Design Challenge

Slack Huddles: AI Summary & Recording

 
 

Often, conversations happen in Slack, and then a subset of participants move to a huddle to continue the conversation. However, some team members who weren’t in the huddle see the thread where it left off with no conclusion and they are stuck with incomplete information or potentially incorrect information. I challenged myself to solve this problem to ensure everyone is in the loop on how a conversation ended.

AI TECHNOLOGY | UX/UI | PROTOTYPE POC

+ Solo design challenge
+ Timeline: 1 week

 
 
 
 

Context

Problem

Huddles are often used to continue a conversation to more easily discuss and come to decisions together. However, there is no way to understand the decisions being made within a huddle if you weren’t in the huddle.

This leads to loss of context, misalignment, and repeated conversations.

PROCESS

I followed the double diamond design process, adapting it to my timeline and project needs. My assumption was that the huddles functionality that currently exists had been well received and the project goal was to add additional functionality rather than redesign huddles.

My goal for this design challenge was to ideate and validate a potential solution to propose a believable POC without getting too far in the weeds with UI design.

 
 
 

Discovery

I explored Slack functionality more deeply by auditing the huddles feature and potentially adjacent features to better understand what I was starting with.

  • Understand huddles functionality

    • Actions

    • States

    • Notifications

  • Understand potentially adjacent features

    • Canvas - In-app channel documentation

    • Files - all shared files

    • Org tools - connected apps & orgs

    • Video & Audio messages

 
 

SLACK PRODUCT RESEARCH

If I were part of Slack’s product design team, I would have context about the company goals, product tenants, and vision for the products and features that I work on. In this challenge, I did some research to better understand Slack’s product direction and values in order to make assumptions about how the project could align and support.

Slack Vision Statement

Make work life simpler, more pleasant, and more productive.
Slack is the productivity platform that empowers everyone with no-code automation and AI, 
makes search and knowledge sharing seamless, and keeps teams connected and engaged. 
Around the world, Slack is the platform companies trust and people love to use.

Research-based Assumptions

Based on my knowledge, I can assume that Slack identified that :

  • Clear decision documentation & knowledge transfer is a key aspect of productivity in teams

  • Connecting teams in the tools they already use reduces the need to pay for additional documentation & meeting tools

  • Using no-code automation and AI to reduce manual processes is an efficiency gain (potential early differentiator)

 

INITIAL OPEN QUESTIONS

As this was a solo design challenge, there were a handful of open questions I would’ve otherwise sought to answer through stakeholders and existing resources.

Q1. Is there any previous research or product usage data around huddle usage that may be relevant?

  • Look for any existing research, data, or internal knowledge about the huddles feature, how it has been received, and any related feedback or usability tickets

  • Review anything that was descoped in previous iteration that might be worth considering now

  • Identify/establish baseline metrics related to feature through usage data tool

    • Total number of huddles made or joined within an org, lasting at least 5 seconds with at least one other member (existing Org-level analytics)

    • % of huddled within an org that had a topic added

    • % of huddles within an org that had notes taken within the huddle thread

    • % of active team members in channel that viewed notes in the huddle thread that had been taken

Q2. Are there any other details about the focus user group?

  • I made the assumption that we’d focus in on teams using Slack that work is a fully remote or hybrid environment, where huddles are a key tool for quick discussions and decisions

Q3. Are there any known constraints that could impact design decisions? (ie, technical, timelines, etc)

  • I made the assumption that there were no known constraints yet

  • Design-led project to identify opportunity based on research

  • Align with product and engineering stakeholders on research findings

  • Identify scope & constraints along the way

 

CURRENT WORKFLOW

  • Understand how huddles are started and what actions can be taken

  • Define workflow for huddle in team channel vs DM

  • Identify opportunities & gaps

 
 

Ideation

I went through the process of rapid ideation, similar to how I might facilitate a stakeholder workshop to generate ideas and build alignment.

  • Post-it brainstorm

  • Crazy 8s sketches

  • Sketching out top ideas

Brainstorming and grouping ideas into themes

Crazy 8s rapid ideation; 1 min per box

Example of sketches of the top ideas in more detail

 
 

Solutioning

DEFINING USER STORIES

User stories helped me realistically think about the user’s needs and flows. I used these to narrow the scope of my project and focus in on the first story within my short timeline.

  • ⭐️ As a slack user in a team channel, we move a conversation to a huddle in the same channel to work out a problem and come to a decision. Only a portion of the team is involved. We need to document the outcome for other’s to be aware of.

  • As a slack user in a large channel, I ask a question to the team. The SME responds and suggests a huddle. We start a direct 1-1 huddle. Information and decision that were made may be relevant to the team channel, so I share the notes and outcomes.

  • As a slack user in a team channel, I was not available/online for a team huddle where important decisions were made and want to see the outcomes of a decision that may impact me.

WORKFLOW OPPORTUNITY

  • Opportunities drive design exploration

  • Document ideas and open questions

  • Refine scope

DESIGN SYSTEM

Utilized this Figma community Slack component library in place of Slack’s real design system.

