< Home

Enterprise SaaS Platform

Dashboard UX

AI Product Design

AI-Powered Coaching Experience for Call Center Teams

A feature inside a post-call speech analytics platform that helps team leads deliver scalable, structured coaching through AI-driven coaching modules and guided 1:1 sessions.

ROLE

Product Designer

TIMELINE

Jan 2026-

Mar 2026

TEAM

Product Manager

Frontend Developer

Backend Developer

SKILLS

UX Research

UI Design

Interaction Design

Usability Testing

ROLE

Product Designer

TEAM

Product Manager

Frontend Developer

Backend Developer

TIMELINE

Jan 2026-Mar 2026

SKILLS

UX Research

UI Design

Interaction Design

Usability Testing

About Product

Lingo is an AI-powered speech analytics platform by Sandlogic Technologies. It analyzes call center conversations to surface customer sentiment, agent performance metrics, and complaint trends for call center agents, team leads, managers, and CXOs.

While Lingo does the analytical heavy lifting, it stops at insights -

There's no path from "this agent is struggling" to actually doing something about it. This project explores what it looks like when Lingo closes that gap.

* Due to an NDA with Sandlogic Technologies, the screens shown here are reproduced for portfolio purposes only and do not represent Lingo's proprietary UI.

The Context

In large call center teams, coaching is triggered reactively - when CSAT scores drop, escalations spike, or communication errors pile up. The process is almost entirely manual: team leads review call recordings and discuss mistakes in 1:1 sessions.

With a single team lead managing anywhere from 20 to 50 agents, this model breaks down fast.

High-risk agents get attention. Average performers don't. And by the time feedback reaches anyone, the call it's referencing is days old.

The User

KARAN: TEAM LEAD

Karan oversees 20-50 agents and is accountable for his team's KPIs.

His day involves jumping between dashboards, reviewing escalations, and running 1:1s - all while trying to stay on top of who needs coaching and what for.

He's data-driven and looks for performance patterns, but his tools makes him do the legwork. Call discovery is manual. Coaching material lives in personal notes. And there's no system to track whether a coaching session actually moved the needle.

Focused on improving calls

Multi-tool workflow

KPI accountable

Data-driven reviewer

Oversees 20–50 agents

Coaches agents

Responsible for team performance

Looks for performance patterns

Limited time for reviews

Meanwhile…

Jyoti, a call center agent in Karan’s team, handles 80-120 calls a day in a high-stress, KPI-driven environment.

She needs fast, specific feedback - not a conversation that happens two weeks after the call she made the mistake on.

Jyoti, a call center agent in Karan’s team, handles 80-120 calls a day in a high-stress, KPI-driven environment.

She needs fast, specific feedback - not a conversation that happens two weeks after the call she made the mistake on.

Core Issue

Team leads don't have enough time to coach consistently - and the tools they have make it slower, not faster.

Pain Points

Manual call reviews

Slow preparation, inconsistent quality.

Coaching gaps for average performers

TLs prioritize high-risk cases; others fall through.

Delayed feedback

Feedback arrives too late for agents to connect it to the call it came from.

Difficult call discovery

Hard to find exact moments worth coaching.

Lack of structured coaching resources

Coaching depends on personal notes, not reusable resources.

Split Coaching Stack

Call insights and coaching live in two separate tools, doubling TL effort.

How Might We

How might we make coaching easier for team leads to deliver and scale?

Brainstorming / Solution Mapping

CURRENT SCENARIO

Card sorting + solution mapping

Market Research

Before heading towards fleshing out our solution we reviewed two existing market solutions which are very prominent in the Indian market- Convin and Observe.AI.

Convin

Observe.AI

Lingo (Proposed)

Coaching trigger

Automated

Convin

Automated

Observe.AI

TL-initiated + AI-assisted

Lingo

(Proposed)

Coaching trigger

Automated

TL-initiated + AI-assisted

Content creation

System-generated

AI-assisted, TL-reviewed

1:1 session support

Limited

Structured + AI-assisted preparation

Call moment curation

Auto-flagged

Auto-flagged, TL-editable

Both platforms offer some form of coaching, but skew toward automated, system-generated paths with minimal room for team lead input.

The coaching experience is largely hands-off - which sounds efficient, but removes the contextual judgment that makes feedback actually land.

Our approach takes a different position: AI assists the work, but people make the decisions.

