quality checks on their internal RAG knowledge bases. It loads a golden Q&A dataset, scores faithfulness and relevance using heuristic metrics, then invokes Perplexity as a cost-effective LLM judge for ambiguous answers. Every run’s spend is tracked, and the suite can fail CI pipelines when scores dip below thresholds — so regressions surface before users notice.

Read the tutorial → · Download the code (zip)

Built with @reaatech/rag-eval-core, @reaatech/rag-eval-dataset, @reaatech/rag-eval-judge, @reaatech/llm-cost-telemetry, @reaatech/rag-eval-gate, on Perplexity · 81 tests / 99.63% coverage.

Vertex AI Multi-Agent Handoff for SMB Field Service Dispatch

This recipe creates a dispatch orchestrator for field service businesses using Vertex AI. When a customer request comes in, confidence-based routing decides whether to handle it autonomously or escalate to a human, while an LLM router picks the cheapest capable Gemini model for each specialist sub-agent (scheduling, inventory, billing). Session continuity preserves the full conversation across handoffs, and per-agent budgets cap spend.

Read the tutorial → · Download the code (zip)

Built with @reaatech/agent-handoff, @reaatech/confidence-router, @reaatech/llm-router-engine, @reaatech/agent-budget-engine, @reaatech/session-continuity, on Vertex AI · 173 tests / 98.55% coverage.

Azure AI Voice Agent for Freshservice ITSM Ticket Creation

This tutorial wires up an Azure AI-powered voice agent that answers a Twilio number and creates a Freshservice ticket from the caller’s spoken description. It uses Deepgram for speech-to-text and text-to-speech, and Azure AI Inference for understanding the request. Structured repair turns the caller’s words into a valid ticket payload, and confidence routing asks clarifying questions when a category or urgency is ambiguous.

Read the tutorial → · Download the code (zip)

Built with @reaatech/voice-agent-core, @reaatech/voice-agent-stt, @reaatech/voice-agent-tts, @reaatech/voice-agent-telephony, @reaatech/agent-handoff, @reaatech/structured-repair-core, on Azure AI · 85 tests / 96.26% coverage.

AI Discovery Doc Review for Small Litigation Firm

Built for the 5-attorney firm handling a $50k case, this pipeline automates document review so cases don't go pro se. It pulls text from PDFs, generates summaries and privilege flags, and compares results against golden datasets to measure accuracy. Every document's review cost is tracked, making it possible to price small matters profitably.

Read the tutorial → · Download the code (zip)

Built with @reaatech/media-pipeline-mcp-doc-extraction, @reaatech/rag-eval-core, @reaatech/agent-eval-harness-golden, @reaatech/llm-cost-telemetry, @reaatech/mcp-server-tools, on provider-agnostic · 105 tests / 100.00% coverage.

Anthropic Voice Agent for Toast Restaurant Order Taking

Restaurants can now take phone orders 24/7 without staff intervention. This voice agent answers calls on a Twilio number, uses Anthropic's Claude to interpret orders and answer menu questions, and submits the completed order to Toast POS. Session continuity maintains context across multiple utterances (e.g., “medium” vs. “large”), and when a request is too complex, the call hands off to a human.

Read the tutorial → · Download the code (zip)

Built with @reaatech/voice-agent-core, @reaatech/voice-agent-telephony, @reaatech/session-continuity, @reaatech/session-continuity-storage-redis, @reaatech/confidence-router, @reaatech/agent-handoff, @reaatech/llm-cost-telemetry, on Anthropic · 86 tests / 98.50% coverage.

Vertex AI Lead Intake for Jobber Field Service Booking

This recipe gives home service companies a 24/7 lead intake line. When a caller describes their need, Vertex AI classifies the intent (new installation, repair, estimate), extracts contact details, and creates a client and booking request in Jobber automatically. Low-confidence calls are routed to a live dispatcher, and every call's cost is tracked.

Read the tutorial → · Download the code (zip)

Built with @reaatech/voice-agent-core, @reaatech/confidence-router, @reaatech/session-continuity, @reaatech/agent-handoff, @reaatech/llm-cost-telemetry, @reaatech/agent-eval-harness-gate, on Vertex AI · 78 tests / 98.24% coverage.

Databricks LLM Observability for SMB Production AI

This integration wraps Databricks model serving calls to emit OpenTelemetry traces with GenAI semantic conventions. Spans carry per-token cost breakdowns, and everything lands in Langfuse for visualization and alerting. SMB teams get latency and spend dashboards without building their own instrumentation.

Read the tutorial → · Download the code (zip)

Built with @reaatech/otel-genai-semconv-core, @reaatech/otel-genai-semconv-exporters, @reaatech/otel-cost-exporter-core, on Databricks · 72 tests / 97.86% coverage.

Browse all solutions →

Browse the full product catalog at reaatech.com/products.