AI-Assisted Partner Applications, End to End
Category
Compliance & Automation
/
/
At UNYX, applying to financial institutions for partnerships was a fully manual job: gathering entity details, preparing regulatory documentation, and tailoring each submission to the institution's compliance requirements. It was slow and repetitive, so I rebuilt it into a structured, AI-assisted system.
The challenge
Every partnership application to a financial institution was assembled by hand, which made the process slow, inconsistent, and hard to scale.
Entity details and regulatory documents (certificates of incorporation, bank statements, source of funds evidence, investment letters) had to be gathered and prepared for each application.
Every institution's due diligence had to be completed by hand, whether an online form or a Word or PDF questionnaire, and re-done from scratch for each one.
Submissions had to be tailored to each institution's specific compliance requirements.
Compliance documents were scattered across multiple Google Drive folders, with no clear view of what was current.
The approach
I rebuilt the process in four phases: a clean documentation foundation, then automation, then a way to choose partners, then full ownership of the workflow end to end.
Phase 1, documentation audit and remediation
Audited all compliance documentation across Google Drive, categorizing every document by type and validity, flagged expired ones and either renewed them with external legal counsel or prepared new ones directly, then organized everything into a single, structured repository ready for any application.
Phase 2, AI-assisted applications
Built AI agents that read the documentation and auto-populated application forms and due diligence questionnaires, adapting the output to the type of partnership, extended them to review incoming membership agreements and NDAs and flag clauses misaligned with the business model, and built in pricing logic across monthly, integration, and rail fees so the team had clear benchmarks to evaluate and negotiate offers.
Phase 3, partner selection intelligence
Catalogued each partner's offering and built a simple tool where any team member can quickly identify the best partner to process a given transaction, weighing not just fees but accepted jurisdictions, transaction categories, …
Phase 4, end-to-end ownership
Tied it together into a structured, repeatable process covering the full lifecycle, from submission to document exchange, agreement review, and pricing negotiation, replacing ad-hoc decisions with a clear workflow.
How it stays accurate: the agents handle the drafting, but a person reviews and approves every application before it goes out. As that work is validated across more cases, the next step is a dedicated AI quality check to review applications automatically, but for now a human stays in the loop.
The impact
The process went from manual and person-dependent to structured, automated, and repeatable.
A single source of truth for compliance documentation, ready to deploy in any application.
AI agents that draft and tailor applications and due diligence questionnaires, cutting manual preparation and keeping submissions consistent.
Faster, better-informed evaluation of partner offers, with clear pricing benchmarks and agreement review built in.
A simple way for any team member to pick the right partner for a transaction, by fees, jurisdiction, and transaction type.
A workflow that no longer depends on one person's knowledge.
A Process That Scales, Not a Bottleneck
Onboarding to financial partners is high-stakes and detail-heavy, and when it lives in one person's head it does not scale. By building a clean documentation foundation and layering AI on top, the work became faster, more consistent, and repeatable, so the team can pursue more partnerships without adding overhead.
