Activity timeline and analyst provenance

What these two footers are

Every decision workspace ends in two optional footers, stacked below the plan and its outcome. The first is the Activity timeline: a plain event log of everything that has happened to this decision, who did it, and when. The second is the analyst provenance footer, headed "Generated by Verinode Business Analyst": it only appears when this particular decision came out of a Business Analyst scan rather than the always-on signal pipeline, and it shows the framework anchors, data reads, and cost behind that specific finding.

Neither footer changes the decision. Both exist so you can trust what you are looking at: exactly what happened on this plan, and, when it applies, exactly how Verinode's AI Co-COO arrived at this particular call. Verinode does not decide anything on your behalf here either, the footers are a record, not a control.

Where to find them

Open Decisions from the sidebar, or go to iq.verinode.ai/decisions. Click any card to open its workspace as an overlay. Scroll to the bottom, past the plan and the outcome section. If the decision has at least one logged event, you will see a row that reads Activity · N events with a small arrow beside it. Click it to expand the timeline. Below that, if the decision was scan-generated, the provenance footer sits open by default (it does not need a click to expand).

The Activity timeline

What it is. Every meaningful thing that happens to a decision's plan, whether you did it, IQ did it acting on your behalf, the system did it, or a counterparty triggered it, gets written to an event log. This footer reads that log for the decision you have open and lists it newest first.

What you see when you expand it. A heading line repeating Activity with the same event count, a short description ("Every action on this plan, captured in the event log. Filter by who did it."), a row of filter chips, then the list of events itself.

The filter chips

Five chips, each with a small count beside its label:

• All, every event on this plan.
• You, actions you took yourself.
• IQ, actions IQ took acting on your behalf.
• System, background actions the platform itself performed (for example, an automatic re-run).
• Counterparty, actions detected from the other side of the plan, a carrier, adjuster, or vendor moving on their own.

A chip only appears if at least one event of that kind exists (the All chip always shows). Click a chip to narrow the list to just that actor; click All to clear the filter.

What each row shows

Reading left to right:

• When, a relative time ("just now", "12m ago", "3h ago", "2d ago"). Once an event is more than a week old, the row switches to a short date instead (for example "Jul 9"). This updates live while the workspace is open, it is not a frozen snapshot from page load.
• Who, a small colored pill: You, IQ, System, or Counterparty, matching the filter chips.
• What happened, one of: Created, Status changed, Edited, Re-ran, Attached evidence, Marked done, or Undid.
• Which step, the subtask's position and title on the plan (for example "#2 Pull the denied supplements"), so you know exactly what part of the plan the event touches.
• A detail, when there is one worth showing:

- A status change reads "→" followed by the new status (Queued, In progress, Waiting on counterparty, Blocked, Done, Obviated, or Skipped), with a reason appended after a middle dot if one was recorded (for example "→ Blocked · Waiting on adjuster callback"). - A "Marked done" event reads "via" followed by whatever closed it out, but only when that was something other than you or IQ. If you or IQ marked it done, no extra text is needed, the actor pill already says who. - An "Attached evidence" event names what kind of evidence landed: Attested, Forwarded email, Doc uploaded, or Auto-detected.

• Where it came from, a small tag on the far right showing the channel the action was taken through: ui, chat, voice, email, or mirror (an automatic update pulled in from an outside system). This tag is left off for plain background system events, since there is no external origin worth naming for those.

Showing more, collapsing

Only the ten most recent events (after any filter is applied) show by default. If there are more, a Show N more link appears below the list; click it to reveal the rest, up to the most recent 100 events on this plan. Once expanded, the link becomes Collapse, which folds the list back down to ten.

Empty state

If a decision genuinely has zero events logged (which, given the gating above, effectively never surfaces inside the workspace since the Activity row itself only appears once there is at least one event), the footer's own built-in message reads:

No actions logged on this decision yet.

Analyst provenance footer

What it is. When a decision was written by a Business Analyst scan, its evidence carries the scan run's identity, the standards it checked the finding against, the internal reads it grounded the finding in, and what that run cost to produce. This footer surfaces all of that, so a scan-sourced decision is never a black box.

This footer only renders for decisions that came from a scan. Decisions from the always-on signal pipeline, forwarded documents, or manual entries do not carry it, there is no run to attribute them to.

