Files

adroit /Fundline· Case study

Free design review

Case study

Fundline

Cash-movement approvals for a brokerage back office, with reports and email

A back-office tool that decides which money may leave or enter a brokerage customer’s account: requests arrive, are checked against the account on file and the sanctions list, are approved by the right desk under rules an admin sets, go to the bank, and leave a trail anyone can read.

Product designBrand identityThe people, maps and journeys on this page are who and what we designed for, drawn from the brief and the product. They are assumptions to check with real users.

Web app for desktop browsers, designed at 1440 wide, with PDF reports and HTML emails

The challenge

Every deposit and withdrawal at a clearing firm needs a check, a decision and proof. Treasury platforms are built for corporate treasury, and clearing services offer the rails without an operator’s approval screen. The back office needed one place where the right person approves the right request, quickly and on the record.

What we did

  • Product design
  • Approval workflow and rules
  • Brand sheet and components
  • PDF report and email templates

What changed

  • An approval queue where the account-match check is the first thing a reviewer reads, and approve or reject sits on the row.
  • Approval rules written as sentences (amount, then the desk that must approve), set by an admin without a developer.
  • A four-eyes setting by desk, kept apart from the amount rules.
  • An audit trail that reads one request as a story, from created to screened to approved.
  • A dashboard per desk, plus emailed digests and PDF reports for the paperwork that follows.

Overview

Fundline is a web app for the cash desk of a brokerage clearing firm. It holds the bank accounts customers may move money to and from, takes deposit and withdrawal requests (ACH, wire and linked-account deposits), routes each one to the right approver, and hands approved money to the bank as a file.

Four desks use it: Admin, Treasury, Margin and Compliance. They see the same screens; what changes is which requests wait for them.

The design file runs from a brand sheet (colours, type, buttons, inputs, icons) to high-fidelity screens for each desk, dialogs for every decision, PDF report templates and notification emails. This case study shows the design; the sample data is invented.

The challenge

Moving a customer’s money is easy to do and expensive to get wrong. A withdrawal to the wrong account, a payment to a sanctioned party, or one person approving their own large transfer: each is a loss or a finding.

Before a request leaves, three questions need answers: is this the account we have on file, is the payee clear of the sanctions list, and has the right desk said yes. Afterwards comes a fourth: can we prove it?

The tool had to answer all four on the screen where the work happens, for desks that clear many requests a day before the bank cut-off.

The landscape

We reviewed seven payment-operations, treasury and clearing products that approve or move money. Every claim links to its source; where a vendor page blocked us, the claim comes from a search summary and says so.

In the matrix, "no" means not shown publicly and "unknown" means we could not verify it. The Fundline column reflects the design file, not a shipped build.

Showing 9 of 9 features

The landscape. Each cell says whether a product has a feature: yes, partly, no, not known, or not public (it may exist but nothing public shows it). Our product is the first column.
FeatureOurs
Amount-threshold approval rulesYesYesYesYesYesYesPartlyNot known
Role-based approversYesYesYesYesYesYesNot knownPartly
Four eyes, no self-approvalYesYesYesPartlyNot knownYesNot knownNot known
Account-match check on each requestYesNot knownNot knownPartlyNot knownNot knownNot knownNot known
Sanctions screening in the flowYesNot knownNot knownYesYesNot knownNot knownNot known
Audit trailYesYesNot knownYesPartlyNot knownNot knownYes
Bank file or wire statusYesNot knownNot knownYesYesNot knownPartlyPartly
Emailed digests and PDF reportsYesNot knownNot knownNot knownPartlyPartlyNot knownPartly
Built for a brokerage cash deskYesNoNoNoNoNoYesYes
  • Yes
  • Partly
  • No
  • Not known
  • Not public

Where we stand out

  • Approval built for a cash desk

    The approvers are the back office’s own desks (Treasury, Margin, Compliance) and requests carry back-office account numbers and firm names. The treasury platforms we reviewed target corporate payments; the clearing services offer rails without a documented approval screen.

