Treasury That Knows Who’s Asking
Employer: Altimetrik | Client: J.P. Morgan Chase
A modernization pursuit and proof of concept: how J.P. Morgan ACCESS could show a small finance team one thing and a global multinational another, without turning into two products.
I DESIGNED
I PROPOSED
TEAM OWNED
THE PROBLEM
ACCESS already did a lot right: broad treasury and payment coverage, deep institutional trust, and a scale most competitors couldn’t touch. Benchmarking it against CashPro, Wells Fargo, Citibank, HSBCnet, Kyriba, and SoFi confirmed that, and turned up the same few gaps every time: personalization behind Wells Gateway and Kyriba, mobile behind HSBCnet and SoFi, accessibility undocumented anywhere.
We weren't building five different products for five different people. We were building one system that reveals more as people are ready for it.
The job wasn’t inventing new capability. It was closing those specific gaps without breaking what already worked.
01 · STARTING POINT
Before proposing anything new, I needed to actually understand what ACCESS contained: payments, liquidity, receivables, FX, administration, security, connected systems. Reverse-engineering the live dashboard made that concrete: a twelve-section global nav, real-time and prior-day balance views, task-segmented pending actions. It reframed the question from “what widgets go on a dashboard” to “what does this user, in this role, at this stage, with these permissions, actually need to see?”
02 · AI AS THE CONNECTIVE LAYER
TREASURY ANALYST
CFO
AP SPECIALIST
SECURITY ADMIN
Same assistant, but it decides what to surface based on who’s actually asking. This built directly on Cash Flow Intelligence, which J.P. Morgan Payments had already started shipping, real forecasting built into the dashboard, extended toward something that adjusts by role instead of showing everyone the same thing.
03 · GROWING INTO COMPLEXITY
Persona research spanned four tiers: large corporate treasury, SMB, governance and compliance, and executive oversight. The insight that held across all of them: two people can do the identical task, like approving a payment, and still need very different amounts of context and control to feel confident.
A small business starts out initiating payments and pulling basic reports. As it grows, it needs layered permissions, foreign exchange, multiple entities, and eventually real compliance requirements. Personalization has to flex the same way. Client onboarding research showed the same idea in miniature.
STAYS CONSTANT
FLEXES
TARGET: cut a manual setup that takes 7 to 10 days to under three with a guided, template-driven version.
04 · PEOPLE + CONNECTED SYSTEMS
People and the systems talking behind their backs
Bigger organizations route most work through ERP systems, APIs, or host-to-host file transfer. Smaller ones still run on manual exports and spreadsheets. ACCESS had to work for both.
FINDING: Only Security Administrators could approve user requests or manage entitlements. There was no self-service option for a regular user to provision their own access. That shaped where automation could help, and where it couldn’t.
Human and automated work collide at three points: approvals, exceptions, and audits. The platform needed to show people what automated systems were doing without making them learn the plumbing, or figure out who had permission to act.
DELIVERABLES
A clearer model of what ACCESS does, personas spanning SMB to enterprise treasury, a maturity framework connecting growth to progressive complexity, and a conversational-AI concept grounded in analytics the platform was already building. A pursuit and proof of concept, not shipped.