Voodoo Oracle

An AI-powered assistant trained on Voodoo Robotics documentation, APIs, and best practices.

What Is the Voodoo Oracle?

The Voodoo Oracle is a custom ChatGPT-powered assistant built specifically for Voodoo Robotics integrators, developers, and warehouse engineers. It has been trained on our complete documentation set — API references, best practices, deployment guides, and common troubleshooting scenarios.

Think of it as an always-available first-line resource that can answer questions about:

  • API endpoint usage, parameters, and examples
  • Integration patterns and code snippets (Python, curl, JavaScript)
  • Turbo configuration, placement, and troubleshooting
  • Server deployment options (www.voodoodevices.com vs Big Block)
  • Best practices for pick-to-light, put-to-light, and sequencing workflows
  • Common error codes and their resolutions

Launch the Oracle

Voodoo Oracle — AI Assistant

Ask questions about Voodoo APIs, devices, integration patterns, and troubleshooting. Powered by ChatGPT with Voodoo-specific training.

Open Voodoo Oracle

Opens in a new tab. Requires a ChatGPT account (free or paid).

What the Oracle Can Do

Build an Integration

"Build a tested Python service that maps our WMS events to Voodoo commands and can be deployed with systemd."

Explain Concepts

"What's the difference between a Sequence and a direct device update?"

Troubleshoot Issues

"I'm getting a 401 error when calling the locations endpoint. What should I check?"

Recommend Architecture

"What's the best way to integrate Voodoo with our SAP WMS for batch picking?"

From a Question to Production Code

The Oracle can go far beyond explanations and isolated snippets. With enough context and follow-up direction, it can help construct a full, deployable integration. Treat the conversation like an engineering project with requirements, implementation, testing, and review.

  1. Start from existing work. Check the Integration Examples and give the Oracle the closest repository link or relevant files.
  2. Ask it to interview you. Have it identify missing details about WMS events, authentication, locations, quantities, Orders or direct commands, acknowledgements, runtime, hosting, and security constraints before proposing code.
  3. Agree on the design. Request a data-flow diagram, field mappings, failure behavior, acceptance criteria, and a file tree. Correct misunderstandings before generating the project.
  4. Request the whole project. Ask for complete source files, pinned dependencies, environment configuration, tests, logging, retries, idempotency, health checks, deployment files, and a README. Explicitly say that snippets and pseudocode are not sufficient.
  5. Run it and keep going. Execute the tests and use a Voodoo test account. Paste exact failures and logs into the same conversation, ask for focused corrections, and repeat until the behavior matches the acceptance criteria.
  6. Perform a human production review. Verify credentials, permissions, input validation, duplicate events, timeouts, recovery, observability, business rules, and rollback behavior before deployment.

A Strong First Prompt

“Help me build a production Voodoo integration for [WMS/ERP]. Use [repository or documentation link] as the starting point. Before writing code, ask detailed questions about event payloads, authentication, location mapping, quantities, callbacks, failure handling, deployment, and our definition of done. After I answer, propose the architecture and then construct the complete project with tests and deployment instructions.”

What the Oracle Cannot Do

Important Limitations

The Voodoo Oracle is an AI assistant, not a substitute for official support. Keep these limitations in mind:
  • No account access — The Oracle cannot look up your account, devices, or API keys. It has no access to your www.voodoodevices.com data.
  • Do not share secrets — Never paste API keys, passwords, access tokens, private customer data, or other sensitive information into the conversation. Use sanitized payloads and placeholder values.
  • No live system interaction — It cannot send commands to your Devices, check device status, or modify your configuration.
  • Not always current — The Oracle's training data may not reflect the very latest API changes or features. Always verify against the official documentation.
  • May produce errors — Like all AI systems, the Oracle can generate incorrect code or outdated information. Test all code before deploying to production.
  • No SLA — The Oracle is provided as-is with no guaranteed uptime or response accuracy. For production-critical support, contact the support team.

Tips for Best Results

  1. Be specific — Instead of "How do I use the API?", ask "How do I create a Sequence with 3 steps using the REST API in Python?"
  2. Provide context — Tell the Oracle what WMS/ERP you're using, your integration method (REST vs GET), and what you've already tried.
  3. Ask for the complete project — Specify the language and runtime, then request every source file, dependency, test, configuration template, and deployment instruction needed to run it.
  4. Iterate with evidence — Paste exact errors, API responses, logs, and failing tests. Ask for the smallest correction and an additional test that proves the problem is fixed.
  5. Request a production review — Ask the Oracle to inspect the finished project for missing retries, idempotency, secret handling, logging, health checks, tests, and rollback steps.
  6. Verify — Always test generated code against a non-production environment before deploying.

Tip

The Oracle works best for "how do I" questions. For "why isn't this working" questions about your specific deployment, contact support — they can access logs and system state.