Primary Goal
Integrate Generative AI into the Cybersecurity Platform to enhance user experience through interactive documentation, basic analyst augmentation and guidance, functioning as a concierge service and real-time analyst support.
Process for Requirement Breakdown and Goal Definition
Connect with PLM (Product Line Management): Collaborate to understand the product goals and vision. Engage with Subject Matter Experts (SMEs): Gather in-depth insights and technical requirements. Dig Deep into Requirements: Analyze user needs and pain points for a comprehensive understanding. Breakdown of Goals: Create smooth, logical paths for users to engage with the platform and AI features.
Key Learnings from Collaborating with PLM, SMEs, and Analyzing Requirements
Easy to consume
Given the broad scope of our cybersecurity platform, AI capabilities should be tailored for different use cases (e.g., threat detection, incident response, policies, etc.) to ensure relevance and ease of access across various modules, while also creating clear connections between them.
Access to Recurrent Tasks
Users should be able to easily pick up where they left off, with the ability to return to relevant areas or tasks across the platform without friction.
Clear Feedback and Trust
Ensure the AI provides clear reasoning behind its suggestions and allows users to give feedback. This builds user confidence in the AI’s recommendations, while also enabling the AI to learn and better understand the user’s needs.
Building on the initial understanding
The UX of creating a general hub incorporates familiar research patterns, drawing inspiration from AI-driven platforms in the market. This approach helped align with users’ existing preferences and expectations. Users needs an intuitive hub, for better content organization, and seamless integration in platform for collaboration across analysts and managers. A key thought was for enabling seamless sharing of information across different teams—such as analysts and managers. Did an brainstorming and card sorting exercise for features, later extending it to the information architecture (IA) below.
Next Steps
Proactively shared IA, wireframes, and initial design drafts with multiple stakeholders to gather feedback and facilitate iterative design changes through several feedback sessions.
When the UX was at a good stage internally, presented to some top Fortune 500 companies customers to gain their insights.
Afterward, trade-offs were made to prioritize the MVP and help users achieve their goals
1. No Need for Last Conversation: Since we already displayed recent interactions, it was redundant to show the last conversation again.
2. Voice Search Feature: Given our diverse user base with different language groups, we decided to add a voice search feature. This wasn't initially considered, but user insights suggested that voice search and audio-based prompts would be more intuitive and effective.
3. Sharing Notes and Favorites: Exploring options to allow users to share notes and favorites became a valuable insight, as it improved collaboration and user engagement.
4. Collaboration with Engineering on Timelines: After collaborating with the engineering team to understand the timeline for the AI feature, it was determined that implementing new graphs and charts in the first version would require significant time and resources. Therefore, we decided to stick with the current graphs, which the PLM agreed upon.
On the similar way Page-aware were also thought of in different areas of product which could support prompts of results
1. Text 2. Tables 3. Nodals 4. Images 5. Graphs
Now for AI in Bespoke UX
Investigate Events Analysis
The bespoke experience would be an AI tailored for specific tasks or areas, designed to provide focused, context-driven responses.
Core Problem
The primary goal in this case is to AI should enable users to write queries using natura language and receive concise, context-driven summaries of the results. This allows analysts to quickly see key insights from large datasets, and at the end, view a comprehensive summary of the results, improving efficiency in investigations without the need to manually parse through all the information.
Breakdown of Problem
Users should be able to query in natural language, making interactions intuitive. Users should receive a summary of results, highlighting key insights for quick understanding.
Thank you for reading.
I hope you liked it.
