If you're energized by a fast pace, high standards, and high accountability for yourself and others, you'll thrive here and we encourage you to apply.
15Five's customer support runs on an AI-native operating model: our AI agent (Fin) resolves routine customer questions, AI assistance lives inside every support tier, and a closed-loop content system turns every wrong or missing AI answer into a published fix. The AI Support Manager role operates that system; the percentages under What You'll Own are approximate shares of time. You own the day-to-day performance of our customer-facing AI agent, the Help Center content it draws from, which you own end to end, and the feedback loop that makes both better every week.
This is an operator role with clear boundaries, real authority inside them, and a defined escalation path for everything outside them. You will be the expert on how our support AI behaves and why. You will not be a lone hero: the role is built around documented runbooks, a shared intake queue, and a named second for every critical system.
What You'll Own
Fin/AI agent operations (25%): our customer-facing AI agent resolves routine questions correctly and the quality-adjusted automation rate holds or improves month over month. Owns resolution quality, deflection with quality gates, topic configuration, prompt and behavior tuning, and escalation-to-human design.
Help Center content ownership (40%): accuracy and AI-readiness of the Help Center as the shared knowledge layer feeding customer self-service and every AI surface. This role is the owner of every published article, where ownership means accountability for every published word, not typing them: AI drafts, this role structures, verifies, and publishes. Owns the taxonomy, article templates, and AI-retrieval writing standards, the Guru collections, the drafting pipeline, and the verify-and-publish workflow with Tier 1. Nothing ships to the knowledge layer without passing this role’s quality bar.
The content flywheel (20%): every detected wrong or missing AI answer becomes a published fix. Triage AI-failure reports from the shared intake queue, trace each to root cause (retrieval failure, content defect, or product-truth gap), route product-truth gaps to their owner, and report gap-closure velocity.
Launch readiness (10%): no release ships without its support content ready. Hold the support seat in release planning and prepare Help Center articles, Fin topics, and frontline teams before go-live.
- Operability (5%): no single point of failure. Maintain the runbooks, configuration documentation, and measurement baselines, and keep a trained second for every critical system. Documentation is part of done.
- Where This Role Sits
- This role is the support side of a larger AI ecosystem. It works best when the seams are clean, so here is who owns what around it:
R&D owns the AI knowledge-base architecture and the in-product AI (Amaya) knowledge stores, with a separate role scoped to that work. This role is their closest consumer: it keeps the knowledge layer they read from accurate and feeds architecture gaps back to them.
Leadership owns org-wide AI-native strategy and internal AI consolidation. This role informs those calls with ground truth from the operation.
Support tier leads own ticketing platform administration and support workflows. This role designs the AI handoffs that land in them.
This role owns the Guru collections and keeps them accurate and AI-ready, since Guru's AI answers from them.
- New cross-team AI initiatives start with your manager, so they get prioritized before work begins.
- How Success Is Measured
- Quality-adjusted automation rate: AI resolutions with no human takeover, no negative feedback, no reopen.
Knowledge gap closure rate: speed from detected wrong/missing answer to published fix, weighted by impact. Top articles are proactively owned and current, not just reactively fixed.
Escalation precision: when the AI hands off to a human, it lands in the right place with full context.
Operability: every critical system has current runbooks and a functioning second; no single point of failure.
- Partner trust: tier leads, content, and product partners would say this function makes their work easier.
- What We're Looking For
- 3–5+ years in support operations, CX operations, knowledge management, or conversational AI operations, with hands-on ownership of a production AI agent (Intercom Fin strongly preferred; Ada, Forethought, Decagon, or similar also relevant).
A proven content engineer: customer-facing help content you owned end to end, plus taxonomy, content verification workflows, structured authoring, and writing for AI retrieval, not just human readers.
An experimenter's method: hypothesize, test small, measure, then scale. Comfortable with AI-assisted workflows as a daily practice.
Strong analytical habits: you can trace a wrong AI answer through retrieval, content, and product behavior to root cause, and you show your work.
Collaboration as a first-class skill: you raise concerns early, directly, and through agreed channels; you disagree openly, commit fully, and treat peers' domains with the respect you expect for yours.
You build systems, not dependencies: your instinct when you become the only person who can do something is to document it and train a second.
Working Style
This role touches systems that many teams depend on, which makes judgment about scope as important as technical skill. The people who thrive in it:
check in before investing significant time in adjacent work,
surface risks to their manager as crisp recommendations rather than broad escalations, and
measure their impact by outcomes the team ships together rather than by territory.
We are explicit about this because the role's leverage is real: operated with focus and trust to deliver meaningful impact.