  • Checks on the row, not in a report

    Account match is the first column of the queue and the sanctions check is its own step in the audit trail. Kyriba and HighRadius document screening as a step in the pipeline; Fundline puts the result where the reviewer reads.

  • Rules a desk can read aloud

    Each rule is a sentence: amount greater than or equal to a figure, then the desk that must approve. A separate setting adds four eyes by desk.

  • The paperwork goes out by itself

    A digest of requests waiting, a daily email with the outgoing bank file as a PDF, and report templates with rules for wide tables. Among the tools reviewed, only alerts are documented.

Where they do better

An honest read: these are the places the others are ahead today.

  • Rule depth: Modern Treasury combines conditions in AND / OR blocks and makes every reviewer approve again after an edit; Atlar adds tiers, fallback chains and a creator-approval switch.
  • Bank reach: Kyriba connects to over 10,000 banks in ISO 20022 and SWIFT formats. Fundline sends to one bank.
  • Screening breadth: Kyriba lists several sanctions lists and anomaly detection; the design file shows one sanctions check.
  • Self-serve simplicity: Mercury’s owners set an approval rule in settings. Fundline needs an admin role and its own setup screen.

Sources

The landscape

Modern Treasury

Companies that move money through bank APIs, and their payment operations teams.

Audience
Companies that move money through bank APIs, and their payment operations teams.
Website
docs.moderntreasury.com/payments/docs/approval-rules-overview (opens in a new tab)

Strengths

  • Approval rules on payments and on new external bank accounts, with conditions in AND / OR blocks.
  • Reviewer groups with a set number of reviewers; the last person to edit a payment cannot review it.
  • Every rule change and review is logged in an audit trail and notifies other admins.

Gaps

  • Amount and account conditions come from a journal post (search snippet); the docs page gives no field list.
  • Sanctions screening, reports and emailed digests were not verified.
  • Built for payments in general, not a brokerage cash desk.
The landscape

Atlar

Corporate treasury teams.

Audience
Corporate treasury teams.
Website
docs.atlar.com/docs/approval-chains (opens in a new tab)

Strengths

  • Approval chains with amount tiers, from automatic approval for small payments to a single senior role for the largest.
  • Each step names roles and a number of approvals; one person cannot approve two steps of a chain.
  • Creator approval is off by default, which enforces four eyes.

Gaps

  • The page does not describe an audit log, sanctions screening or reports.
The landscape

Kyriba

Large corporate treasury.

Audience
Large corporate treasury.
Website
kyriba.com/products/payments (opens in a new tab)

Strengths

  • Multi-level, rule-based approval workflows with digital signatures.
  • Sanction screening, bank account validation and anomaly detection on every payment.
  • Connects to over 10,000 banks in ISO 20022, SWIFT and local formats, with full audit trails.

Gaps

  • Up to nine approval levels and the OFAC, UN and EU lists come from a fact sheet and press coverage (search snippets).
  • Dashboard design and emailed digests were not verified.
The landscape

HighRadius

Corporate treasury and accounts payable.

Audience
Corporate treasury and accounts payable.
Website
highradius.com/software/treasury-risk/treasury-payment/payment-workflow-and-approval (opens in a new tab)

Strengths

  • Approval rules on amount, type, method, bank account, entity and currency, routed by role.
  • Fraud detection and sanctions screening before batched payments go to the bank.
  • A dashboard of payment status and history, with alerts for approvers and for failed payments.

Gaps

  • The page promises transparency but does not describe an audit log screen; email is not mentioned.
The landscape

Mercury

Startups and small businesses banking with Mercury.

Audience
Startups and small businesses banking with Mercury.
Website
support.mercury.com/hc/en-us/articles/31277592110356-Navigating-approvals (opens in a new tab)

Strengths

  • Amount-threshold approval rules that stack, with any admin, a department lead or a named person as approver.
  • Dual admin approval above a daily limit, requested by email and in the dashboard.
  • Separation of duties stops people approving their own payments.

