# How Relevance AI Runs Support Agents: Copilots → Autopilot at Scale Source: https://www.youtube.com/watch?v=wd6P9eUDQOk Recap page: https://rapidrecap.app/video/wd6P9eUDQOk Generated: 2026-01-15T06:33:36.072+00:00 --- ## Quick Overview The presentation concludes that support teams should start by building AI co-pilots to identify and fix issues in real-time before implementing fully autonomous auto-pilot agents, citing past negative experiences with auto-pilots that hallucinated and lacked necessary human oversight, leading to the recommendation to empower subject matter experts (SMEs) on the support team to build these initial co-pilot agents. **Key Points:** - The critical decision when building AI support agents is whether to start with auto-pilot or co-pilot, with most teams incorrectly choosing auto-pilot first. - Early auto-pilot attempts failed due to hallucinations (incorrect information without guardrails), lack of support SME thinking, and no escalation pathways, resulting in lost customers and declining CSAT. - Co-pilots, like the agent 'Amy', require human approval before sending responses, allowing teams to identify and fix issues before moving to autonomous systems. - Relevance AI promotes the right approach: Agent builders equal the support team, empowering SMEs to build agents without engineering involvement, prioritizing quality before scaling to auto-pilots. - The evolution to 'Jack the Auto-Pilot Agent' involved moving Amy's refined tooling to Jack, scaling auto-pilot in two weeks, deploying across free and paid tiers, and establishing clear guardrails and escalation rules. - Co-pilots are essential for revealing process gaps in real-time, as demonstrated by a customer asking about available LLMs, which Amy (co-pilot) couldn't answer, leading the team to build a tool to address the gap immediately. - The final takeaway encourages committing to learning and evolving in 2025 by using Relevance AI, joining the community, and becoming an AI Agent Builder. ![Screenshot at 00:19: The slide titled 'SUPPORT AGENTS: AUTO-PILOT VS CO-PILOT' establishes the central theme of the presentation, contrasting the two AI implementation strategies for support teams.](https://ss.rapidrecap.app/screens/wd6P9eUDQOk/00-00-19.jpg) **Context:** The presentation, delivered by Alex Witts of Relevance AI, addresses the strategic choice between deploying fully autonomous AI support agents (auto-pilot) versus human-in-the-loop assistants (co-pilot). Alex shares personal experience from building support workflows for a large tech company with 1,500 support agents, emphasizing the necessity of a co-pilot approach before scaling to full automation to ensure quality and avoid costly errors. ## Detailed Analysis The presentation argues against immediately implementing fully autonomous AI support agents (auto-pilot) for complex use cases, advocating instead for a co-pilot approach first. Alex recounts past failures where an initial auto-pilot agent, 'Sunny', hallucinated answers, lacked subject matter expertise, and had no escalation pathways, leading to lost customers and declining CSAT scores. The key takeaway is to 'Stop jumping to auto-pilot' for risky use cases. The recommended strategy involves empowering Subject Matter Experts (SMEs) on the support team to become 'AI Agent Builders' using tools like Relevance AI, ensuring quality before scaling. The evolution from the co-pilot 'Amy' to the auto-pilot 'Jack' shows successful scaling in two weeks after establishing clear guardrails and escalation rules. A real-world example highlights how co-pilots catch gaps: when a customer asked about available LLMs, Amy (co-pilot) flagged the gap because she lacked the tool to answer, allowing the team to build the necessary tool immediately. The presentation concludes by emphasizing that support teams are evolving into AI Agent Builders, managing AI teammates and orchestrating more tickets, and encourages the audience to commit to learning AI by starting with Relevance AI, joining the community, and becoming an AI Agent Builder in 2025. ### Introduction and Core Conflict - For those new to Relevance AI, Alex introduces herself as leading the support team, leading to the core conflict: Support Agents: Auto-Pilot vs Co-Pilot - Alex emphasizes that most teams choose the risky auto-pilot route first and fail. ### Co-Pilot Definition and Examples (Amy) - Co-pilot is defined as 'Human-in-the-loop, approval required,' contrasting with Auto-pilot's 'Fully autonomous, zero human approval' (e.g., self-driving cars, autonomous chatbots) - Co-pilot examples include documentation drafter, feature request reporter, case reviewer, community content creator, and response reviewer. ### Failures of Auto-Pilot First - The initial auto-pilot agent, 'Sunny,' hallucinated, lacked support SME thinking, and had no escalation pathways, resulting in lost customers and declining CSAT. - This led to the realization that co-pilots are necessary for identifying issues before full automation. ### The Right Approach - Agent builders equal the support team, empowering SMEs to build agents without engineering overhead, focusing on 'Quality before scale' (co-pilots first, then auto-pilots). - The speaker cites her previous company's experience where support agents handled ticket work while building agents. ### Co-Pilots Reveal Gaps in Real-Time (Example) - A customer asked about available LLMs; Amy (co-pilot) couldn't answer, flagging a gap that was immediately fixed by building an endpoint query tool. - This shows co-pilots allow for iterative refinement before full autonomy. ### Evolution to Auto-Pilot (Jack) - Amy's refined tooling was moved to 'Jack the Auto-Pilot Agent,' which scaled rapidly in two weeks across free and paid tiers with clear guardrails and escalation rules. - Jack handles high-volume, low-risk tasks like account deletions automatically, while complex issues are escalated. ### Final Takeaways - 1. Stop jumping to auto-pilot first, as risky use cases often fail. 2. Empower support teams (SMEs) with no-code tools to build agents. 3. Use co-pilots to find gaps in real-time before scaling. - Commitment for 2025: Start using Relevance AI, join the community, and become an AI Agent Builder. ![Screenshot at 00:19: The slide titled 'SUPPORT AGENTS: AUTO-PILOT VS CO-PILOT' establishes the central theme of the presentation, contrasting the two AI implementation strategies for support teams.](https://ss.rapidrecap.app/screens/wd6P9eUDQOk/00-00-19.jpg) ![Screenshot at 01:49: The slide 'The critical decision' highlights that most teams choose the auto-pilot route first, which often leads to failure due to hallucinations and lack of human oversight.](https://ss.rapidrecap.app/screens/wd6P9eUDQOk/00-01-49.jpg) ![Screenshot at 02:01: The slide 'Understanding the Difference' clearly defines Auto-pilot as fully autonomous with zero human approval, versus Co-pilot requiring human-in-the-loop approval.](https://ss.rapidrecap.app/screens/wd6P9eUDQOk/00-02-01.jpg) ![Screenshot at 04:16: The slide 'Co-Pilot: Your AI Teammate' lists examples of co-pilot roles like Documentation drafter and Case reviewer, emphasizing human approval before sending responses.](https://ss.rapidrecap.app/screens/wd6P9eUDQOk/00-04-16.jpg) ![Screenshot at 06:16: The slide 'Auto-Pilot: Autonomous Operations' lists examples of fully autonomous tasks, noting that customer-facing support agents always run on auto-pilot.](https://ss.rapidrecap.app/screens/wd6P9eUDQOk/00-06-16.jpg)