About the role
About Dust We're creating a new AI operating system that has the potential to change the way companies operate . Our mission at Dust is to transform how work gets done by letting any team and employee shape the exact agents they need to accelerate their work. With 70%+ weekly active users, people stick with Dust as much as they do with Slack and Notion. We already are a core part of their jobs. We're at an exciting stage of our journey and growing fast. We're serving great customers like Cursor, Clay, Whatnot, and Persona, and aim to 5x our growth by the end of 2026. Dust is a Sequoia-backed company with an experienced and determined team of optimists (coming from companies like Stripe and OpenAI) that likes to focus on users, getting great things done by shipping fast, and doesn’t take itself too seriously while doing so. TLDR Dust is building the AI operating system that transforms how companies work, already used weekly by 70%+ of our customers' teams. We're looking for a Support Engineer to define what AI-native support looks like in practice.
The role The Customer Success team at Dust partners with our customers to transform how they work with AI. We're looking for a Support Engineer to work alongside our support agents to deliver fast, precise, and delightful user experiences. Your
responsibilities will center on three pillars: quality assurance of agent-generated responses until the vast majority require minimal human intervention, knowledge gap closure by ensuring missing information is systematically documented, and debugging issues through to a well-defined hypothesis. You will define what AI Agentic support looks like in practice.
What you'll do Resolve our users issues: Investigate issues across historical patterns, logs, code, and our back-office tooling. Develop hypotheses, identify root causes, and provide clear answers. Quality assurance of conversations: Systematically analyze agent responses for inconsistencies, redundancies, and ambiguities. Continuously iterate on prompts, documentation, and tooling until human intervention is minimal. Knowledge gap identification and closure: When the agent fails due to missing or incorrect information in documentation, code, or internal resources, ensure the appropriate stakeholders document the missing knowledge. Maintain accountability for closing gaps through persistent, collaborative follow-up. Inbox coverage: Handle complex cases escalated by the agent. Produce responses that are simple, legible, obvious, short, and self-explanatory. Function development: Contribute to the tooling, automations, and prompt engineering that improve agent performance. The objective is a single, elegant system capable of handling the majority of support interactions.