Gaps

  • All claims come from search summaries; the help centre blocked direct reading.
  • A bank’s own approval layer, not a back-office tool.
The landscape

Apex Cash

Broker-dealers, advisers and fintech apps that clear through Apex.

Audience
Broker-dealers, advisers and fintech apps that clear through Apex.
Website
apexfintechsolutions.com (opens in a new tab)

Strengths

  • ACH deposits and withdrawals, scheduled ACH, wire and check withdrawals, and cash journals between customer accounts.
  • Transfers above a risk threshold go to Apex for review.

Gaps

  • All claims come from search summaries and an SDK readme; the overview page now redirects.
  • No operator approval screen, configurable rules or audit view was found in public.
The landscape

BNY Pershing

Advisers, operations teams and clients of a custodian.

Audience
Advisers, operations teams and clients of a custodian.
Website
bny.com/pershing/us/en/solutions/wealth-technology/custody-brokerage-account-management.html (opens in a new tab)

Strengths

  • ACH, wires, checks, journals and account transfers, recurring, same-day or immediate.
  • Built-in approvals for multi-step account processes, centralised alerts and complete audit trails.

Gaps

  • The page ties approvals to account changes, not to money movement.
  • Sanctions screening in the interface was not verified.

Who we designed for

Someone we designed for, drawn from the brief and the product. Not a real person, and not yet checked with real users.
  • Treasury approver who clears deposits and withdrawals

    “Show me the risky ones first. Let me clear the rest in a click.”

    Goals
    3
    Frustrations
    3
  • Compliance officer who owns sanctions checks and four eyes

    “If it isn’t in the trail, it didn’t happen.”

    Goals
    3
    Frustrations
    3
  • Operations admin who sets limits, desks and banking instructions

    “I want to read the rules out loud and have them make sense.”

    Goals
    3
    Frustrations
    2
Who we designed for
Someone we designed for, drawn from the brief and the product. Not a real person, and not yet checked with real users.

The release desk

Treasury approver who clears deposits and withdrawals

“Show me the risky ones first. Let me clear the rest in a click.”
Illustrative line, not a recorded quote

Context

Works the queue for most of the day, with peaks before the bank cut-off; approves from a list, not from a form.

Goals

  • Clear the safe requests fast, before the bank cut-off.
  • Never release money to an account that is not on file.
  • Know at a glance what is waiting for this desk and what is not.

Frustrations

  • Opening each request to find out whether the account matches.
  • Approve buttons on requests that are already decided.
  • A queue that mixes every desk’s work.
Who we designed for
Someone we designed for, drawn from the brief and the product. Not a real person, and not yet checked with real users.

The screener

Compliance officer who owns sanctions checks and four eyes

“If it isn’t in the trail, it didn’t happen.”
Illustrative line, not a recorded quote

Context

Reviews flagged requests, reads the audit trail and answers to regulators and auditors.

Goals

  • Every request screened before money moves.
  • No single person releasing a large amount alone.
  • Answer an auditor’s question in minutes, not days.

Frustrations

  • Proving afterwards who checked what, and when.
  • Screening results kept outside the approval tool.
  • Rules that changed without anyone noticing.
Who we designed for
Someone we designed for, drawn from the brief and the product. Not a real person, and not yet checked with real users.

The rule keeper

Operations admin who sets limits, desks and banking instructions

“I want to read the rules out loud and have them make sense.”
Illustrative line, not a recorded quote

Context

Visits the setup screens weekly; answers to the head of operations.

Goals

  • Change an approval limit without asking a developer.
  • Add a user or a bank account and know it is right.
  • Keep the rules short enough to explain.

Frustrations

  • Rules that live in people’s heads or in a spreadsheet.
  • Settings spread across several screens.

What they say, think and feel: The release desk

What we designed for, drawn from the brief and the product. An assumption to check with real users.

