Autonomous Collections

The problem

A decade-old product fighting its own users

Collections analysts chase overdue invoices all day: find who owes, work the risky accounts first, make the calls, and log every promise to pay. The tool meant to help was designed almost ten years earlier. It was precise with numbers but blind to insight, cluttered with flashy elements, and built on an aging Ext JS stack that capped what the experience could become. So I treated this as a fresh product, not a reskin, and named it CLS 2.0.

Research

Listening before redrawing

I started where the friction lived. I ran user interviews and contextual inquiries, read product analytics, and studied competitors to separate real needs from assumptions. The principle we held onto: design for the user, not for ourselves. A few findings set the direction.

• Outdated and flashy

  Almost everyone described the look and feel as dated and poorly structured, with type and elements that felt too loud for all-day use.

• 89% wanted more at a glance

  Nearly nine in ten asked for more of the data points they need to finish daily tasks, surfaced up front instead of dug out.

• 76% wanted to work in place

  Three in four wanted lightweight tools, like notes, inside the product rather than scattered across sticky notes and calendars.

• An aging platform

  The legacy stack limited performance and interaction, which made a ground-up rebuild the honest answer.

The redesign principles

Assist, enable, motivate, manage, measure

From the opportunities we mapped, the product made a single promise: take the busywork off the analyst so they can focus on the conversation that actually recovers cash. AI reads the raw data and surfaces it thoughtfully, the interface stays calm and predictable, and progress is always visible. Those goals became five design principles the whole system answers to.

Information architecture

One spine for the whole journey

Before any pixels, I stitched the flow together so navigation stayed obvious from login to wrap-up. A workboard opens the day, a worklist orders the work, every customer rolls up into one detail spine (overview, invoices, payments, communications, details), and calls resolve into a post-call wrap-up and summary, then the loop repeats.

The product, screen by screen

From a busy table to a guided day

The redesign is ten core screens working as one system. It opens on a command center, narrows to a prioritized worklist, then drills into a full customer 360 and the tools to work a call end to end.

Freeda, voice & touch

An assistant built for how people actually work

Analysts increasingly work on touch-capable laptops and want to move without hunting through menus. So the product was optimized for a touch-and-voice framework, and given its own assistant, Freeda. Ask in plain language, like "show me all customers with past due over ten thousand dollars," and the worklist filters itself; Freeda can open an account or kick off a task, by voice or by tap.

Craft decisions

The details that made it usable

• Insight over raw data

  Periodic trends and behavioral signals surfaced as concise, AI-driven findings, so analysts decide faster instead of reading tables.

• CTAs with room to breathe

  Primary actions sit where the task expects them, with negative space around them, making the flow predictable and quick.

• Readable in the dark

  A dark base drawn from the brand blue and black, tuned to hold a 4.5:1 contrast minimum so long sessions do not strain the eyes.

• A linear path

  The journey simplified toward a near-straight line, cutting fuss at each step and clarifying the critical touchpoints.

Visual system

A calm, dark-first palette

Analysts live in this product, so the visual system had to feel fresh without tiring the eyes. I derived a dark base from the brand's blue and black, lifted surfaces with a 5% wash of bluish white for elevation, and reserved a warm orange for the actions that matter most. Type is Ubuntu, set for legibility at small sizes across dense screens.

Outcome

A product analysts could actually live in

The collections product was fully redesigned: a refined interaction model, new functionality, resolved accessibility issues, and AI-driven insight woven through. Disparate tools, the sticky notes, reminders, and calendars, folded into one workspace, so analysts could focus on the work instead of the workaround.

1. Replatform

   Rebuilt a decade-old product as CLS 2.0 on a modern stack, redesigning experience and interface together.

2. Consolidate

   Folded notes, reminders, and calls into one screen, removing the disparate side-tools analysts relied on.

3. Recover faster

   HighRadius reported up to 75% faster receivables recovery in a 2021 survey, with stronger automated dunning and calls.

4. Land well

   The redesign was warmly received across global collections teams.

There may be a better solution out there, but I designed an experience that works well for the people who use it every day, and that is the part I am proudest of.