SlideAI

Designing a GenAI-powered workflow to accelerate powerpoint slide creation for consulting teams

CLIENT

McKinsey’s Technology & Digital (T&D) group, responsible for developing, evaluating and launching tech solutions for use by the rest of the organization.

IMPACT

In 6 weeks:

  • Identified key user behaviors, use cases and needs

  • Defined clear product value proposition and feature backlog for future development

  • Delivered UX for Beta release

Consulting teams that were part of Beta testing reported:

  • Spending ~20% less time on upfront desk research

  • 85% being able to fulfill visual layout and design needs with minimal to no involvement from Presentation Design Team

ROLE/TEAM

UX Lead (myself) + Project Manager, Tech Lead, ML/AI engineering, full stack engineering

RESPONSIBILITIES

As UX Lead, the scope of my work covered:

  • Workstream scoping, planning and management

  • Leading qualitative user discovery and research

  • Defining product strategy and differentiation between existing and net-new tools

  • Generating feature backlog and collaborating with Project Manager on Beta feature prioritization

  • Working across multiple product teams to define a cohesive user experience

Context

From initial ideas to final deliverables, Powerpoint slides are the primary medium of communication consultants use throughout a client engagement. Hundreds of hours are spent on slide creation and iteration - which makes accelerating this process an ongoing focus for McKinsey’s Technology & Digital (T&D) Group.

Problem/Brief

T&D leadership wanted to:

  • Understand the opportunity to expedite the slide creation process through the use of GenAI

  • Define any net-new features and tools that would be needed to be built to support a Gen-AI enabled workflow

Finding targeted opportunities for GenAI to have a meaningful impact

As the potential field of discovery was undefined, I took a clean-sheet approach to user discovery - bearing in mind that it would be key to find specific moments where GenAI could tangibly improve the current experience, rather than try to apply it in a one-size-fits-all manner across the entire current state journey.

To quickly identify these key moments, I ran qualitative and quantitative research in parallel - launching a survey of consultants at all levels of seniority and targeting key individuals for qualitative interviews. This enabled the team and I to better understand the phases and activities a team goes through in Powerpoint slide generation.

As part of both research methods, I included co-creation activities to prioritize their tasks along the axes of ‘time taken to complete’ vs ‘amount of original/creative input needed’. This helped identify the colloquial “grunt work” - tasks most ripe for automation. Together with the PM and T&D leadership, we downselected to 2 key use cases for initial exploration:

  • Synthesizing raw desk research into slide content that is easily digestible (e.g. summarizing; turning text into bulleted lists/columns)

  • Basic visual alignment and layout refinement of basic slides into Firm-approved output (e.g. applying visual templates; aligning text; inserting illustrative iconography). While teams typically “outsource” this to a small central pool of Presentation Designers, a sharp increase in requests in recent years has turned what used to be an overnight turnaround into a multi-day wait.

I mapped out a North Star user journey around these two use cases, and built a list of features required to enable the future state experience. I then worked with the PM, Tech Lead and ML/AI engineers to understand time and tech constraints and narrow the scope for Beta testing.

Getting creative with Beta delivery

Good news: Leadership was on board. Less good: They want it tomorrow.

Once Leadership was aligned on initial research and overall feature scope, they set a target of starting Beta testing within the quarter. This obviously posed problems with regards to timeline for design and build - and I was challenged to find a way to execute on as many Beta features as possible in the limited timeframe while maintaining a good user experience.

Working closely with ENG and ML/AI leads, I iterated on multiple possible user flows, eventually proposing a solution to split the features and reduce the size of any net new build. This involved leveraging Lilli, McKinsey’s existing GenAI tool, in tandem with a Powerpoint plugin that would be built from scratch.

While the PM worked with Lilli product leadership to prioritize new features for release, I held twice-weekly working sessions with Lilli designers and F/E engineers to (a) understand their UX/UI and ML/AI constraints and (b) refine my design output. I did this in parallel with UX development of the Powerpoint plugin, ensuring a cohesive user experience across both apps.

As I was scheduled to transition to a new project, I created a package of detailed UX and UR artifacts to ensure a seamless onboarding for future designers. This included early wireframes and information architecture explorations, calling out areas and interactions that required more investigation and development - this was particularly effective as it:

  • Allowed the engineering teams to kick off exploration and development in spite of missing a dedicated UX/UI resource

  • Expedited the amount of UI design work needed to get to Beta, with many of the initial wireframes and screen designs being implemented in the release.