Topogy (formerly Finte.Ai)
app.topogy.com/ recommendations
Role
Principal/Founding Product Designer
Team
CEO, CTO, & Me
Timeline
1 month, 2025
Impact
Discovered 200k overages for our biggest client
TAGS
FinOps
Information Architecture
Prototyping
Shipped
Tools
Check out the main flow
01
The Challenge
“What should we fix first to make the biggest impact on our spend?”
Engineering teams weren’t struggling to collect infrastructure data. They were struggling to understand where to act first. Finance could identify rising costs, but engineers were responsible for determining what could safely change, which systems would be affected, and who should own the work.
We kept hearing the same painpoints from our user research:
"Our AWS bill jumped $70K in one month."
"We burned through 60% of our logging credits in March."
"I think we're wasting 30% of our infrastructure spend."
Existing tools showed teams where money was being spent, but offered little guidance on what to fix first. We saw an opportunity to shift the experience from reporting costs to helping teams reduce them.
02
The Goal
"How might we turn infrastructure data into clear, prioritized action?"
I set out to design a recommendations experience that would help engineering teams understand where their money was going, identify the highest-impact savings opportunities, and know what to fix first.
The experience needed to:
Prioritize recommendations by potential savings
Put the ownership of cost optimization in the hands of those who have the ability to improve it
Provide enough evidence for engineers to trust the recommendation
Guide teams from discovery through implementation
81%
of respondents said their cloud costs were about where they should be when engineering had some level of cost ownership.
03
Design Principles
Before exploring solutions, I grounded the work in a few principles that could keep a complex feature human, scalable, and easy to recognize across the product.
💡
Insights over information
Simplify complex data into actionable, intuitive visuals that enable teams to quickly identify opportunities and act with confidence.
💪
Maximize impact, minimize effort
Streamline workflow to reduce manual processes, allowing users to focus on driving results.
🧩
Empower curious problem-solvers
Enable users to discover opportunities through flexible exploration and hierarchical information so they know where to look & how to make a difference.
04
The Solution
I designed a zero-to-one recommendations platform that analyzed infrastructure spend across providers like AWS, Datadog, Snowflake, and OpenAI, then surfaced prioritized opportunities to reduce costs.
Instead of asking engineers to manually investigate billing reports, the platform identified potential savings, explained why each recommendation mattered, estimated financial impact, and guided teams through implementation.
05
The Process
We built quickly, then reorganized around how engineers actually work
Research first.
For the first few months, I worked with the founders to interview engineering and finance leaders, test our assumptions, and learn from design partners.
Get real data into production.
Once the pain points were clear, we moved quickly. We needed a live product ingesting real infrastructure data so we could see where the design broke down.
Our first version grouped recommendations by spend category, such as Compute, Storage, and Networking. The structure made sense internally, but testing showed that it did not match how engineers actually owned the work. Engineers were trying to lower their AWS bill before renewal, not fix their compute. Their mental model centered around vendors and workflows, not infrastructure taxonomy.
We also found that most recommendations fell into a small set of repeatable jobs, including right-sizing, commitments, elasticity, and data cleanup.
06
The Iteration
Key insight
Engineers did not organize their work by spend category. They organized it around the systems they owned and the job they needed to complete.
That insight changed the information architecture. I reorganized recommendations around two layers: the type of optimization work and the provider or system affected.
1. Group by optimization job
Organize recommendations around repeatable work like right-sizing, commitments, elasticity, and data cleanup, so you can do multiple jobs with one effort.
2. Narrow by provider and system
Let engineers quickly find recommendations tied to the vendors, and services, so they can tackle cost spikes and anomalies as they come up or upon renewal.
This redesign allowed users to complete the entire workflow more efficiently. The interface shifted from a reporting tool into an operational workflow.
07
The Results
We launch quickly and became the first native-AI FinOps company, securing 2 customers within one month.
$200k
In Datadog overages discovered post launch.
2
We signed our first 2 customers!
"This is exactly what we need right now.. insights out of the box, tracking real outcomes."
L
CTO
Legal Tech Company
Would I change anything?
This project came at the beginning of my shift toward AI-assisted design. I used AI mostly for copy, while spending too much time perfecting Figma mockups. If I revisited the work today, I would use AI to prototype earlier, explore more directions, and validate ideas before investing in polish.
NEXT PROJECT
Calculated Fields
Designing a more human way to configure business logic.
Read next