  • Library isn’t maintained and hadn’t been updated to Slack’s current UI & functionality

  • For the sake of the exercise, I used it as is in order to focus on the design problem

  • I built any additional local components used to align with the system using defined token and patterns

 

Design Options

OPTION 1: Prioritize huddles & note-taking

Low LOE enhancements to promote the use of existing features within team channels. These would be easy enhancements to add into any direction chosen.

 

OPTION 2: AI-generated huddle summary

In line with Slack’s vision statement, utilize AI to automate transparency and dissemination of information.

 

OPTION 3: Record huddles

Meeting apps have introduced recording for transparency and asynchronous work. Adding this into Slack would keep users from needing to switch apps to keep records of discussions during meetings.

 
 

Validation

Quickly validate the problem, workflows, and design options with user feedback.

  • 3 participants from remote or hybrid company

methodology

  • 30 min remote calls

  • Workflow interview questions

  • Review designs & validate direction

 

AI Summary Insights

  • Initial impression was positive

  • Participants expected the generated summary to come from the audio & thread, not just the thread.

  • Overall, this option was favored for easily catching up on what happened in a huddle

“If we’re not diligent about adding notes from the meeting into the thread, it would have less value.”
- Amanda, Eng tech lead

“It feels bad when the whole team makes a decision but the one person that wasn’t there is out of the loop.”
- Colby, Product Manager

 

Recording Insights

  • Recording in huddle is clear; maybe more testing/iteration on message indicator in channel for folks not in huddle

  • Participants like seeing timestamps for recording(s) in thread and playback

“If there was like a one hour huddle when I was OOO, I don’t have time to go back and watch the whole thing.”
- Colby, Product Manager

“What if you want to go back and look at something, but theres a lot of messages...is it saved somewhere else as well? like Google Drive?”
- Suzanne, Program Manager

 

Key Findings

  • The AI summary & Recording ideas solved different issues around catching up on huddles

    • AI Summary favored for efficiency – understand at high-level what was decided

    • Recording favored for transparency –document and look back on exactly what happened

  • Participants wanted a blended option including recording, transcripts, and AI summary based on audio, notes and things shared.

Research findings would be documented in PRD, shared with stakeholders to align on opportunities, as well as disseminated to relevant functions in the org.

Final POC

Prototype of final solution

I incorporated user feedback to propose a direction for huddles enhancements, ideally combining the options into a single solution to:

  • encourage users to use huddles functionality

  • solve for transparency of what was talked about in the huddle

  • allow users to efficiently understand what happened

Product strategy

Based on the discovery work and research, with the proposed solution we would expect to see:

  • Slack as a leader in the market in utilizing AI technology to make async work more efficient

  • Teams that work in Slack can host meeting in team channels, reducing the need for companies to pay for other meeting tools

 

Subtle usability enhancements encourage users to start huddles in team channels for visibility.

In addition to the 2 pop-out states for huddles, in-footer huddles default the huddle thread open and encourage users to jot down notes and links.

Introducing recording functionality to huddles alleviates the need for teams to leave their communication tool in order to use a meeting application which can record and document important conversations.

Clear timestamps and record of recordings make it easy to understand what happened, look back on what was decided, and search for past huddles.

  • Slack has already built great tools around media viewing, media thumbnails, and transcriptions. The huddles feature could leverage that and re-use some existing patterns used in the app.

Leveraging AI technology in Slack by allowing users to generate a summary of their huddle and eliminating manual tasks. This would improve knowledge sharing for busy teams and drive efficiency in communication.

Analytics Dashboard

Add additional data to the analytics dashboard with new functionality to let orgs see their huddle productivity metrics.

For example:

  • Total number of messages added to a huddle thread

  • Total number of AI summaries generated after a huddle

  • Total number of audio or video recordings from huddles

 

Next Steps

  • Stakeholder alignment

    • Timeline/scope, product direction, & tech constraints

    • Update workflow & designs with any changes

    • Define success metrics for aligned project

  • Refine designs & add consider the details

    • Tool tips, loading states, error handling

    • Searching historical recordings

    • Editing/deleting summary or recording

    • Recording settings

    • Recording UI & interactions

Additionally, there were items that I wasn’t able to explore due to timeline constraints. These are items that could be explored as part of this project or as future iterations post-release:

 

Project Reflections

Design systems for design efficiency

By using and adhering to an existing design system, I could focus on solving user pain points and product feature gaps. Finding an existing component library helped me rapidly design and test mid- & hi-fidelity design options.

Process is important, but it shouldn’t limit you

My solid design process served as a foundation to help me through this 1 week design exercise but didn’t hold me back from coming to a reasonable design solution within the given time.

Leverage data in decisions

In lieu of stakeholder feedback, I leaned into user feedback and product research to guide my design decisions in this design-led project. By gathering data, disseminating, and aligning with stakeholders on the outcomes, I can advocate for the user and push back where it matters against constraints and scope.