The Lab
The Lab is where you build your AI Twin. Maya, your AI Success Specialist, guides you through the entire process — from your initial interview to publishing your trained Twin. No technical knowledge required.
Maya
Maya lives in the Lab and walks you through everything you need to create a high-quality Twin. You can chat with Maya by typing, or start a voice call for a more natural conversation.
Maya handles the entire build process: interviewing you, generating your Twin’s Guidelines, training your Twin, and helping you iterate until it’s ready to publish.
The Interview
When you first enter the Lab, Maya interviews you to understand who you are and how your Twin should behave. She’ll ask about:
- Your name, background, and bio
- Your business and what you do
- Who you’re building the Twin for (yourself or someone else)
- The purpose of your AI Twin
- Your target audience
- Your primary language
- How you think and communicate
- Your voice and tone, with examples
- Topic boundaries — what your Twin should not discuss
- Ethical boundaries and disclaimers
- Language preferences and phrases to avoid
- Example questions your users will ask
- Conversation style and response length preferences
The interview is conversational — Maya adapts based on your answers and guides you through each topic naturally.
Guidelines
After your interview, Maya generates your Twin’s Guidelines — a structured summary of everything she learned about you. This takes less than 30 seconds.
Review your Guidelines before training begins. You’ll see exactly what Maya captured, so you can confirm it’s accurate or ask her to make changes.
Training
When you’re satisfied with your Guidelines, tell Maya to start training your Twin. Training uses both your interview responses and any content you’ve uploaded to your Knowledge Base — including speaking patterns, Q&A examples, and coaching style.
- First training run — about 15–20 minutes
- Iterations — less than 5 minutes each
You can watch training progress in real time with status updates as your Twin is built.
Questions
If training doesn’t reach 100%, Maya will surface Questions for you to answer. These help resolve areas where your Twin needs more clarity or detail. Answer them and training continues.
Publishing
Once training is complete, publish your Twin to make it live and available to your users.
Iterating
After publishing, test your Twin and provide feedback. Maya uses your feedback to update the Guidelines and retrain — each iteration is faster than the last.
The Lab supports a continuous cycle: Interview → Guidelines → Train → Publish → Test → Iterate
Questions? Contact support@steno.ai.