0-to-1
AI SaaS
Rapid Prototyping
Enterprise Cybersecurity faces workforce shortage, high stress and threats of ever-evolving attacks. On top of that, attackers don't quite follow a 9-to-5 schedule, keeping security operators constantly on edge.
Graphen was an AI solution start-up founded by ex-IBM researchers. It aimed to bridge the gap in workforce shortage and solve advanced threat detection with graph-based AI. It partnered up with a national bank to incubate and validate the MVP.
How can we enhance the efficiency of enterprise cybersecurity operations — improving monitoring, triage, and investigation processes — to enable operators to manage more with less, swiftly detect and address threats, and provide trust-worthy AI predictions in a high-stress, high-stakes environment?
As the start-up's founding designer, I designed and shipped the company's first MVP from inception, including design discovery, user research, product vision alignment, user validation and execusion to secure the company's first paying customer ($2M contract).
Team & Context
Impact
70%
Increased Incident Analysis Volume
40%
Improved Threat Detection Accuracy
$2M
Contract Secured - First Paying Customer
Graphen’s AI-assisted investigation suite, unlike typical cybersecurity solutions targeting expert users, utilizes ergonomic principles for efficient security monitoring and triage, tailored to users’ knowledge levels and organization's workflows.
Discovery & Prioritization
Challenge
Identify User Needs and Pain Points
Approach
I mapped out the design journey and identified two potential personas and their pain points and goals. I shared my findings with the product team and prioritized tasks with a mix of short-term fixes and long-term bets for prioritization.
Result
I quickly gained allies in product, engineer and research team to shift the waterfall process to be customer-centric, introduced clarity on product vision and roadmap.
This paved the way for future swift product pivot and rapid prototyping to meet shifting customer demands.
In design discovery, I identified core personas and charted out the investigation workflow, and workshoped with my team to emphasizing user pain points and ideate opportunities.
Key Insights
Junior operators’ pain points were repetitive routine inspections and interpreting alert signals with a handbook; they focused on meeting quotas and comply with policy. "Confidence score" means little to them.
Senior operators are experienced and prioritized trust and transparency over convenience. They scrutinize the reasoning and sources behind AI predictions because how much stake is on the line.
To drive product-led adoption, I needed to shift the team’s focus from developing for themselves (AI researchers) to bridging the gap between customers and their goals. Boring, non-AI solutions can bring a lot of value.
Based on insights, I proposed several opportunities and shared the initial wireframes with the team. We prioritized these based on complexity and importance.
My design leadership transformed the waterfall, feature-factory approach and established a unified product vision. This strategic focus accelerated our path to product adoption, brought clarity and increased team confidence in our product's success.
Define
Streamline Threat Monitoring to Drive Adoption
Challenge
The team had built a prototype that produced AI predictions but that tool received poor adoption and engagement.
Approach
I had the insights that junior operators were struggling with technical jargon and non-intuitive workflows, so I simplified AI outputs and created a streamlined information hierarchy that fit into the existing workflow.
Result
The reimagined investigation dashboard received quick adoption with its reduced complexity and seamless workflow integration. This saved my team development time on dedicated onboarding features and reduced the need for extensive corporate training.
Based on these findings, I simplified the AI model outputs to suit the operational level of junior operators and developed a structured information system to integrate smoothly with existing workflows, significantly easing the onboarding process and enhancing adoption.
To streamline the monitoring workflow and reduce complexity, I abstracted away the complexity in AI outputs to better align with their technical proficiency, and structured the solution's
Prototype & Validate
Build Trust With the AI Blackbox
Challenge
AI’s blackbox nature hindered user trust in a high-stake environment with its unreliable predictions and lacked of explanations. Close to launch, I found the AI team hadn't make progress in resolving this, which could hinder adoption and jeopardize product launch.
Approach
I worked with AI researchers to understand the model’s complexities and brainstormed solutions. With the in-depth understanding of AI model, I proposed a human-in-the-loop visual investigation tool to allow security experts validate AI predictions with a more efficient deep-dive.
I rallied the whole project team behind this product pivot, developing a functional prototype in two weeks to address technical and scope concerns.
Result
The functional prototype tested positively with our users and validated our hypothesis. The proactive pivot addressed key stakeholder's concern, delivered the product's promised potential and saved 3 month+ development time.
Deliver
Design System
Challenge
I needed to establish a brand from scratch, built quality into it to convey quality and trust, with limited time and resource (me as a solo designer and working with a full-stack developer).
Approach
Drawing inspiration from IBM Plex Sans typography, I crafted a command-line inspired interface that resonates with expert users' familiarity with retro, dark command-line environments. I directly implemented designs in CSS and JavaScript/React to ensure precise translation from design to code.
Result
My hands-on approach significantly accelerated design velocity, enabling our two-person team to deliver the design in just six weeks. The UI was praised by customers and the Director of Product as a key differentiator, setting a high-quality standard that facilitated future product expansions. I also guided the offshore remote team in adopting this design system through documentation, design guidelines and training.
Impact
Graphen's AI monitoring solution revolutionized enterprise cybersecurity by integrating a graph-based AI, tailored and intuitive interface and a innovative way to validate predictions, leading to a 70% increase in incident analysis and a 40% improvement in threat detection. My product pivot, design leadership and design to execution secured a $2M contract, paving the way for future expansions and product adoption.
In just nine months, I spearheaded the design and implementation of Graphen's AI monitoring suite, built trustful partnership with product, engineering and research and transformed the organization into a customer-led, high-functioning team. This strategy significantly sped up product development and positioned Graphen for ongoing success in the competitive cybersecurity market.
Reflection & Next Step
I was surprised by how much of the problem lies in optimizing and understanding of the human process, rather than pushing the boundaries of technologies. I have adopted human empowerment and collaboration as my core design tenant, and it has kept me humble ever since.