Daily recap for June 15, 2026

Small medical practices can now auto-extract EOBs and insurance claims from PDFs with a production-grade pipeline, thanks to our new tutorial. That’s one of nine solutions we published today, each with a ready-to-run codebase and a full walkthrough.

What we shipped

Azure AI Document Pipeline for SMB Medical Claim Processing

For medical billing teams and small practices, this pipeline ingests EOB and claim PDFs, uses Azure Document Intelligence to extract raw data, and applies a multi-stage process to fix malformed LLM outputs with structured repair. Low-confidence extractions get routed to a human review queue, and cost-aware caching avoids re-processing similar documents. It’s a turnkey way to cut administrative overhead and reduce errors.

Read the tutorial → Download the code (zip)

Built with @reaatech/structured-repair-core, @reaatech/confidence-router-core, @reaatech/llm-cost-telemetry, @reaatech/llm-cache, @reaatech/llm-cache-adapters-redis, @reaatech/agent-budget-engine, on Azure AI · 98 tests, 95.52% coverage.

Anthropic AI Runbook Automation for SMB DevOps Incident Recovery

Small teams without dedicated DevOps can now generate and test incident runbooks automatically. This solution scans your service repos, builds structured runbooks with alert definitions, runs chaos scenarios to validate recovery steps, and summarizes findings with Anthropic Claude. Dirty workflows and DynamoDB state keep everything durable. If your AI agents fail, this gets them back online fast.

Read the tutorial → Download the code (zip)

Built with @reaatech/agent-runbook, @reaatech/agent-runbook-cli, @reaatech/agent-runbook-alerts, @reaatech/agent-runbook-analyzer, @reaatech/agent-chaos-scenarios, @reaatech/session-continuity-storage-dynamodb, on Anthropic · 74 tests, 95.83% coverage.

Vertex AI Sandboxed Data Analysis with Budget Guardrails for SMBs

Analysts can now generate and run data analysis code safely inside a sandbox, with per-session budgets enforced. This recipe uses Vertex AI for model routing, structured repair to fix malformed code, and an E2B sandbox for execution. Spend limits and cost tracking keep you from blowing the budget on a runaway query.

Read the tutorial → Download the code (zip)

Built with @reaatech/agent-budget-engine, @reaatech/llm-router-engine, @reaatech/session-continuity, @reaatech/structured-repair-core, on Vertex AI · 110 tests, 100.00% coverage.

Cohere Insurance Policy Data Extraction for SMB Brokers

Independent insurance brokers can stop re-keying policy data. This pipeline uploads a PDF, parses it with OCR when needed, extracts structured fields with Cohere’s LLM, repairs any malformed JSON, tracks spend per broker, and enforces monthly budgets. It’s a single API that replaces hours of manual entry.

Read the tutorial → Download the code (zip)

Built with @reaatech/media-pipeline-mcp-doc-extraction, @reaatech/structured-repair-core, @reaatech/llm-cost-telemetry-aggregation, @reaatech/agent-budget-engine, on Cohere · 77 tests, 98.83% coverage.

Automated Referral Letter Generator from Veterinary Medical Records

Veterinary practices can draft specialist referral letters in under two minutes. The system reads a PIMS record (PDF or DOCX), extracts relevant clinical data, and formats a complete referral letter using AI. Agent memory and context-window planning keep the output accurate, while guardrails ensure no critical detail is missed.

Read the tutorial → Download the code (zip)

Built with @reaatech/agent-memory, @reaatech/agent-memory-embedding, @reaatech/agent-memory-retrieval, @reaatech/llm-cache, @reaatech/guardrail-chain, @reaatech/context-window-planner, on agnostic provider · 72 tests, 99.24% coverage.

After-Hours Pet Emergency Triage Agent for Veterinary Clinics

Stop after-hours burnout. This voice agent answers client calls outside clinic hours, asks structured questions, transcribes responses with speech-to-text, and uses an LLM to assess urgency. It speaks back a severity score and flags critical cases for immediate vet callback, all over a Twilio phone call.

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-mesh, @reaatech/llm-router-core, on agnostic provider · 102 tests, 95.63% coverage.

Google Gemini RAG Product Search for Square Online SMB Stores

Square Online merchants can replace their default keyword search with a natural-language product finder. The solution indexes your catalog into a Qdrant vector database and answers queries like “red sneakers size 10” using Gemini. Semantic caching, cost-aware model routing, and conversation state keep the answers relevant and affordable.

Read the tutorial → Download the code (zip)

Built with @reaatech/llm-cache, @reaatech/llm-cache-adapters-qdrant, @reaatech/llm-router-core, @reaatech/llm-router-engine, @reaatech/session-continuity, on Google Gemini · 112 tests, 96.01% coverage.

Code Sandbox for Jira Service Management SMB IT Teams

Small IT teams can run custom incident-trend analysis scripts directly from Jira tickets. A Jira webhook triggers an isolated E2B sandbox to execute Python, with a circuit breaker to stop runaway scripts and a budget engine to cap per-user costs. Structured repair validates outputs before posting results back to the ticket.

Read the tutorial → Download the code (zip)

Built with @reaatech/agent-budget-engine, @reaatech/circuit-breaker-core, @reaatech/structured-repair-core, on agnostic provider · 71 tests, 95.60% coverage.

Mistral AI Security Guardrails for Etsy SMB Marketplace Messaging

Etsy sellers can intercept incoming buyer messages and scan them for PII, abuse, and phishing before they reach the inbox. A guardrail chain uses Mistral’s moderation prompt and Presidio for PII redaction, then quarantines high-risk messages. Budget tracking keeps the cost predictable, and a dashboard lets you review flagged messages.

Read the tutorial → Download the code (zip)

Built with @reaatech/guardrail-chain, @reaatech/guardrail-chain-config, @reaatech/guardrail-chain-guardrails, @reaatech/agent-budget-engine, on Mistral AI · 73 tests, 96.25% coverage.

Browse all solutions →

Browse the full catalog at reaatech.com/products.