Back

Back

Generative AI - Design sprint

Generative AI - Design sprint

Generative AI - Design sprint

Integrating KOSMO with agent search was designed to reduce average call handling times for call centre executives, improving efficiency and lowering operational costs. With stringent timelines, we decided to run a 2 day Design sprint workshop

Role

Role

Product Designer

Approach

Approach

Design sprint

Design sprint

Tools

Tools

Figma, Miro

THE WHAT

In a condensed 2-day design sprint, we gathered multiple stakeholders for an in-person workshop to tackle our project goals under tight deadlines. On Day 1, we kicked off with a meeting to clarify the core problem and key questions, followed by an ideation session where participants sketched and discussed their solutions, ultimately narrowing them down to a few promising concepts.

On Day 2, we held critique sessions to evaluate and refine those concepts, encouraging constructive feedback and collaboration among stakeholders. Although we couldn't create a prototype or conduct user testing due to time constraints, we concluded with a debriefing session to review the ideas generated and outline actionable next steps for the final product development.

THE PLAN

A key stakeholder was identified who had the final say after every session. He acted as the decision maker and his vote ensured we were setting the expectations right without getting carried away. In this workshop, the decider was “Head of transformation”.

Post every session, we had voting session, where every participant was assigned certain number of votes. This ensured every participant had equal opportunity to review the ideas/suggestions and the best one was picked.

The below steps outline a rough outline of how the workshop was conducted followed by the results.

Agenda
Goal setup
How are we measuring success?
What might stop us from reaching our goal?
What are our competitor’s doing?
How Might We?
Our customers and their journey
It’s time to for idea generation

Workshop Output

Key aspects of the workshop following each step and activity are outlined below, with each activity highlighting the 'Decider votes' section, which reflects the final decision maker's chosen takeaway.

The activity ensured that every member of the workshop was heard, with their suggestions and concerns taken into consideration.

Goals, Metrics and Risks

Goals

Quick, Fast and Accurate. something they would use happily.

Metrics

30% reduction in average handle time on service calls

100% advisor usage and reduced Average Handle Time (AHT)

95% of agents say they prefer using KOSMO while helping customers

Risks

Can we roll out without upsetting the workforce?

What is the data is rubbish?

Competitor analysis

Bard - Direct competitor

Bard is an AI tool designed to help customer service teams quickly resolve inquiries by providing intelligent, real-time suggestions based on data, improving efficiency and customer satisfaction. It continuously learns from user interactions to enhance its recommendations. Bard streamlines communication and reduces response times, making it an invaluable asset for customer su

Advantages

Lengthy answers with too much information.

No validation question to refine response

Disadvantages

Lengthy answers with too much information.

No validation question to refine response

Chat GPT - Direct competitor

ChatGPT is an AI language model that generates human-like responses in conversations, useful for tasks like answering questions, content creation, and providing explanations across a wide range of topics. Trained on diverse datasets, it adapts to different conversational contexts and improves its responses over time, offering versatile support for various industries and use cases.

Advantages

Simple and uncomplicated user interface.

Easy to understand what the tool is and it’s capabilities.

Disadvantages

User interface is non-intuitive and boring.

Option to get feedback and the questions need to be structured and planned.

How Might We statements (HMW)

During the workshop, the we grouped all the HMW statements into categories and formed specific themes. The Key stakeholder the had a chance to pick 5 Important HMW statements which he felt were more important for the MVP launch. The key themes and a few HMW statements are listed below.

Themes

Accuracy of the AI tool

User Interface

Content

Feedback

HMW statements

HMW ensure content given as a response is 100% accurate?

HMW integrate wider one help tools?

HMW ensure easy feedback?

HMW promote and engage usage of the tool?

Journey Map

The journey mapping activity helped us visualise the key players involved during the end-end process and also understand the key expectations from every player involved.

Journey Map

The journey mapping activity helped us visualise the key players involved during the end-end process and also understand the key expectations from every player involved.

Key Players

Sky customers

Call centre executives

Data scientists

Content editors

Workshop images

A brief look into the output of the workshop. We had a diverse list of participants which also included 4 call centre experts who were the end users of the tool

Final output

How did the workshop help me in finalising the designs?

WORKSHOP OUTPUT

Key aspects of the workshop following each step and activity are outlined below, with each activity highlighting the 'Decider votes' section, which reflects the final decision maker's chosen takeaway.

The activity ensured that every member of the workshop was heard, with their suggestions and concerns taken into consideration.

Goals

Quick, Fast and Accurate. something they would use happily.

Metrics

30% reduction in average handle time on service calls

100% advisor usage and reduction in Average Handle Time

95% of agents say they prefer using KOSMO while helping customers

Risks

Can we roll out without upsetting the workforce?

What is the data is rubbish?

Bard - Direct competitor

Bard is an AI tool designed to help customer service teams quickly resolve inquiries by providing intelligent, real-time suggestions based on data, improving efficiency and customer satisfaction. It continuously learns from user interactions to enhance its recommendations. Bard streamlines communication and reduces response times, making it an invaluable asset for customer su

Advantages

Lengthy answers with too much information.

No validation question to refine response

Disadvantages

Lengthy answers with too much information.

No validation question to refine response

Chat GPT - Direct competitor

ChatGPT is an AI language model that generates human-like responses in conversations, useful for tasks like answering questions, content creation, and providing explanations across a wide range of topics. Trained on diverse datasets, it adapts to different conversational contexts and improves its responses over time, offering versatile support for various industries and use cases.

Advantages

Simple and uncomplicated user interface.

Easy to understand what the tool is and it’s capabilities.

Disadvantages

User interface is non-intuitive and boring.

Option to get feedback and the questions need to be structured and planned.

Competitor analysis

During the workshop, the we grouped all the HMW statements into categories and formed specific themes. The Key stakeholder the had a chance to pick 5 Important HMW statements which he felt were more important for the MVP launch. The key themes and a few HMW statements are listed below.

THE PLAN

A key stakeholder was identified who had the final say after every session. He acted as the decision maker and his vote ensured we were setting the expectations right without getting carried away. In this workshop, the decider was “Head of transformation”.

Post every session, we had voting session, where every participant was assigned certain number of votes. This ensured every participant had equal opportunity to review the ideas/suggestions and the best one was picked.

The below steps outline a rough outline of how the workshop was conducted followed by the results.

Agenda

Goal setup

How are we measuring success?

What might stop us from reaching our goals?

What are our competitors doing?

How Might We?

Our customers and their journey

It’s time to ideate

Themes

Accuracy of the AI tool

User Interface

Content

Feedback

HMW statements

HMW ensure content given as a response is 100% accurate?

HMW integrate wider one help tools?

HMW ensure easy feedback?

HMW promote and engage usage of the tool?

Key Players

Sky customers

Call centre executives

Data scientists

Content editors

How Might We statements (HMW)

During the workshop, the we grouped all the HMW statements into categories and formed specific themes. The Key stakeholder the had a chance to pick 5 Important HMW statements which he felt were more important for the MVP launch. The key themes and a few HMW statements are listed below.