Voyance Vision
Leveraging AI to automate document processing, reducing manual intervention by 80% and accelerating processing speed by 90%
Situation
Businesses generate and process thousands of documents daily — invoices, contracts, compliance forms. Managing these manually is time-consuming, prone to human error, and costly. Traditional document management systems often struggle with low-quality images, poor scanning, and inefficient search capabilities.
The fintech sector in Nigeria faces stringent Know Your Customer (KYC) regulations, requiring businesses to verify customer identities rapidly and accurately. The existing manual processes were slow, expensive, and error-prone — highlighting an urgent need for an intelligent, automated solution.
Task
As lead designer, I defined the overall design direction, developed prototypes, and crafted research documents to guide the project. I created a design strategy document that outlined our vision and principles, ensuring alignment across teams.
Beyond design, I worked closely with Project Managers to align the design with business objectives, Data Scientists and Engineers to accommodate technical requirements, Frontend and Backend Developers to ensure accurate implementation, and the CEO — a software and machine learning engineer — to gain deeper technical insights and refine the user experience.
Action
Research and discovery
To create a solution tailored to real-world business challenges, we conducted research with fintech businesses. Our findings highlighted critical pain points: manual document processing was slow and error-prone, businesses wanted to minimise manual intervention for cost reduction, and KYC and other regulatory processes needed to be streamlined for compliance.
These insights guided our two-phase development approach: improving speed and accuracy by refining AI-driven data extraction, and enhancing automation by building workflows that reduce manual effort. We adopted rapid prototyping to test concepts and iterate quickly based on feedback.
Trainable AI models
The core product allowed businesses to train custom models or use pre-trained models developed by Voyance's data engineers. This enabled domain-specific document extraction without engineering overhead — a fintech company could train a model to recognise their specific KYC documents, while a logistics company could train one for shipping manifests.
Seamless upload, export, and annotation
Documents could be uploaded from local or cloud storage, with extracted data downloadable in multiple formats. Advanced annotation tools let users label and annotate documents to improve data accuracy — creating a feedback loop where the system got smarter with use.
Workflow integration
To reduce human effort through automation, I designed a workflow system that could instruct the AI to extract data during predefined events. Rather than requiring someone to manually trigger extraction, the system processed documents automatically as part of existing business workflows — cutting out an entire layer of manual intervention.
Testing and iteration
During prototype testing, we collaborated with external data scientists, engineers, and machine learning experts. Their feedback led to two major enhancements: initially, we assumed all documents would be in image format, but user testing revealed a significant need for PDF upload support. Businesses also wanted to integrate our AI models into their own systems, prompting us to develop an API integration layer.
By iterating based on real user feedback, we ensured Voyance Vision met industry needs effectively — not just the needs we assumed existed.
Bridging knowledge gaps
As a designer without a background in OCR, machine learning, or AI, I had to rapidly upskill to create intuitive interfaces for technically complex workflows. Regular documentation and sync sessions helped keep product, design, and engineering teams aligned — ensuring the interface reflected the true capabilities and constraints of the underlying technology.
Result
The launch was a resounding success. The implementation of Voyance Vision delivered substantial impact across the board.
- 80% reduction in manual intervention — documents processed automatically without human prompting
- 70% decrease in human resource dependency — teams redeployed to higher-value work
- 90% faster document processing — what took hours now took minutes
- Instant approvals and seamless customer activations, significantly improving end-user experience
- API integration enabled businesses to embed Vision's AI into their own systems
Reflection
Voyance Vision was a challenging but rewarding project. By blending AI, user-centric design, and automation, we delivered a system that transformed how businesses handle documents. The experience reinforced the importance of user research, iterative design, and technical collaboration in building impactful products.
While Vision started as a document management solution, its AI capabilities opened doors to broader applications — fraud detection, healthcare record management, legal contract analysis, and insurance claims processing. The foundation we built was designed to scale.
Automate the tedious. Empower the human.