
Stimma
Stimma is an AI agent built for the consumer, designed to save time, money, and hassle by calling customer service lines and resolving issues on behalf of users.
Conversational AI • Consumer Tools • Startup MVP
The Team
UX Team
• 1 (me) Product Design
Engineering Team
• 5 Developers
Business Team
• 1 Product Management
• 2 Finance
• 1 Marketing
Our cross-functional team brought expertise from:
My Role
I was the sole Product Designer on a cross-functional team of 10.
End-to-End UX – Defined the journey from bill upload to savings confirmation
UI & Web Design – Designed the MVP web app and marketing landing page
Visual Storytelling – Created pitch deck slides and demo assets for investors
Cross-Functional Collaboration – Aligned engineering constraints with user needs, and translated business strategy into user-facing design
Pitch Support – Contributed visuals and presented in two investor rounds
Outcome & Impact
Real Savings
✓ Negotiations with major internet service providers including Xfinity, Spectrum, Quantum Fiber.
✓ Average savings: $16/month ($192/year).
✓ Featured in GeekWire↗ after winning “Top VC Pick” in a UW Startup Showcase.
The Problem
Final Solution
Stimma is an AI agent that calls customer service lines and negotiates better rates.
By leveraging low-latency conversational AI technology, Stimma:
Engages in real-time voice-based negotiations with service providers
Navigates phone menus and converses with both AI agents and human customer service representatives
Proactively monitors each service provider's promotional deals and alerts customers when new promotions might save them money
Suggests potential savings opportunities when a promotional rate secured via a previous negotiation has likely expired
Requires minimal user intervention, handling the entire negotiation process from start to finish
We have successfully demonstrated our technology with major service providers including Xfinity, Spectrum, and Quantum Fiber, with customers already seeing real savings.
We have successfully demonstrated our technology with major service providers including Xfinity, Spectrum, and Quantum Fiber, with customers already seeing real savings.
Research Insights
From 30 user interviews who pay monthly internet bills, we uncovered key barriers:
AI trust: is it safe to provide my information?
Loss of control: anxious about letting AI handle everything
Design Implications
Transparency → progress updates, clear confirmation screens.
Trust → emphasize security + human in the look monitoring in MVP phase.
Low effort → minimal steps (bill upload → confirmation → result)
We also ran experimental Ad campaigns on Facebook and Instagram, and we found that no one likes calling their ISP. In fact, the click through rates outperform the average CTRs on these platforms by about 2%.
The Challenge
Business challenges
Users want lower bills but hate negotiating → frustration, long hold times, upselling.
Market has competitors (DoNotPay, Billshark), but they’re expensive or manual-heavy.
Goal: validate “real savings” quickly to attract customers and investors.
UX challenges
AI Trust: “Can I trust an AI with my sensitive info?”
Loss of Control: “Shouldn’t I be involved if someone negotiates my bills?”
Simplicity: Users wanted minimal steps, but needed enough feedback to feel safe.
Early-stage Demo
MVP Flow
In order to validate the idea, we bootstrapped a process where users send bill details via text or email, and our team manually kicked off the AI-powered negotiations. To maximize learnings in this early phase and to minimize risk of negative outcomes, a human team member monitored a live transcript of each call. Following the call, a team member sent a message to share the outcome of the negotiation. The calls themselves were guided by custom prompts, which changed depending on the provider. We used third-party services to place the calls, navigate the phone menus, and generate an AI voice.
Users texted/email their bill.
AI bot called providers and negotiated.
A human(our team) monitored live transcripts for quality.
Users received outcome report by text/email.
Here’s a real recorded call between an actual customer and a live service agent, fully handled by Stimma’s AI.
Note: This recording has been edited for the sake of time and customer privacy.
MVP UI
Web Flow
Based on our findings from the demo, I created a UI since we will mature our processes and our product, the user experience will be tighter. The general flow will be the same: the user securely shares information, goes about their day, and receives a notification about the outcome of the negotiation. The screens below showcase how this flow will take shape in a web app, though users will have the option to submit requests to Stimma via text message as well.
I designed a 5-step journey for the MVP web app:
Upload Bill
The user begins by uploading a recent internet bill, allowing Stimma to collect the account information required for negotiation.
Bill Preview & Extraction
Once uploaded, Stimma displays a preview of the bill and extracts relevant details.
Confirm Details
The user reviews the extracted information, such as provider name, account number, and current rate. They can make corrections or add any missing information before proceeding.
Review & Submit
Stimma presents a confirmation screen showing what action will be taken (e.g., negotiating with Comcast for a better rate). The user selects a voice (or records their own), adds any additional details (e.g., “I’m willing to downgrade my service by up to 100 Mbps in order to save more”), and submits the request.
Success Confirmation
Once the negotiation begins, the user sees a confirmation page. The user can opt to track the progress in real-time or go about their day. Regardless, they’ll receive an update following the negotiation.
Business Context
Market Opportunity
$50B+ annual U.S. consumer overpayment (internet, phone, utilities).
Existing players focus on subscription cancellation or one-time deals, not automated ongoing negotiations.
Potential to scale into a universal “bill autopilot” platform.
Competitive Strategy
Most competitors rely on manual negotiation and charge upfront or subscription fees. For example, DoNotPay requires a $36.99 bi-monthly subscription with no guaranteed savings. Stimma eliminates this risk: if we don’t save users money, we don’t take their money. This approach lowers barriers to adoption and attracts a larger customer base.
Learnings & Reflection
Trust > Features: Trust-building is design’s main role. Users cared less about features and more about security & transparency.
Manual-AI hybrid = faster validation. Early human-in-the-loop gave us user insights without breaking trust.
Storytelling drove adoption: My pitch visuals and landing page were critical in converting both investors and customers.
This project pushed me to go beyond UI and own the entire product narrative: user flows, brand trust, investor decks, even ad creatives. As the sole designer, I had to wear multiple hats—UX strategist, UI builder, storyteller—while ensuring design consistently supported both business growth and user confidence.
It reinforced my belief that in emerging tech, design isn’t decoration—it’s the bridge that makes disruptive ideas feel usable, safe, and human.