Extends ExistingPhase 3 — IntelligenceWF-08

Report Dashboard

A dashboard that sits there waiting for someone to look at it is not intelligence — it is a filing cabinet. This one connects every system, feeds live data automatically, and sends the warning before the problem becomes visible.

The Gap I Found

The dashboard existed.
It just waited for people to find the problems.

The business had a dashboard with the right metrics — AR balance, gross margin, utilisation, pipeline value, 90-day cash forecast. Leadership opened it when they needed to make a decision. And every time they did, the numbers were behind.

Sometimes by hours, sometimes by days. Because the dashboard updated when someone refreshed it — not when the underlying data changed. In a business where people are busy, that refresh step gets skipped constantly. The problem wasn't the layout or the metrics. The problem was that the dashboard was passive. It held the data. It didn't watch it.

A traditional dashboard sits and waits for a human to look at it, notice something is wrong, and decide to act. By the time that loop completes, the golden window to fix the issue has often already closed. The over-commitment has already been made. The cash has already left. The client has already pushed back on the invoice.

Why It Happens

No push events. No analysis layer.
No one watching the numbers.

Financial DB, Capacity View, CRM pipeline, and People DB each hold a piece of the picture — but each was built independently. No push event connected any of them to the dashboard, so data aged silently in the source until someone pulled it manually.

And even when the dashboard was current, the monitoring was entirely human. Someone had to open it, scan the numbers, recognise that something looked off, and escalate. That chain depends on attention and availability at exactly the right moment — which is precisely when it is least reliable.

What I Designed

Two layers: live data in,
AI analysis out.

The first layer connects all four source systems through defined push events. Financial DB fires on every write. Capacity View on refresh. CRM on stage change. People DB on payroll approval. The dashboard stops being a snapshot and becomes a live view — always current, no manual refresh required.

The second layer is where it separates from any traditional dashboard. An AI layer scans every incoming data point against defined thresholds. When a number drops below target — AR balance, gross margin, utilisation, pipeline value — the system doesn't wait for someone to notice. It routes an alert to Slack, email, or both, depending on the severity and the stakeholder. On a fixed schedule, it also sends a weekly digest summarising the state of the operation — whether or not anything is wrong.

The dashboard went from passive to active. It no longer needs a person to look at it to be useful. It tells you when you need to look.

Layer 1 — Live feed architecture

Financial DB
on writeAR balance · gross margin · cash forecast
Capacity View
on refreshUtilisation %
CRM
on stage changePipeline value
People DB
on payroll approvalCash outflow · labour cost
Report Dashboard— always current

Layer 2 — AI watchdog + alert routing

AR balance drops below 30-day thresholdSlack + Emailcash risk
Utilisation hits 90% capacitySlackover-commitment
Pipeline drops below monthly targetEmailrevenue gap
Gross margin falls below set floorSlack + Emailmargin erosion
On schedule: weekly digest sent automatically to stakeholders — whether or not anything is wrong

Before

Data changes in source systems
Dashboard sits unchanged — nobody notices
Someone opens the dashboard to check numbers
Numbers are hours or days old
Decision made on stale data
Problem surfaces after the fact
Data age: hours to days · Warning system: none · Discovery: too late

After

Data changes in source systems
Dashboard updates automatically — live
AI scans every update for threshold breaches
Warning fires to Slack and/or email immediately
Stakeholder acts while there is still time to fix it
Weekly digest lands in inbox on schedule
Data age: seconds · Warning system: always on · Discovery: before it breaks

Systems Involved

Report Dashboard (Notion / Airtable / custom)Financial DBCapacity ViewCRM (Notion / Airtable / HubSpot)People DBAI layer (Claude / GPT)SlackEmail (Gmail / Outlook)Automation (Make / Zapier / n8n)

What Changed

The operation is visible.
And it speaks up when something is wrong.

Live

Dashboard, always

Every write event in any source system triggers an update. Leadership sees the current state of the operation — not the state it was in the last time someone opened a tab and hit refresh.

AI watching every number

The AI layer scans each incoming data point against defined thresholds. When a number drops below target — AR, margin, utilisation, pipeline — it does not wait for someone to notice. It sends the alert.

0

Manual reports required

The weekly digest goes out on schedule automatically. Stakeholders get the summary in their inbox whether or not they remember to look at the dashboard — and whether or not anything is wrong.

Golden time to act

The gap between a problem appearing and a stakeholder knowing about it used to be measured in days. Now it's measured in minutes. That gap is where the damage gets done — or gets prevented.

The human experience changed on two levels. First: the dashboard became trustworthy. Before, opening it meant a quiet mental check — when did someone last refresh this? That question no longer exists. Every metric reflects the current state of the operation.

Second, and more importantly: stakeholders stopped having to go looking for problems. The system now finds them first. A Slack message arrives before the over-commitment is locked in. An email lands before the cash shortfall becomes a crisis. The golden time to act — the window between when a problem appears and when it becomes irreversible — is now the window the team actually operates in.

How It Connects

The terminal point for every
workflow — and the earliest warning system.

Every other workflow in this system feeds the dashboard downstream. The quality of what it shows — and what the AI watchdog can detect — depends entirely on how accurately each upstream workflow writes to its source. A clean pipeline here means the alerts are trustworthy, not noisy. A broken upstream feed means the warning system has nothing to watch.

Collections & ARPayment events update the AR balance live — and if it drops below threshold, the alert fires. Cash risk surfaces in minutes, not at the next manual check.
Time Approval & PayrollApproved payroll posts as cash outflow. Locked, rate-applied time entries flow into gross margin. The AI watchdog monitors both — an eroded margin triggers a warning before the next invoice goes out.
Project ManagementTeam assignments feed the utilisation view. When capacity approaches the ceiling, the system alerts before over-commitment is made — not after the proposal has already been sent.
Email TriageEvery confirmed lead writes a CRM record that feeds pipeline value. If pipeline drops below monthly target, stakeholders know immediately — with enough time to act on it.

Linh Pham

Operations Analyst

Ready to look at your operation?

pdlinh.vyco@gmail.com

© 2025 Linh Pham · Based in Canada · Remote Worldwide