Solution Feature Highlights

Two features, scoped to the team lead's workflow:

Coaching Packs

Structured, self-guided learning modules that TLs curate and deploy to agents, assisted by AI that suggests relevant call moments, common mistakes, and best-practice examples.

1:1 sessions (PLATFORM-Assisted Manual Coaching)

A scheduling and session support tool that helps TLs prepare for and run structured coaching sessions with high-risk agents, backed by AI-suggested and TL reviewed call data.

The main differentiator between the two features is that 1:1 sessions can tackle multiple issues in a single session (which would be crucial for high-risk agents), whereas coaching packs are built to target specific issues.

Solution Feature Breakdown

Information Architecture

Feature flows

Coaching Packs - Creation + Deployment

Flow

Details

Contains a ranked bar chart of the most frequent performance issues detected across the team, Each bar represents:

number of agents affected
frequency of the issue in calls

This helps the Team Lead quickly identify team-wide coaching opportunities.

The detailed view in a modal shows:

agents most affected by this issue.
number of agents impacted
total call occurrences of the issue
common patterns detected

AI automatically suggests:

mistake call snippets
good example call snippets
recommended coaching tips

The Team Lead can edit and review the AI suggestions.

The Team Lead can assign it to:

specific agents
agents affected by the issue
the entire team

1:1 Sessions - Scheduling + Preparation

Flow

Details

High-Risk Agents table generated from speech analytics signals.

Table Columns:

Agent, Risk Level, Key Issue, Calls Affected, Last Flagged, Action

This table helps the Team Lead quickly identify agents who require 1:1 coaching.

Opens a scheduling panel. The Team Lead can choose Session Type, fill detail fields

Online session- The platform generates a meeting link automatically using the Team Lead’s connected calendar.
In-person session
Fields include: session date, time, coaching objective, optional notes

Once scheduled, the agent and the Team Lead receives a calendar invite and notification.

AI automatically prepares a draft coaching kit for the Team Lead.

AI surfaces

flagged call snippets
transcript highlights
common mistake patterns
example good calls from the team

Team Leads can refine AI-curated snippets with their own examples, notes, and supporting materials.

Coaching Packs - Creation + Deployment

Flow

Details

Contains a ranked bar chart of the most frequent performance issues detected across the team, Each bar represents:

number of agents affected
frequency of the issue in calls

This helps the Team Lead quickly identify team-wide coaching opportunities.

The detailed view in a modal shows:

agents most affected by this issue.
number of agents impacted
total call occurrences of the issue
common patterns detected

AI automatically suggests:

mistake call snippets
good example call snippets
recommended coaching tips

The Team Lead can edit and review the AI suggestions.

The Team Lead can assign it to:

specific agents
agents affected by the issue
the entire team

1:1 Sessions - Scheduling + Preparation

Flow

Details

High-Risk Agents table generated from speech analytics signals.

Table Columns:

Agent, Risk Level, Key Issue, Calls Affected, Last Flagged, Action

This table helps the Team Lead quickly identify agents who require 1:1 coaching.

Opens a scheduling panel. The Team Lead can choose Session Type, fill detail fields

Online session- The platform generates a meeting link automatically using the Team Lead’s connected calendar.
In-person session
Fields include: session date, time, coaching objective, optional notes

Once scheduled, the agent and the Team Lead receives a calendar invite and notification.

AI automatically prepares a draft coaching kit for the Team Lead.

AI surfaces

flagged call snippets
transcript highlights
common mistake patterns
example good calls from the team

Team Leads can refine AI-curated snippets with their own examples, notes, and supporting materials.

Coaching Dashboard

The coaching dashboard gives team leads a single view of their team's health - who's at risk, where the skill gaps are, and how coaching activity is tracking.

So they can decide where to act without digging through multiple tools.

The coaching dashboard gives team leads a single view of their team's health - who's at risk, where the skill gaps are, and how coaching activity is tracking.

So they can decide where to act without digging through multiple tools.

High-Risk Agents Table

Flags agents who need immediate 1:1 attention, with their risk level, key issues, calls affected, and allows the TL to directly schedule a coaching session.

AI ASSIST Panel

Surfaces AI-generated team insights and suggested next actions, so the TL always has a starting point.

The panel is context-aware, once an action oriented task is initiated it actively supports the work in progress.

Coaching KPI Cards

Snapshot of four key numbers regarding agent coaching.

Coaching pack Tracker

Tracks the status of all assigned coaching packs across the team - who has started, who is overdue, and who has completed their assignment.

