Abdallah Elraai

Case file Exhibit A·1

CV ↓

Exhibit A·1Syntheia (opens Syntheia in a new tab)

The MFN App

A contract-review pipeline that ran on re-keying and an email chain. I mapped it, then argued for deleting most of it.

Matter
Syntheia (opens Syntheia in a new tab) — Legal-tech SaaS · enterprise → self-serve B2B/B2C
Role
Product analysis · IA · interface
Domain
Enterprise contract review
Term
Jan 2025 — present
  • Workflow analysis
  • Process design
  • Requirements
  • Enterprise
  • Legal tech

The workflow existed long before the software did. A contract arrived, someone typed its terms into a tracker, someone else circulated it for sign-off by email, and someone typed the outcome back in. Every organisation I have seen with this problem asks for the same thing: a better tracker.

I mapped the path a contract actually took — not the documented one — by sitting with the people doing it. What came back was not a list of screens to design. It was a list of steps that did not need to exist.

The redesign turned five handoffs into one guided flow. Capture happens once, at intake. Approval happens in place, inside the same context that shows you what you are approving. The tracker stopped being a destination you visit and became a by-product of doing the work.

Re-key the terms into the tracker.Capture once, at intake.

Every re-key was a fresh opportunity to be wrong, and the errors surfaced weeks later in the wrong document. This did not need a better interface — it needed to not exist.

Circulate for sign-off by email.Approve in the flow, in place.

The email chain was where status went to die. Nobody could answer "where is this contract" without interrupting a person, so the real cost was two people, not one.

User flow · before What a contract actually did
yes no Contract in email attachment Re-key terms into the tracker Circulate sign-off by email Approved? Re-key outcome into the tracker Filed eventually Chase status unknown

Two of the six steps were the same person typing the same terms twice, weeks apart. The loop had no status of its own — “where is this contract” could only be answered by interrupting someone.

Before — 5 handoffs
  1. Receiveintake
  2. Re-keymanual entry
  3. Emailstatus unknown
  4. Sign-offchase
  5. Re-keymanual entry

After — 1 guided flow

  1. Intakecapture once
  2. Reviewin place
  3. Approvein flow
  4. Fileautomatic
User flow · after What it does now
yes no Intake capture once Review terms in place Approve? Filed automatic Request change in thread

Same decision, one context. Capture happens against the document instead of after it, approval happens where the terms are visible, and the record writes itself. Five handoffs became one guided flow.

Screen structure · redrawn Report → MFN → provision record
Compendium report
MFN mode
Provision record
  1. The mode is entered from the report it belongs to, not from a separate app. One toolbar action, and the URL and the border both change — no new place to learn.
  2. Status is a tri-toggle in the cell: electable, disclosable, excluded. It is the one place in-cell interactivity earns its keep, because the decision is the row — putting it behind a panel would mean opening 41 panels.
  3. Everything that is not the decision lives in a right panel over the same table, so the list you were reading is still behind it, still in the same order.

Three screens, one surface. The classification happens on the row that already holds the provision, so the record is written by doing the work rather than typed into a tracker afterwards. Structure and flow are the real ones; every value shown is a placeholder.

  • −40%

    Legal processing time

  • 0

    Manual data-entry errors

On reflection — The measurable win came from removal, not addition. If I ran it again I would push harder, earlier, for access to the people doing the work — the first two weeks of assumptions cost more than they should have.

Syntheia is under NDA. Every interface on this page is a redrawn abstraction — real structure and real decisions, no real screens and no client data.