Road to Autopilot
Building your AI Agents is like onboarding a new team member
Last updated
Was this helpful?
Building your AI Agents is like onboarding a new team member
Last updated
Was this helpful?
In Conduit, Autopilot represents the ultimate goal: an AI capable of handling your communication for you, responding to common inquiries and complex ones too.
By default, your AI begins on Copilot, assisting with responses but always seeking human confirmation before sending. Moving from Copilot to Autopilot is a journey—a gradual process of training, refining, and trusting your AI to take on more responsibilities without needing you to step in.
Imagine the Road to Autopilot like onboarding a new team member to manage your communications. At first, you’d guide them closely, sharing essential details, policies, and specific procedures. Over time, as they learn the intricacies of your brand and processes, they start working independently, handling more situations with minimal oversight. This is exactly how Autopilot and your Knowledge Base work: the more knowledge you provide, the more capable and confident you become in your AI managing communication autonomously.
Progressing toward Autopilot is a process of building, refining, and optimizing your AI. This journey happens in two key phases:
Before your AI can handle guest conversations effectively, it needs the right knowledge. In this phase, you’ll focus on:
Connecting Sources – Import data from your PMS, spreadsheets, PDFs, and other sources to centralize all essential information.
Filling Required Info – Ensure each listing has the structured details needed for complete and accurate responses.
Adding Knowledge – Manually refine and supplement gaps to improve response quality and depth.
This phase ensures your AI has access to the right knowledge, forming the foundation for accurate, reliable communication.
Once the AI has been taught, it’s time to test and fine-tune how it applies that knowledge in real interactions. This phase focuses on:
Sandbox Testing – Act like a guest, ask questions, and adjust knowledge blocks to refine AI responses.
Inbox Proposed Replies – Review AI-generated responses in live conversations, modify sources, and add missing knowledge directly from the inbox.
Advanced Rules – Create AI rules to guide responses in more complex situations, ensuring consistency in decision-making and tone.
By continuously teaching and tuning, you’re shaping your AI into a smarter, more independent assistant, moving it toward Autopilot—where it can handle guest communication confidently without your intervention.
Sources
Required Info
Sandbox