Out loud, in meetings and messages

  • “Does the account match?”
  • “Is this one mine or Margin’s?”

What goes unsaid

  • One wrong release costs more than a slow afternoon.
  • Most requests are routine; a few are not.

Observable behaviour

  • Scans the queue for anything not matched.
  • Approves from the row and opens only the odd ones.

The emotional undertow

  • Pressed before the cut-off.
  • Calm when the risky ones stand out.

Pains

  • Checks hidden behind a click.
  • Decided requests that still show buttons.

Gains

  • Match status as the first column.
  • Approve and reject on the row, outcomes as chips.

What they say, think and feel: The screener

What we designed for, drawn from the brief and the product. An assumption to check with real users.

Out loud, in meetings and messages

  • “Who approved this, and on what basis?”
  • “Was it screened before it went out?”

What goes unsaid

  • The trail is the product an auditor sees.
  • Four eyes only works if the tool enforces it.

Observable behaviour

  • Reads a request’s events in order.
  • Sets which desks need a second approver.

The emotional undertow

  • Exposed when evidence is scattered.
  • Confident when one row per event tells the story.

Pains

  • Reconstructing decisions from email.
  • Screening results in another system.

Gains

  • An audit trail with created, screened and approved as separate rows.
  • A reason recorded with every approval.

Release a withdrawal before the cut-off

What we designed for, drawn from the brief and the product. An assumption to check with real users.

The journey we designed for: The release desk6 stages

  1. Know there is new work and how much.

    Feeling today: Frustrated (−1)After the redesign: Neutral (0)

  2. See what is waiting for this desk.

    Feeling today: Frustrated (−1)After the redesign: Pleased (+1)

  3. Be sure the money goes where it should.

    Feeling today: Very frustrated (−2)After the redesign: Pleased (+1)

  4. Approve or reject, on the record.

    Feeling today: Neutral (0)After the redesign: Pleased (+1)

  5. Know the money left.

    Feeling today: Frustrated (−1)After the redesign: Pleased (+1)

  6. Answer any question about this request later.

    Feeling today: Neutral (0)After the redesign: Delighted (+2)

Key design decisions

  1. The match comes first

    Problem
    The costliest mistake is sending money to an account that is not on file, and the check for it was easy to skip.
    Choice
    Account match is the first column of the queue and the first chip on every dashboard card, written as Matched or Not matched in green or red.
  2. Decide on the row

    Problem
    Most requests are routine, and opening each one wastes the morning.
    Choice
    Open requests carry Approve and Reject on the row and on the dashboard cards. Decided rows swap the buttons for an outcome chip.
  3. Restate before you approve

    Problem
    An approval made from a list can miss a detail.
    Choice
    The approve dialog repeats every field of the request (amount, date, type, name, bank details, address, account type) and asks for a reason before Yes, approve. Reject and cancel have their own dialogs.
  4. Rules as sentences

    Problem
    Approval limits lived in people’s heads.
    Choice
    An admin writes rules as sentences in a table (amount greater than or equal to a figure, then the desk that must approve), with an editor row on top. Four eyes is a separate setting by desk.
  5. Status in words

    Problem
    Requests pass through many states, and colour alone cannot tell them apart.
    Choice
    Every state is a word in a coloured chip: Matched, Not matched, Approved, Rejected, Created, Pending, OFAC check, Auto approved, Scheduled for transmission, Transmitted.
  6. One shell for four desks

    Problem
    Four desks need different work but the same tool.
    Choice
    The same screens serve Admin, Treasury, Margin and Compliance; the header names the desk and the queue holds what waits for it.

Design targets

What the approval flow was designed to make true, written as figures to test it against.

Targets the design was built to reach, not measured results.

  1. Design target

    < 60 s

    Time to decide a routine request

    The account match and the screening result sit on the row beside Approve and Reject, so a routine decision never waits on another screen.

