IQ preferences & memory: what IQ knows about you

What this page is

IQ is your AI Co-COO. It reads what flows in from your jobs, invoices, carriers, and inbox, and it surfaces decisions and drafts actions on your behalf. Two things make that trustworthy instead of unsettling: you can see the pattern IQ has picked up on how you engage with different kinds of decisions, and you can see (and edit, or forget) the durable facts it holds about you and your business. Both live under IQ Activity Log, in the same audit surface as every email IQ sent and every state change it made. Nothing on either tab is a peer benchmark or a network number, it is entirely about you and your own history with IQ.

Where to find it

Open Settings from the sidebar, then under Data & Privacy, click IQ Activity Log (subtitle: "Every action IQ took for you"). That takes you to /settings/iq, which redirects to the Activity log tab. Across the top of the page, six tabs run in this order: Activity log, Outbound emails, Inbound classifications, Conversations, Memory, Preferences. The two this article covers are the last two:

• Preferences (/settings/iq/preferences), IQ's inferred read on which kinds of decisions you act on versus pass on, and your override controls.
• Memory (/settings/iq/memory), every durable fact IQ has been told to keep in mind about you, your team, and your business, plus the tools to add, forget, or restore one.

The page header above the tabs reads IQ Activity, with the line: "Every action IQ takes on your behalf, every email IQ sends, every state change in your decisions. This page is the audit you can verify against, if anything here looks wrong, that is a signal we want to see." Preferences and Memory are part of that same audit posture, applied to how IQ decides what to show you and what it remembers.

Preferences: which decisions you want more or less of

What it is. Every decision IQ surfaces to you, benchmark flags, adjuster follow-ups, cost alerts, and everything else IQ classifies by signal type, and it watches whether you act on that kind of decision or consistently pass on it. When you keep passing on a kind, IQ quietly stops pushing it and sets it aside instead of nagging you with the same kind of item every week. The Preferences tab is where that inference becomes visible and editable instead of invisible.

The page. The header reads: "Which decisions you want more or less of," with this explanation underneath: "Kinds of decisions you consistently pass on get set aside instead of pushed at you. If that is wrong for any of them, say so here. Your choice always wins." If IQ has recomputed your profile at least once, a second line appears: "This updates as you act on decisions. A recent change may take a day to show."

The list. Below that sits a section headed "Based on how you've handled decisions (N areas)," where N is the count of decision kinds you have enough history with to show a rate. Only signal types where at least one decision has actually been acted on, resolved, or dismissed show up here, thin history does not get a row.

Each row shows:

• The decision kind, humanized (for example, a signal type like cert_expiring reads as "Cert expiring"). If it carries a technical coo_ prefix internally, that prefix is stripped before display, you never see raw system tokens.
• A status pill, only when one applies:

- Always surface (green), when you have told IQ to keep showing you this kind regardless of your own history with it. - Always park (amber), when you have told IQ to stop surfacing this kind. - Currently parking (amber), when you have not set an override, but IQ has quietly started setting this kind aside because your action rate on it has been consistently low. - Engaged (green), when you have not set an override and your action rate on this kind is solidly high, IQ reads this as a kind you actively want to see. - No pill at all when the pattern is too thin or too mixed to call either way.

• A one-line history summary, for example "6 acted on · 2 dismissed · 75% action rate." The first number is decisions you actually acted on or resolved; the second is decisions you dismissed; the percentage is acted-plus-resolved divided by the total of both, so it is a plain read on how often you engage with this kind versus wave it off.

The controls. Each row has two buttons, Always surface and Always park, plus a Clear button that appears once an override is set. Clicking Always surface tells IQ to keep bringing you this kind of decision no matter what your history says, useful when you genuinely care about a kind but have been too underwater to act on it lately. Clicking Always park tells IQ to stop surfacing this kind even on the weeks you do act on one, useful when a kind is technically actionable but simply is not where your attention should go right now. Clear removes your override and lets IQ's own inference take over again for that row.

A footer note explains the mechanics in plain terms: "How this works. When you keep passing on a kind of decision, IQ stops pushing it and sets it aside instead. Your choices above always win: 'always surface' brings a kind back, 'always park' keeps it quiet even if you act on it sometimes."

Empty state. If you have no decision history with any signal type yet, the section reads: "Nothing here yet. This fills in as you act on, set aside, or pass on decisions over time." That is expected for a new account, it is not a sign anything is broken.

