Back to work
NextIQNextiva2024 — 2026

A layered AI copilot that coaches agents in real time

NextIQ sits beside the agent while they talk to a customer — surfacing knowledge, suggesting responses, and reading sentiment as the conversation unfolds. I designed it as a layered experience: a passive assistant, inline recommendations, and a supervised Autopilot, so teams could dial up automation at their own pace with a person always in control.

RoleProduct Designer — end-to-end
TeamProduct, Engineering, Data
SurfacesLive coaching panel, dashboard

The problem

Agents juggle live conversations while hunting through knowledge bases, second-guessing the right response, and writing up notes afterward. New hires take months to ramp. The opportunity was an AI copilot that delivered expert-level guidance in the moment — without taking the keyboard away from the person or eroding their trust in the system.

A layered experience

  • The assistant. A passive coaching panel running alongside the conversation — never an intrusive overlay. It listens, stays out of the way, and is there when the agent looks for it.
  • Inline recommendations. Suggested responses, relevant knowledge articles, and next steps surfaced the moment they're useful — each one traceable to an approved source, so suggestions are grounded rather than guessed. The agent decides whether to use, edit, or ignore.
  • Supervised Autopilot. The agent can let NextIQ take routine work end-to-end — context retrieval, summaries, note-taking — while supervisors monitor, adjust, and audit every action. Automation scales up only as trust does.

Designing for trust

The core tension was control: how much should the AI do, and how visible should it be? I designed clear states for monitoring, suggesting, and acting, plus real-time signals — sentiment shifts, and sensitive-data prompts that redirect payments to secure channels. Every suggestion and action is logged and reviewable, giving supervisors a full audit trail and agents the confidence to lean on the tool.

Approach

I owned the work from discovery through pixel-level UI and dev-ready prototypes, partnering closely with product, engineering, and data. Components were built on the team's token-based design system, and I prototyped the live coaching interactions in code so stakeholders could feel how the copilot behaved mid-conversation, not just in static screens.

What it delivers

  • 3Layers, from assist to Autopilot
  • Real-timeCoaching & sentiment
  • AuditedEvery action supervised

Designed as a human-in-the-loop copilot — faster ramp for new agents, less after-call work, and expert guidance on every conversation. Full case study and process walkthrough available on request.

Like what you see? Let's build something thoughtful.

I'm open to Senior, Staff & Lead product design roles at AI-first and enterprise SaaS companies. Happy to walk you through the full process behind this work.

Get in touch