AI-Powered Knowledge Sharing

Pulse Knowledge Assistant

COMPANY

Pulse

ROLE

Product Designer

Industry

Enterprise AI Tool

YEAR

2025

Background

In 2025, AI tools for work hit full speed. Around the same time, I took a Coding with AI course that made me rethink how teams manage knowledge. Updates of important context sometimes passed on without clarity, and information was scattered across tools.

That sparked the idea for Pulse. I wanted a tool that helps teams find what they need, when they need it and make knowledge easier to access at the right moment.


Challenge and Opportunity

Challenge:

Fragmented workflows, with scattered context across tools, undocumented decisions at times, and poor traceability, leading to misalignment, delays, and repeated rework.


Opportunity:

Explore how AI can reduce workflow by retrieving relevant information instantly, enabling teams to work more efficiently and independently across time zones.


How might we help product teams access the right context at the right time without digging through Figma, JIRA, or Confluence?


Outcome


  • Reduced the need to switch between tools to locate information.


  • Helped teams get the right information right away and not on tracking down scattered context.


  • While still in concept phase, Pulse reflects my ability to design AI-driven solutions that address workflow inefficiencies and improve operational clarity.

Problems I Identified

Everyone has their own way of organising information, some bookmark tabs, others compile links into Confluence pages as their source of truth.

Problem 1

Scattered Knowledge Across Tools
Important context is split across Figma, JIRA, and Confluence, making it time-consuming to locate what’s needed.

Problem 2

Slow Documentation Access and Limited Visibility
Documentation exists, but it’s challenging to locate or unclear if it should be updated. Timezone differences sometimes make it harder for teammates to access the right information independently, leading to unnecessary back-and-forth effort.

Problem 3

Misalignment Between Roles
Designers, engineers and PMs are often looped in ad-hoc. Due to shifting responsibilities, it’s unclear who to align with at times, leading to fragmented communication and fragmented user journeys.





Current Tools and Workflow


Most rely on JIRA, Confluence, and Excel for managing workflows, while designs are primarily used in Figma and Adobe. To find what they need, they often depend on bookmarks or digging through message threads. Alternatively, they reach directly to their co-workers to ask for the things they needed.


How Competitors Address These Challenges

While exploring how teams manage knowledge today, I looked at tools like Glean, Guru, ClickUp Brain, and Notion. Each offered different strengths in making knowledge more accessible and collaborative:


  1. AI-Knowledge Management
    Help users access, organise, and retrieve information efficiently. They support features like search, summarisation, and recommendations to streamline workflows.


  2. Integration with Other Tools
    Each platform integrates with a range of external apps (e.g., Google Drive, Slack, Confluence), aiming to centralise knowledge and reduce context-switching.

  3. Collaboration Features
    These tools support team collaboration, allowing multiple users to contribute, edit, and verify content, which is essential for large organisations.



Glean stood out for its ability to connect knowledge across tools and deliver relevant insights through a work assistant.

It inspired me to rethink how similar intelligence could support product design teams, and Pulse builds on that idea, helping teams retrieve relevant information the moment they need it.

Prototype Using v0

I decided to incorporate AI tools into this project and generated a rough prototype using v0 to explore layout and component ideas. It helped accelerate early thinking, but I still had to redesign the experience to better align with user needs and apply GenAI UX principles more intentionally.

User Flow

Starting from a personalised homepage, the user can explore or create prompts, run them through AI, and receive a clear summary with sources and confidence level.

They can then give feedback, share results, or save them for future use, making it easier to stay align.


Solution

From exploring suggested prompts to generating context-rich summaries with linked sources, each screen is designed to reduce friction, improve clarity, and support collaboration across product, design, and engineering teams.

The interface focuses on trust, usability, and efficiency, bringing scattered information into one actionable workflow.


Home: Showcases clarity as it shows personalised, actionable prompts upfront, includes metadata to help decide what's useful and tooltips added to reinforce discoverability.


Users can browse and filter prompts by topic, department, or creator, making it easy to discover relevant workflows and reuse existing knowledge across teams.


To create a new prompt, enter prompt details accordingly.


An example of a user who selected Marissa's prompt to find a component update in the design library. Prompt with specific details ensure the AI delivers more accurate, context-aware results tailored to their task.


Pulse actively retrieving context from connected tools like JIRA, Figma, and Confluence, turning a simple prompt into a powerful way to uncover relevant, cross-functional knowledge without manual digging.


Results screen demonstrates transparency by showing the AI’s confidence level, clearly attributing the DRI, and linking directly to original sources helping users understand, verify, and trust the information provided.


I thought it would be refreshing to display who have recently collaborated a prompt with you, or perhaps used your prompt recently under 'Directory'. This supports visibility, collaboration and reconnection for alignment.


Knowledge: Organises prompt results, documents, and shared insights into curated collections, making it easy for teams to find, reference, and reuse key information with clarity and consistency. I have referenced this to Glean's Collections.


Before and After Journey

By comparing common pain points before Pulse like scattered documentation, tool-switching, and unclear collaboration with streamlined outcomes after implementation, it highlights how AI-driven prompts help teams find information faster, reduce rework, and work more transparently across roles and tools.



Defining Success Metrics

Although this is a self-initiated project that highlights just one feature of an internal tool with an AI assistant, my goal is to outline realistic success metrics that demonstrate its potential business impact.



Pulse represents the idea of staying in sync with the knowledge, decisions and people that drive progress.

Risk Management

What I’ll Look Out For

AI tools are unpredictable by nature. So as a designer, I can’t just focus on static screens, I need to observe how the AI behaves in context, and how users respond to that behaviour. This helps build a product that’s not just smart, but usable and trustworthy.

Even though this is a learning project, it’s taught me to go beyond UI, and start thinking about AI behaviour, user trust, and real-world interactions.



Reflections

This self-initiated project revealed just how complex it is to create effective AI-powered experiences. While the concept felt valuable, I recognise that bringing it to life would demand a deeper assessment of technical feasibility.

One key area is usability testing. AI-powered tools presents unique challenges beyond traditional software. Users' mental models for interacting with AI assistants are still evolving rapidly. It could be a scenario whereby early feedback indicates users might overestimate what the assistant can do, then underuse it once they hit limitations.

Even when AI truly adds value, I still have many questions around designing for AI and alongside AI. This experience made me realise how much more there is to consider.

I'm keen to deepen my skills in areas like prompt engineering as a UX tool, ethical foundations of AI, and testing AI concepts through research. There’s still a lot to learn, and that’s what keeps me curious.