Real time Onboarding
Fintech
Image to intelligence
Growing 30% faster month on month with real time onboarding powered by Metaphi's vision AI platform
Learn how Metaphi helped one of the largest payment companies in the world increase new acquisitions by 30%, with 75% lower manpower
30% month on month growth
24,000+ man hours saved
100% precision in compliance
"Fintech is a UX driven business, and every improvement in user experience from onboarding speed to personalized offers improves margin efficiency in the business. We have a strategic roadmap to invest in AI to continuously improve our customer experience."
V.P operations
Today, our new merchant onboarding process is dependent on a manual human team to validate documents and conduct quality checks on the place of business. This is a problem because it creates friction, and second is not scalable for our hypergrowth face. But the problem was complex because it involved building a real time vision generative AI solution, which would require significant resource investment and experience in house. With Metaphi's visual generative AI platform, we are achieving a 75% success rate in auto onboarding. It is a game changer.
The problem
Our customer, one of the world’s largest payment companies, had an onboarding nightmare. Their core target segment is fast growing merchants and the onboarding process is largely assisted journeys. Naturally in a heavily regulated industry like payments, the onboarding process requires an auditable trail of quality checks conducted on:
The physical place of business
Eligibility of the merchant based on established criteria including line of business aka Merchant category code (MCC)
Verification of the business ID documents
Onboarding merchants for digital payments with a human based quality check process is a nightmare
24,000
man hours
Time spent in onboarding today
10-30
minutes
Onboarding time which creates a bad UX
30%
inaccurate
Compliance risk from human errors
The process today is highly human dependent with a backend human operations team that manually validates 20+ parameters for over 20,000 merchants onboarded every single day! That is over 800 man hours of work that goes in for an extremely repetitive, mundane task. The implication? Merchants do not get onboarded easily, and the ground field sales agent has to wait 30+ minutes before receiving status on the application for the merchant they have acquired.
Bad UX for new merchants
Inefficiency in sales process
High manual costs
Searching for a solution
The fintech company had been searching and experimenting with many off the shelf computer vision solutions to replace manual quality agents with AI agents, but none of the solutions worked out of the box. Some were very expensive, some very inaccurate and almost all very inconsistent with no reliable confidence intervals that would enable the business leadership to trust an AI solution.
"Our strategic goal is to reduce our manual onboarding process with an AI led journey, but we cannot afford any false positives. Even if we start with a 40% reduction, that is a good start but we are yet to come across a solution that delivers on our requirement of zero false positives."
V.P operations
How did Metaphi solve the problem?
Metaphi's implementation of a Mixture of Experts (MoE) vision model in a real time, low latency environment with up to 100 requests per minute (100 RPM) is a one of it’s kind solution. The solution delivered 76% recall, and 100% precision enabling the fintech to solve 75% of their problem and change the game for millions of merchants.
Metaphi's image to intelligence solutions enables our fintech customers to transform their onboarding
Real time intelligence from images delivers unprecedented
UX benefits - creating faster growth at lower costs
Customize the parameters with 1 click
Merchant category code-aligned with visa mcc system
Is shop open?
Is shop movable or not?
Type of shop sign board?
Is inventory visible?
Is shop board fixed?
Is GST (unique tax number) number visible?
Line of business as per inventory
Line of business as per sign board
Validation on business id and proof documents
"Honestly, we were surprised that we could even get all the required parameters in the first place. At first, it was just an experiment for us, but seeing how quickly the Metaphi team executed on our custom parameters made us understand how differentiated their approach to vision AI is"
Lead PM, Onboarding
Metaphi implemented a Mixture of Experts vision model to draw intelligence from human generated images
Zero false positives
100% precision, 76% recall
7 seconds latency
Only such solution in the market
99.99%
uptime
Handle rate limits with Ensemble approach
AI that integrates as easily as an API
"We were worried about integration and implementation - specifically because of our low latency requirements. But with Metaphi, integration was just like any other standard API. Implementation was straight forward because of the sub 10 seconds latency which mean we could deliver intelligence at the point of onboarding without needing any change in the behaviour of the current field agents"
Lead PM, Onboarding
Vision Al that integrates as easily as a single API, right into your field environment
