AI/ML
Workflow Automation
Investigation Platform
SaaS
CISO Auto-Trace
Automates fund tracing, detects suspicious patterns, and generates reports in one click.
Year
2021-2024
Role
At a Glance
Auto-Trace automatically traces crypto funds from source to destination across multiple wallets and hops, helping financial crime investigators follow the money trail and detect suspicious activities in minutes instead of hours.
As the founding designer, I led the 0→1 design of Auto-Trace, owning the end-to-end UX, collaborating with data scientists and engineers to design the tracing logic within heavy-data constraints. I validated the workflow with investigators using real case data and iterated the product through multiple development phases.
Impact
85%
faster investigations
30%
customer growth
Problem
As crypto-related financial crimes surged in 2021, investigators had to manually trace wallet hops to follow laundering paths and fund destinations, the process was slow and difficult to scale.
Solution
Turning manual investigations into a fast, automated process that follows funds across blockchains, identifies key flows, helping investigators make faster, more confident decisions.
Context
CISO was built to help financial crime investigators visualize complex crypto transactions between wallets. It’s a powerful graph tool where analysts can expand nodes to make the connections with other wallets and customize the view to trace money flows. But as cases grew larger, efficiency hit a wall. Investigators were still expanding wallets one by one, manually filtering and guessing which paths mattered.
Data Overload with No Clear Guidance
The graph was just overloaded with data. Most of it came from unknown accounts, everything looked the same, and there was no guidance on where to focus. Investigators had to use filters and sort out the key paths, going down rabbit holes that often led to dead ends, which means no exchanges found, just wasted time.
Investigators spent most of their time on manual work instead of real analysis and reporting
Delays case delivery
Reduces client trust
Slows business growth
Findings from Observation
Funds Jumped Through Too Many Wallets
The graph relied on manual expansion. Money often flowed through dozens of wallets, forcing investigators to click each step, which was slow and time‑consuming.
Hard to Follow the Story of the Money
After expansion, analysts often faced overwhelming number of transactions per node. They had to sort by time and amount to figure out what mattered.
High-Risk Paths Hard to Spot Quickly
Analysts wanted the system to flag or rank risky paths automatically, so they didn’t have to dig through everything manually.
Analysts Created Workarounds
Noise forced analysts to document key paths in spreadsheets to build cleaner investigation reports, which made the process slower and harder to maintain.
Goal
Fast
Make tracing faster and surface key fund flows instantly.
From Manual to Automated Tracing
Introducing Auto-Trace
One-Click Tracing, Instant Results
Enter a wallet or transaction hash, and Auto-Trace follows the funds, highlights the most relevant flows, and identifies liquidation points.
No Risk Left Outside the Trace
Auto-Trace pulls in historical suspicious activity back in the graph, letting investigators review risky behavior alongside the flow.
Actionable Insights
Auto-Trace provides a clear next step for every address at the last hop, offers one-click export, monitoring for inactive wallets, or Continue Tracing for active ones.
Clearer Graphs to Reduce Noise
Large graphs stay readable with clustering, filters, and dimming of low-risk nodes, keeping focus on what matters most.
Approach and Iteration
STEP 1
Investigation Workflow Observation
I held working sessions with investigators to observe decision making in live investigations. By mapping their actions and reviewing case reports, I saw that each decision wasn’t random, it followed a pattern.
Synthesized decision paths into an automated, multi-scenario workflow
✅ Outcome
Investigator constantly made the same four judgment calls:
STEP 2
Why Auto-Trace Needed Its Own Space
I explored two directions: adding Auto-Trace directly into the main graph or creating a separate mode.
At first, embedding it seemed more natural for users, one of the users even suggested to add custom rule logic to the graph, but testing with frontend engineers showed major issues. In The manual graph, all transactions are aggregated under one unique node, but Auto-Trace needs to duplicate nodes to show the real flow of funds. I chose to build it as a separate two-step flow, to make the result graph easier to follow.
✅ Outcome
Launched the Auto-Trace MVP in 3 months! 🚀
STEP 3
Tested MVP with real data to validate results and experience
PROBLEM
Setup felt unreliable, no clear guidance on trace limits, frequent failed traces, and slow loads when limits were maxed out.
SOLUTION
Simplified setup, refined trace limits, and added Auto-Detect for smoother, faster tracing.
Address or Transaction
Time Range
Sending or Receiving Limits
Depth and Breadth
Hops
Address expanded
Tracing Direction
INITIAL SETUP
Address or Transaction
Time Range
Tracing Direction
TRACE LIMITS
Expanded Addresses Limit
Hop Depth
+ Sort Transaction By
Date (Newest/ Oldest)
Value (High/ Low)
+ Amount Cap
TRACE RESULTS
+ Continue Tracing
Refined setup limit for better control and faster performance
Setup page simplified to essentials, Auto-Detect prevents failed traces.
✅ Outcome
These changes made tracing faster, reduced failed trace by 95%, and gave investigators flexibility to trace deeper only when needed
STEP 4
Cleaning Up the Result View
PROBLEM
Large cases looked messy, too many nodes and colors made it hard to focus on key results. Setup, results and graph cramped in one view.
SOLUTION
Grouped addresses into clusters, highlighted only exchanges and high-risk wallets, and simplified the layout with collapsible panels for a cleaner, more focused graph.
Improved results page with more graph space, customization options, and clustered nodes.
✅ Outcome
Cleaner layout and clustering cut visual noise, making large traces easier to read.
High-risk entities and exchanges highlighted for quick focus on critical paths.
Customazition let investigators hide irrelevant nodes and stay focused.
STEP 5
Guiding the Next Step
PROBLEM
Investigators often got stuck at the last hop with no clear next step, leading to inconsistent workflows.
SOLUTION
Added clear next-step actions, one-click exports for exchange cash-outs, monitoring for wallets holding funds, and a Continue Tracing option for when the funds move further.
✅ Outcome
Every endpoint became actionable, no more dead ends.
Investigation kept flowing without interruption.
The Impact
Since launching Auto-Trace, we’ve seen a significant shift in how our clients approach investigations. Auto-Trace evolved through multiple phases of prototyping and iteration, tested with investigators at every step. The feature quickly became the go-to starting point for both new and experienced investigators, saving hours of manual work and helping teams focus on what matters most.
Investigation Time Reduced by 85%
Auto-Trace provides us with real time-saving functionality. I have compared this to some live jobs here. It recreated investigations in 10 minutes when it took the investigating officer hours to complete using other tracing software.
Senior Investigator, UK Law Enforcement Agency
Strong enforcement of forward and backward Auto-Trace helped us to create a better analytical framework for blockchain tracing.
Analyst, Federal Financial Enforcement Agency
Customer Growth Increased by 30%
Adopted as a Screening API
For some of our largest clients in high-security and defense, being able to trace the flow of funds at scale is critical. We delivered Auto-Trace as an API, allowing quickly screen batches of addresses to see where funds came from and where they ended up, especially if they reached the liquidation points.
Future Vision
The story isn’t finished. One of the toughest challenges is still cross-chain tracing, where tokens move across different blockchains. This requires manual handoffs and separate data sources, and building a seamless multi-chain solution is our next frontier.
For me, Auto-Trace shifted how I design complex tools: not just making them faster, but keeping them manageable while ensuring decisions stay clear. As we continue to improve, our vision is to bring Auto-Trace directly into the manual graph experience, so analysts can blend automation with hands-on review in a single workflow.
This gave investigators control up front while keeping automation fast and explainable.