Team Skills Gap Chart

Ranked bar chart of recurring performance issues across the team, showing how many agents are affected and how often each issue appears in calls.

Coaching KPI Cards

Snapshot of four key numbers regarding agent coaching.

Team Skills Gap Chart

Ranked bar chart of recurring performance issues across the team, showing how many agents are affected and how often each issue appears in calls.

AI ASSIST Panel

Surfaces AI-generated team insights and suggested next actions, so the TL always has a starting point.

The panel is context-aware, once an action oriented task is initiated it actively supports the work in progress.

High-Risk Agents Table

Flags agents who need immediate 1:1 attention, with their risk level, key issues, calls affected, and allows the TL to directly schedule a coaching session.

Coaching pack Tracker

Tracks the status of all assigned coaching packs across the team - who has started, who is overdue, and who has completed their assignment.

Journey 1:

Karan opens the coaching dashboard to check where his team is slipping
- and ends up deploying a targeted coaching pack in minutes.

Journey 2:

Karan spots a high-risk agent on his dashboard and schedules a structured,
data-backed coaching session with minimal manual efforts.

Usability Testing

A structured usability testing, evaluating the two core team lead coaching flows across 10 participants, modelled on the Lingo coaching dashboard.

Participant briefs

Task-A

coaching Pack Deployment

Scenario + task brief

"You are a team lead managing a call centre team of 20 agents. Your coaching dashboard has flagged that several agents are consistently missing compliance statements on calls. Your task is to create and deploy a coaching pack to address this issue - starting from the dashboard."

Step-wise flow breakdown:

Select “Missed Compliances” Skill Gap from the graph.

Go through the skill gap details and start creating a coaching pack.

Go through pre-filled pack details and create a pack with the provided details.

Review/edit call snippets.

Assign pack to all the affected agents.

Review & Deploy the coaching pack.

Task-B

1:1 Session Scheduling

Scenario + task brief

"You are a team lead managing a call centre team. One of your agents, Arjun Verma, has been flagged as high-risk for escalation mishandling and dead air. Your task is to schedule a 1:1 coaching session with Arjun and review the AI-prepared coaching kit before confirming the session."

Step-wise flow breakdown:

Locate Arjun Verma in High-Risk Agents table.

Fill session scheduling details as provided.

Review coaching kit and confirm session.

Methodology

Participant pool

10 users, Ages 24–37, tech comfort rated 1–5.

outcome structure

Outcomes scored as Success or Fail.

TASK-A scope

6 steps (A1-A6) from dashboard hover interaction through to coaching pack deployment. Entry point: Skills Gap bar chart, not the Coaching Packs section nav.

Task-A

coaching Pack Deployment

TASK-B scope

3 steps (B1-B3) from High-Risk Agents table through to session confirmation and kit curation. Entry point: Coach CTA in the High-Risk Agents table.

Task-B

1:1 Session Scheduling

Overall results

100%

Task-A overall success rate

80%

Task-B overall success rate

Step-level findings

Step

Description

Success Rate

Difficulty

Select “Missed Compliances” Skill Gap from the graph.

100%

Go through the skill gap details and start creating a coaching pack.

100%

Go through pre-filled pack details and create a pack with the provided details.

100%

Review/edit call snippets.

100%

Assign pack to all the affected agents.

100%

Review & Deploy the coaching pack.

100%

Locate Arjun Verma in High-Risk Agents table.

100%

Fill session scheduling details as provided.

80%

Medium

2/10 users abandoned -

Both with low tech comfort

2/10 users abandoned -

Both with low tech comfort

Review coaching kit and confirm session.

80%

Key design implications : task-a

Priority 1: Critical

Priority 2: Moderate

Priority 3: Minor

Select “Missed Compliances” Skill Gap from the graph.

Observation:

4 of 10 users missed the bar chart on first hover attempt.

Tooltip and hover state are working. Consider reinforcing with a subtle animated pulse or underline on the top bar on first dashboard load to draw attention proactively.

Low Effort

Medium Impact

Review/edit call snippets.

Observation:

6 of 10 users accepted all AI-suggested snippets without reviewing.

Add a subtle 'Review & edit' label above the snippet list to surface editability. Does not need to be mandatory - just visible enough to prompt awareness.

Low Effort

Medium Impact

Go through pre-filled pack details and create a pack with the provided details.

Observation:

2 of 10 users skipped AI Assist panel entirely.

