Customer empowerment
AI helps our teams build and run better services — improving the quality, speed, and reach of the services we provide. It is designed to support human judgment on decisions that matter, not to stand in for it.
The EPCVIP AI Trusted Framework
How EPCVIP works to keep AI-assisted work safe, accurate, and accountable — the principles, standards, and safeguards built into the agents and sessions our teams use.
EPCVIP is a lead-generation and affiliate network in the fintech industry — we help people find and compare financial products. We are not a lender, and we do not lend, offer, price, fund, or hold money ourselves. We use AI within our own teams to build and ship the systems and processes behind these services.
This page describes how our engineering teams use AI coding agents to build and ship software. It does not describe how consumer information is collected, used, or shared.
The principles behind every agent
Eight principles shape how EPCVIP’s teams use AI to build software. They are the reason the standards and safeguards below exist.
Data stewardship
Customer data is entrusted to us, and the AI tools our teams use are built with that in mind. This page does not describe how it is collected, used, or shared.
Transparency
When AI helps with the work, our standards call for the people who ship it to be able to explain how it was made, in plain terms.
Fairness
Fair treatment of the people we serve is a standard our work is held to, whether or not AI helped produce it.
Security
The AI tools our teams use are designed with security in mind from the start — protection is designed in, not added later.
Accountability
A person answers for every change an AI tool helps make. AI supports people; it does not replace them on decisions that matter.
Continuous improvement
Every shortfall teaches us something. We are building ways to measure how well our safeguards work, and we keep improving them as AI changes.
Reliability
AI-assisted work is held to the same standards as anything written by hand, and our standards call for review before it ships.
Trust, built into the path of least resistance
The safe way is the easy way. EPCVIP’s standards and safeguards come built into the agent itself, so good results depend less on remembering the rules.
01
Shared principles
Eight principles set the direction. Every standard and safeguard below traces back to one of them.
02
Built-in standards
Around fifty coding and security standards travel with the agents we operate, so the right way is already loaded — nothing to find or set up.
03
Automatic safeguards
Quality and security checks run as you work to help verify each change before it ships, leaving a clear, auditable record.
04
Ready playbooks
If something slips, a documented response is ready, with clear roles for who runs it.
Safeguards for customer data, built in
We help people find and compare financial products — work that runs on sensitive data — so stewardship is built into how our teams use AI, and it is the commitment the rest of the framework is built to keep.
Safeguards throughout the work
Stewardship is built into how our agents work — not bolted on afterward.
- Agents connect to company data through reviewed, approved tools that are designed to limit what they can see.
- For testing, our standards call for synthetic or anonymized data rather than real customer records.
- We scan for secrets before each commit our AI agents make, to help keep credentials out of the codebase.
- We check the third-party code our software relies on for known vulnerabilities, and our standards call for addressing the serious ones before a change ships.
A person is responsible
AI is a tool, not the accountable party — responsibility stays with people.
- The developer who submits a change is responsible for it, whether or not an agent helped write it.
- Our standards call for review before changes ship, so quality doesn’t rest on the agent alone.
- Our standards call for each codebase to have a designated owner, so there’s a clear person to turn to.
- Our standards call for each change to be explainable by the person who submits it.
Informed by recognized frameworks
The principles, standards, and safeguards described here are informed by recognized frameworks like the NIST AI Risk Management Framework (AI RMF 1.0) and ISO/IEC 42001. We do not claim certification to either.
EPCVIP is a lead-generation and affiliate network, not a lender; it does not make, price, or fund lending decisions. This page is informational, describes our internal practices for how we use AI, and creates no contractual right or warranty.