How to set up spending policies

A spending policy in Mylo isn't a settings form; it's your actual policy document. You add it as text or a PDF, Mylo reads it, and from then on every new expense is checked against it automatically. Flags show up with a reason; nothing is ever blocked. Here's how to set it up, end to end.

Who can do what

• Owners and admins manage the organisation policy, per-role policies, and per-person policies for their org.
• Every member can create and manage their own personal policy.
• A policy an admin assigns to a member is read-only to that member: they can see it, not change it.
• A connected accountant can manage a client org's policies on the client's behalf.

Members also see the org and role policies that apply to them (read-only), so everyone knows what their expenses are measured against.

Step 1: Add a policy

1. Open Settings → Spending policy.
2. Tap Add a policy.
3. Pick who it applies to:

- Organisation: everyone in your org. - By role: owners, admins, or members. - By person: one specific member. - Personal: just your own expenses.

1. Add the content: paste or type the policy text, start from a template, or upload a PDF (up to 10 MB; Mylo extracts the text).
2. Give it a title and save.

That's the whole setup. No rule builder: plain English like "Meals are capped at $75 per person per day. Receipts are required over $75. Alcohol is not reimbursable." is exactly what Mylo wants.

Step 2: Review the compiled limits

After you save, Mylo analyses the text and compiles the hard numbers into deterministic rules: spend caps per day/week/month/report, per-expense limits, "receipt required over…" rules, and outright bans. These are checked by arithmetic over real expense totals, not by a model guessing.

Open the policy to see the rules it extracted. You can:

• Edit a rule Mylo got slightly wrong.
• Add a rule of your own.
• Delete one that doesn't belong.

Rules you create or edit are pinned: if you later edit the policy text and Mylo re-compiles, your manual rules are never overwritten.

Step 3: What happens when an expense comes in

Every new or edited expense is checked automatically:

`` new expense → checking… → Policy OK / Policy flag ``

• The check judges the expense against every applicable policy at once: org + the creator's role + their personal policy, read together.
• The fast arithmetic rules run first; then Mylo reads the prose for judgment calls, and can search the submitter's own expense history for context: duplicates, split purchases, running daily totals.
• The verdict lands as a badge on the expense. A flag always includes a one-line reason, and when sibling expenses drove it, Mylo cites the exact expenses it looked at.

The check runs while the expense is still a draft, so submitters see the verdict before anyone else does. If your drafts are trending over a cumulative cap, Mylo warns you before you submit.

Step 4: When something is flagged

A flag is a heads-up, not a wall:

• The expense still submits normally; Mylo never blocks submission.
• In the approval queue, flagged expenses carry a Policy flag chip with the reason, so reviewers spot them instantly among the clean ones.
• The reviewer decides: approve anyway (the flag was fine in context) or deny with a reason so the employee can fix and resubmit.

<img src="../features/images/approval-queue-accountant.png" alt="Approval queue" width="900" />

Waivers: write your exceptions into the policy

Real policies have exceptions, and Mylo honours the written ones. Say your policy caps meals at $75/day but adds: "During approved conferences the meal cap is $150 per day."

When the $75 limit fires on a conference dinner, Mylo checks the written policies for an explicit exception, verifies its conditions actually hold (it can look for the conference registration in the person's expenses), and waives the flag only by quoting the exact sentence from your policy. The expense then passes with an audit note showing which limit fired and which written exception cleared it.

What never happens: a waiver out of fairness, precedent, or common sense. No written exception, no waiver: the flag stands and a human decides.

Conflicting policies get flagged too

If two applicable policies contradict each other on an expense (the org policy caps meals at $75/day while a role policy says $150/day, with no stated precedence), Mylo flags the expense and describes the conflict, naming both policies. The fix is editing the documents to say which one wins (a policy that explicitly overrides another is not a conflict).

Good to know

• Flag-only. The policy check never blocks, auto-denies, or edits an expense. Humans make every final call.
• Edits recompile. Change the policy text and the compiled limits are re-extracted automatically; title-only edits don't trigger a re-analysis.
• Deleting a policy stops its checks. New expenses are only judged against policies that currently apply.
• Currencies aren't converted. Limits compare against the expense in its own currency.
• Everything leaves a trail. Reasons, cited evidence, and waiver notes stay on the expense for audit.

Want the bigger picture on why we built it this way? Read Write your expense policy in plain English. Mylo enforces it. For what happens after approval, see Submitting, approving, and denying expenses.