Surface one AI suggestion inline in the pack name or objective field. Rename panel to 'AI Context' or add a subtle badge on first visit.

Low Effort

Low Impact

Review & Deploy the coaching pack.

Observation:

3 of 10 paused at the Jyoti conflict warning before deploying.

Reframe warning tone: 'Jyoti has an active pack - you can still proceed.' Add a 'Manage assignments' link that opens a drawer, so users feel they have options without being forced to navigate away.

Low Effort

Low Impact

Key design implications : task-B

Priority 1: Critical

Priority 2: Moderate

Priority 3: Minor

Fill session scheduling details as provided.

Observation:

2 of 10 users could not complete the session scheduling form.

Date/time fields were unfamiliar to low tech comfort users. AI pre-filling of the coaching objective helped but didn't offset overall form complexity.

Medium Effort

High Impact

Review coaching kit and confirm session.

Observation:

5 of 8 users who reached B3 confirmed kit without deep review.

Consider a kit summary line above the confirm button: '3 call moments · 4 talking points · ready to go'. Creates a natural moment to reflect without adding friction.

Low Effort

Medium Impact

What this project taught me

AI-assisted flows don't behave like standard task flows

When AI is part of the workflow, the design problem shifts from "how do users find information" to "how do users trust, verify, and own a decision that was made for them".

Intent and affordance are separate problems

Clarity of intent on a screen doesn't guarantee clarity of action, they are two separate problems that have to be solved independently.

Next steps

Iterate on coaching kit review

Add a review layer to the kit confirmation step that introduces enough friction to prompt genuine engagement - whether through inline snippet context, a lightweight gate, or behavioral signals - without making the flow feel like an audit.

Design the agent-facing screens

The TL flows cover one side of the product. The agent receiving a coaching pack or 1:1 session invite is the other - and the design work there is still ahead. That's where this project goes next.

Thank you for reading!

Don't Click here

Key design implications : task-a

Priority 1: Critical

Priority 2: Moderate

Priority 3: Minor

Select “Missed Compliances” Skill Gap from the graph.

Observation:

4 of 10 users missed the bar chart on first hover attempt.

Tooltip and hover state are working. Consider reinforcing with a subtle animated pulse or underline on the top bar on first dashboard load to draw attention proactively.

Low Effort

Medium Impact

Review/edit call snippets.

Observation:

6 of 10 users accepted all AI-suggested snippets without reviewing.

Add a subtle 'Review & edit' label above the snippet list to surface editability. Does not need to be mandatory - just visible enough to prompt awareness.

Low Effort

Medium Impact

Go through pre-filled pack details and create a pack with the provided details.

Observation:

2 of 10 users skipped AI Assist panel entirely.

Surface one AI suggestion inline in the pack name or objective field. Rename panel to 'AI Context' or add a subtle badge on first visit.

Low Effort

Low Impact

Review & Deploy the coaching pack.

Observation:

3 of 10 paused at the Jyoti conflict warning before deploying.

Reframe warning tone: 'Jyoti has an active pack - you can still proceed.' Add a 'Manage assignments' link that opens a drawer, so users feel they have options without being forced to navigate away.

Low Effort

Low Impact

Key design implications : task-B

Priority 1: Critical

Priority 2: Moderate

Priority 3: Minor

Fill session scheduling details as provided.

Observation:

2 of 10 users could not complete the session scheduling form.

Date/time fields were unfamiliar to low tech comfort users. AI pre-filling of the coaching objective helped but didn't offset overall form complexity.

Medium Effort

High Impact

Review coaching kit and confirm session.

Observation:

5 of 8 users who reached B3 confirmed kit without deep review.

Consider a kit summary line above the confirm button: '3 call moments · 4 talking points · ready to go'. Creates a natural moment to reflect without adding friction.

Low Effort

Medium Impact

The User

KARAN: TEAM LEAD

Karan oversees 20-50 agents and is accountable for his team's KPIs.

His day involves jumping between dashboards, reviewing escalations, and running 1:1s - all while trying to stay on top of who needs coaching and what for.

Focused on improving calls

Multi-tool workflow

KPI accountable

Data-driven reviewer

Oversees 20–50 agents

Coaches agents

Responsible for team performance

Looks for performance patterns

Limited time for reviews

Meanwhile…

Jyoti, a call center agent in Karan’s team, handles 80-120 calls a day in a high-stress, KPI-driven environment.

She needs fast, specific feedback - not a conversation that happens two weeks after the call she made the mistake on.