Memory: what I remember about you

What it is. Facts you or your conversations with IQ have established, who runs which crew, which carrier you dropped, what you never want interrupted about, sit in a running memory that IQ reads silently before it drafts a plan, writes an email, or answers you in chat. This tab is the plain-English window into that memory: every fact currently in force, and a form to add, forget, or restore one.

The intro. The tab opens with: "What I remember about you," followed by: "Facts you have told IQ to keep in mind. IQ uses these silently to shape plans, emails, and recommendations. Edit or forget anything here and the next IQ output reflects the change. Nothing here is shared outside your operator account."

Adding a fact

The Add a fact form sits at the top. Fill in:

• Add a fact, a free-text box, capped at 500 characters (the counter below the box tracks you: "127 / 500"). Placeholder text reads "One short sentence IQ should remember."
• Kind, a dropdown with five categories, each with its own helper text shown under the label:

- Person, "Someone on your team, a key contact, or a counterpart IQ should know about." - Preference, "How you prefer IQ to work. Voice, scope, what to ignore, what to surface." - Business fact, "A durable fact about your business IQ should bring back when relevant." - Commitment, "Something you promised to do, or something a counterpart owes you." - Context, "Background that shapes how IQ should read situations."

• Tags (optional), a comma-separated field, up to 8 tags, with the helper "Up to 8, comma-separated. Helps IQ pull this when the topic comes up." (placeholder example: "team, apex, large-loss"). Tags are how IQ matches a fact to a specific job or carrier conversation later, beyond the standing set it always keeps in view.

Click Save fact (it reads "Saving..." while the write is in flight) and the new fact appears immediately at the top of the Active facts list below, no page reload needed.

A fact you add this way is trusted at full confidence, because you are the source, there is no "maybe" about something you typed in yourself.

Active facts

Below the form, Active facts lists everything currently in force, newest first, with a running count ("14 total"). Each row shows:

• The fact itself, in your own words.
• The kind, labeled the same way as the dropdown above (Person, Preference, Business fact, Commitment, Context).
• Tags, if any were set, shown as small pills.
• Where it came from and when, one of: "Added by you [relative time]" for anything you typed in the form yourself, "From chat [relative time]" for a fact you stated in a conversation with IQ, "From email [relative time]," "From voice [relative time]," or "IQ inferred [relative time]" for a fact IQ picked up rather than one you stated directly. Relative time reads as "today," "yesterday," "N days ago," "N weeks ago," "N months ago," or "N years ago."
• A Forget this link, right-aligned, which reads "Forgetting..." while the removal is in flight.

Empty state. With nothing saved yet: "IQ does not remember anything specific about you yet. Add a fact above and IQ will use it the next time it writes on your behalf."

Forgetting a fact

Click Forget this on any active row and it moves out of Active facts immediately. This is a soft delete, the fact is not destroyed, it is marked forgotten and kept for audit, and IQ stops using it in any new plan, email, or answer from that point forward.

Forgotten facts (the audit trail)

If you have forgotten at least one fact, a Show forgotten facts (N) toggle appears below the active list, with the note "Audit trail. Restore one to bring it back." (This entire block is absent if you have never forgotten anything, there is nothing to show.) Expand it and each row shows the fact's text struck through, its kind, and how long ago it was forgotten, plus a Restore button ("Restoring..." while in flight). Restoring puts the fact straight back into Active facts, and IQ starts using it again on the next output.

How Preferences and Memory work together

Preferences shapes what IQ chooses to surface to you. Memory shapes how IQ writes and reasons once something is in front of you. A decision kind you have set to Always park stays quiet regardless of what Memory knows; a fact in Memory ("don't loop me in on sub-$3K supplements") can influence how IQ frames a decision it does still choose to surface. Both are entirely about you: neither tab shows anything about a peer operator, and nothing you enter here leaves your own operator account.

Best-practice example

Say your Preferences list shows "Cert expiring" carrying a Currently parking pill with a summary of "2 acted on · 9 dismissed · 18% action rate." If that is genuinely a kind you do not want to see (you handle certifications on a separate calendar), leave it, IQ has read you correctly. If instead you have been meaning to get to it but have been buried in a large loss, click Always surface so it comes back regardless of the recent pattern. Then, in Memory, add a fact under Preference: "Only escalate cert expirations inside 30 days, I handle the rest on my own calendar." Now Preferences controls whether the kind shows up at all, and Memory shapes how IQ frames it once it does.

Related reading

• The decision workspace
• The Feed
• Acting on decisions
• Connecting your data