Vedika

Design

About Me

Resume

Coreply

Let the AI Think. Let the Agent Lead.

Designing Coreply an AI-enhanced support dashboard that helps agents respond faster with trustworthy, context-aware suggestions.

Timeline

May 2025 - August 2025

Platform

Desktop

Overview

Coreply is an AI-powered support dashboard built to help customer service agents respond faster and more accurately.
It combines smart reply suggestions, auto-generated customer summaries, and a streamlined resolution form — all in one focused workspace.
Designed with clarity, speed, and trust in mind, Coreply transforms traditional support tools into intuitive AI co-pilots.

The why

Support is stuck.

Agents lose precious time rewriting replies and jumping across tabs to find context. Existing dashboards are cluttered, slow, and built for backends — not people.

The solve

AI that assists, not replaces.

Coreply brings together smart reply suggestions, instant customer insights, and a streamlined resolution flow — all within one clean, assistive workspace that never takes over, only helps.

The impact

Faster replies. Smarter workflows. Happier agents.

Coreply transforms AI from novelty to necessity reducing response time, improving consistency, and making customer support feel like a conversation again.

My Role & Approach

I led the full design process — from defining the user problem, to mapping workflows, prototyping interactions, and designing the final UI.
My goal was to design with real support behavior in mind — not just what the tools do, but how people work under pressure.

Research & Discovery

Understanding the Problem

Since this was a self-initiated project, I didn’t have access to live support agents — but I still wanted to ground my design in reality.
I started by analyzing job postings, support tool reviews, UX case studies, and YouTube videos from actual customer support agents.
My goal was to understand not just what they do, but how they think and feel during a busy shift.

From this, I mapped patterns across 3 key areas:

Common responsibilities

Most agents handle 2–4 chats at once while juggling internal tools, macros, and forms. They’re expected to respond quickly and accurately — under pressure.

Workflow constraints

They work within strict SLAs (Service Level Agreements), often without time to personalize replies or double-check info — which leads to errors and burnout.

Frustrations

Agents often rewrite the same answers, hunt for context in multiple tabs, and rely on outdated macros that feel robotic. AI suggestions feel impersonal or unreliable.

I then distilled these insights into a focused persona:
Emily Reyes — Tier 1 Agent, juggling multiple chats, aiming for speed, and craving tools that feel like true assistants, not barriers.

This persona guided every design decision from the layout of the reply panel to the tone of the AI suggestions.

Emily Reyes

Tier 1 Customer Support Agent | 26 | 2 years in a hybrid support center

"I just want to get through my queue faster without sounding robotic"

Goals

Respond to Customers Faster
Understand the Customer’s Context Quickly
Avoid Mistakes While Multitasking
Stay On Top of Her Queue
Feel in Control of the AI

Frustrations

Repetitive tickets
Switching tabs for context
High response time pressure

Competative Analysis

Existing tools are fast but not agent-first.

Before designing Coreply, I explored platforms like Zendesk, Intercom, and Freshdesk to understand how AI is currently integrated into support tools. While these platforms offer powerful automation and macros, most of them fall short in supporting real-time agent thinking, clarity, and trust.

I wanted to understand what worked, what didn’t, and where Coreply could stand out as a better co-pilot for human agents.

Insight 1

Dashboards are system-oriented not agent-first

Most tools are built around backend logic and ticket management — not the agent’s mental model. Agents end up switching tabs and guessing where to find relevant info.

Insight 3

Automation ≠ real assistance

Macros and auto-responses help with speed, but they aren’t flexible. They don’t adapt to tone, intent, or the nuances of live conversations — making agents sound robotic.

Insight 2

AI suggestions lack transparency

AI replies often appear without clear context or explanation. Agents don’t know where suggestions come from or whether they can trust them — leading to confusion or rejection.

Insight 4

Cognitive overload is rarely addressed

Agents juggle 3–4 chats at once, but existing UIs are cluttered, text-heavy, and poorly prioritized. Important tasks compete for attention, increasing errors and stress.

Define

Combining the above insights into four high-level goals

Based on user research, competitive analysis, and support agent workflows, I outlined four design goals to shape the experience of Coreply:

Goal 1

Reduce cognitive load

Design a focused, clutter-free interface where agents can access everything they need — without switching tabs or losing context.

Goal 2

Build trust in AI

Ensure suggestions are clearly sourced, editable, and never auto-sent — giving agents confidence and control in every interaction.

Goal 3

Keep replies flexible, not robotic

Let agents personalize AI-suggested responses with ease — preserving their tone while still saving time.

Goal 4

Support real-time workflows

Design for speed and responsiveness — enabling agents to resolve issues faster while maintaining clarity under pressure.

Ideate

Exploring Layouts That Empower Support Agents

To structure the dashboard in a way that supports real-time decision-making and minimizes cognitive load, I explored 3 distinct layout directions. I mapped the pros and cons of each and ultimately chose the one that balanced clarity, speed, and familiarity.

Classic 3-Panel

This layout balanced usability and information density best. It gave agents full visibility without extra interaction cost — ideal for fast-paced environments.