How we run product

Product managers at Intempt are builders, not coordinators. If you only manage Jira tickets and schedule meetings, this is not the role. PMs here write specs engineers can build from without back-and-forth, prototype in Lovable before engineering picks anything up, and evaluate AI output quality across the product.

🛠️ What product owns

• Specs engineers can build from without ambiguity. Design and UX decisions
• AI prompts -- write, iterate, evaluate output quality, flag degradation. Prototype in Lovable before engineering touches it
• Product metrics -- adoption, retention, engagement, not just ship dates
• Incident war room and stakeholder communication during Critical and Major incidents

🤖 AI-native product work

PMs cowork with Claude on claude.ai for spec writing, user research synthesis, and competitive analysis. Prototypes are built in Lovable before engineering touches them. AI quality evaluations: scoring outputs on relevance, accuracy, format, and helpfulness against defined rubrics. Every AI feature ships with a quality bar. If the model output degrades, product catches it first.

🔥 Incident role

During Critical and Major incidents, product is in the war room from minute zero. PMs own status page updates, stakeholder communication, and post-incident summaries. This is organizational judgment, not a support role.

🎧 Customer support

Customer support is AI-first. Product handles escalated tickets that AI cannot resolve -- typically complex product questions, billing edge cases, or account-level issues. Product reviews unresolved tickets as part of normal work, not as an additional burden.