How Reme classifies risk levels

Note: This feature is part of the AI TC add-on. See Plans and pricing.

Every action Reme wants to take gets classified as LOW, MEDIUM, or HIGH risk before it ever reaches you. This article covers how that classification actually happens. For how to read and act on a risk badge in the Feed itself, see Reviewing and approving actions in the Feed.

How the classification works

Reme checks each action against a fixed set of rules, in this order:

1. Anything that requires a signature is automatically HIGH risk.
2. Anything addressed to your brokerage's document intake (for Your Castle agents, documents@yourcastle.org) is automatically HIGH risk.
3. Certain action types are always HIGH risk, regardless of content: an inspection objection, a contract amendment, a Commission Disbursement Authorization (CDA), or a wire fraud warning.
4. Certain action types are always MEDIUM risk: a disclosure package, an earnest money reminder, a request for a copy of insurance conditions (CIC), a lender follow-up, scheduling an inspection, a social media post, a home warranty transfer, or forwarding repair receipts. A lender-related email whose subject mentions a loan is also treated as MEDIUM.
5. Certain action types are always LOW risk: a deadline reminder, a routine status check-in, a thank-you email, marking a task complete, an MLS update prompt, adding a calendar event, or the daily digest.
6. Anything Reme doesn't recognize defaults to MEDIUM risk. When it's not sure, it asks rather than assumes.

What each level means for approval

Risk level	Supervised mode	Autonomous mode
LOW	Waits for your approval in the Feed	Executes automatically
MEDIUM	Waits for your approval in the Feed	Waits for your approval in the Feed
HIGH	Waits for your approval in the Feed	Waits for your approval in the Feed

Only LOW-risk actions are ever eligible to run without your review, and only on transactions set to Autonomous mode. MEDIUM- and HIGH-risk actions always wait for you, on every transaction, regardless of autonomy mode. See Supervised vs. Autonomous modes for how to set that per transaction.

Why this matters for CDAs and objections

Because a Commission Disbursement Authorization and an inspection objection are always HIGH risk by type — not just by content — they will never auto-execute, even on a transaction you've set to Autonomous. Both carry real financial or legal consequences if the wording is wrong, so Done-Deal keeps a human in the loop on both no matter what.

FAQ

Can I change what risk level an action type gets?

No. Risk classification is fixed by action type and content — it isn't a per-agent or per-brokerage setting. This keeps the approval requirements for legally or financially significant actions consistent for every agent.

What if Reme misjudges the risk of something?

Every action, at every risk level, is fully editable before you approve it — see Reviewing and approving actions in the Feed. If something reads MEDIUM but feels like it should have been HIGH to you, review it just as carefully before approving; the badge is a guide, not a substitute for reading the draft.

Does the risk level affect where an item shows up?

No — every item shows up in the Feed with its badge visible regardless of risk level (unless it auto-executed as a LOW-risk Autonomous action, in which case it's logged rather than queued). The risk level only affects whether it waits for you first.