  2. Design target

    2

    Checks read beside the Approve button, with no extra click

    The account match and the screening result sit on the row, because a check behind a click gets skipped on a busy day.

  3. Design target

    0

    Approvals without a written reason

    The approve dialog asks why before Yes, approve, so every decision carries its reason into the audit trail.

  4. Design target

    0

    Developer changes needed to edit an approval rule

    Typical starting pointLimits kept in people's heads

    An admin writes rules as sentences in a table, so compliance can read them aloud and the desk can change them.

From request to decision

The dashboard splits today’s requests by desk and outcome, shows last week’s deposits against withdrawals, and lists the queue as cards a reviewer can clear in place.

The approval queue leads with the account match; approving opens a dialog that restates the request beside its checks. All screens are rebuilt as live HTML and SVG with sample data.

Operations dashboard
Approval queue
Approve a funds request

Rules and proof

An admin sets who approves what as a list of sentences, and which desk needs a second approver.

The audit trail keeps one row per event, so a single request reads from created, to screened, to approved.

Approval flow setup
Audit trail

A deep navy carries the product: the rail, titles and primary buttons. Sky blue is its partner for quiet states and secondary data; mint marks where you are.

Green and red mean approve and reject, nothing else. Every status pairs its colour with a word, on a soft chip.

The paperwork has its own voice. PDF reports switch to a warm peach so a printed page is never mistaken for the screen.

Brand: colour

  • Primary. Rail, titles, primary buttons, deposit bars

  • Rail icons, withdrawal bars, selected filters

  • Date badges and soft highlights

  • The current page in the rail

  • Secondary brand colour

  • Approve buttons and approved text (read from the frames)

  • Reject buttons and rejected text (read from the frames)

  • Tertiary. Export actions

  • Body text

  • Table headers and quiet pills

  • Page background and alternate rows

  • Messaging

  • Messaging

  • Messaging

  • PDF report headers and totals (read from the frames)

Brand: type

  • Inter

    Everything in the app. Headings bold at 40, 36, 32, 24 and 20; body semibold at 18, 16, 14 and 12; paragraphs regular at 14/22

    $48,200.00 · Matched

    • 400
    • 500
    • 600
    • 700

Brand: logo notes

  • The client’s logo is not shown, because the client is anonymised. The mark on the rebuilt screens is a neutral codename wordmark drawn for this case study, in the product’s navy, sky and mint.
  • The original pairs a small mark with a two-tone wordmark; the emails set it in a white header card between two dot-grid patterns.
  • Colour and type values come from the design file’s variables; the approve, reject and report colours were read from the rendered frames.

What the file covers

33 parts in 6 groups

The design file holds a brand sheet and components, high-fidelity screens drawn once per desk, the dialogs for every decision, PDF report templates and notification emails.

The groups below are its sections and the screens and states inside them.

    • Colours and messaging states
    • Type scale
    • Buttons and their states
    • Input fields
    • Icon sets
    • Card shadows
    • Banking instructions
    • Funds requests
    • Post a funding request
    • Approval queue
    • Wires transmitted
    • Automatic transmission
    • ACH file transfer
    • Approve dialog
    • Reject dialog
    • Cancel dialog
    • View details
    • Delete confirmation
    • Approval flow setup
    • Four eyes by desk
    • Audit trail
    • Configuration
    • Users and roles
    • Outgoing file report
    • Audit report
    • Wide report grids for 10, 13 and 17 columns
    • Digest of requests waiting
    • Outgoing file email
    • Login
    • Forgot password
    • New password
    • Notifications
    • Announcements

Reflection

Speed and control usually pull against each other on a cash desk. Here they meet on the row: the check sits beside the button, so the fast path is also the safe one.

Next, these design assumptions need sessions with a real cash desk, and the four-eyes setting needs a sharper definition: whether it asks for a second person on the same desk or a second desk.

The rule table is deliberately simple. If the desk outgrows it, the competitors show where to go next, with combined conditions and approval tiers.

Want something like this?