The header line

"Generated by Verinode Business Analyst", with a relative timestamp beside it ("2h ago", "3d ago", or, once it is more than a week old, a full date). If the scan was the automatic weekly pass, the timestamp is followed by "· weekly sweep". If you started it yourself from Run analysis, it reads "· you ran this" instead. Directly under the header:

"Verinode read your business across every section, applied LEAN + restoration-framework expertise, and surfaced this finding with the framework anchors below. Hover any citation for the rationale."

Framework citations

The standards the analyst checked this specific finding against: LEAN, IICRC, OSHA, EPA, State regulation, or Carrier program. Each citation lists the standard, a short code, and, underneath, the rationale explaining why that citation applies to this finding.

Empty state. If the analyst did not attach any citations to this finding, the section reads, in italics:

No framework citations were attached to this finding.

Grounded in (tools called)

A row of small pill labels under the heading Grounded in, one per internal data read the analyst pulled before writing this finding up, humanized into plain language (for example, a margin-status read shows as "Margin", a compound-focus read shows as "Compound Focus"). This whole section is left out entirely if no reads are recorded for the finding, rather than showing an empty label.

Cost telemetry

A small line of text: the input and output token counts for the model call that produced this finding, the dollar cost of that call (to the cent), and which model ran it. This only appears when the underlying run row still exists and has a recorded cost; if the run was never billed out (for example, it is still in progress) or the run row itself has since been deleted, this line is simply omitted along with the timestamp and sweep/manual tag in the header above it.

Dismiss all open findings from this run

A single link at the bottom: Dismiss all open findings from this run. Click it to bulk-clear every decision this same scan produced that is still open (still Pending, in other words), in one move, without hunting each one down individually on the log. Decisions from this run that you have already acted on, parked, or dismissed one at a time are left exactly as they are, the bulk action only touches the ones still waiting on you.

While it runs, the link reads Dismissing… and cannot be clicked again. When it finishes, a toast confirms the result:

• "Dismissed N finding(s) from this run." if it cleared any.
• "No open findings remain from this run." if there was nothing left to clear (you had already worked through them all).
• An error toast, "Couldn't dismiss the run." (or a more specific message), if the action failed.

The decision log refreshes automatically once the toast lands, so any decisions you just dismissed drop out of your open lanes immediately.

How to use these together

The Activity timeline answers "what happened and who did it" for any decision, scan-sourced or not. The provenance footer answers "why does Verinode believe this" for the subset of decisions that came from a scan. Read Activity first if you are trying to reconstruct a plan's history, whether IQ re-ran a step, whether a counterparty moved on their own, whether evidence has already been attached. Read the provenance footer when you want to check a scan finding's grounding before you act on it, especially one carrying a real dollar figure, so you can see the standard it is anchored to and what data actually fed it, not just take the number on faith.

1. 1Open a decision from /decisions and scroll to the bottom of its workspace.
2. 2If an Activity · N events row is present, click it to see the full trail: who touched this plan, what they did, and when.
3. 3Use the You / IQ / System / Counterparty chips to isolate one actor's actions, useful when you want to answer "did IQ actually re-run this?" or "did the carrier respond?"
4. 4If the workspace also shows "Generated by Verinode Business Analyst," check the framework citations and the Grounded in pills before you act, they tell you exactly what standard and what data produced this finding.
5. 5If a whole run turned out to be noise, use Dismiss all open findings from this run rather than clearing each decision one at a time.

Best-practice example

Say a vendor overcharge decision has been sitting Acted for two weeks and you want to know what has actually moved. Open it, expand Activity, and filter to IQ: you see IQ re-ran the vendor comparison once and attached a forwarded invoice as evidence three days ago. Filter to Counterparty and you see nothing, the vendor has not responded. That tells you the ball is still in your court to follow up, not Verinode's. Separately, if this same decision came from a scan, the provenance footer below shows it grounded in a "Vendor Pricing" read and cited against LEAN, giving you the standard behind the dollar figure before you push it further.

Related reading

• The decision workspace for everything else in the workspace these footers sit at the bottom of.
• Running a Business Analyst scan for how a scan runs, what it costs, and how its findings land on your board.
• Acting on decisions for what generates the plan steps this timeline logs.
• The Decisions section: overview for the log, lanes, and workspace this footer belongs to.
• Forwarding documents for how a forwarded email becomes an "Attached evidence" event on a plan.

Data sources