Before Pulsar7
Before the plan-vs-actual system we worked the way everyone does. Morning call, the whole team. Each person, in turn, reports what they did yesterday and what they are going to do today. An hour goes by just so that twenty-six people can check in. Actual management value: zero. Someone says "everything is fine, revenue is up". Someone else cannot even remember their own numbers.
The fallback option was Excel. You make a shared sheet, hand out access, ask people to fill it in every day. Two weeks later the formulas have drifted, the charts do not render, half the columns are empty. A month in, no one fills it any more. Familiar story.
The core problem is that a plan-vs-actual system without a live product is a one-way game. No one wants to fill it in because no one looks at the numbers. No one looks at the numbers because they are perpetually out of date. Closed loop.
How we use Pulsar7 today
It works differently now. Every morning at 10:00 the standup runs for five to ten minutes. We open the "Standup" view, look only at the red metrics. Greens get a silent nod. Yellows we mention briefly. For reds we decide what to do, in the room.
The numbers are already in by the time the standup starts. Each employee logs their actuals during the day in the "Dashboard" view. It takes two minutes: glance at the plan, type what got done, set the status. Done. When the team opens the app in the morning, the numbers are ready.
The team adopted the ritual on its own. No reminders. They saw their own metric in the red zone and wanted to move it to green. Gamification for free, without points or badges.
Financial figures live in a separate tool, FP: cash flow, P&L, payroll. They only surface in Pulsar7 if there is an anomaly (cash gap, drop in a client's revenue).
Albert: the AI director inside the product
Albert is the AI assistant built into Pulsar7. Every morning it looks at the numbers across all departments and writes a short brief. Not "everything is fine, keep going". Specifically: "Production has been down for four days. The bottleneck is contractor downtime. Action for today: call three subcontractors."
The most useful thing about Albert is that it spots what we miss. One morning it flagged that one of our key clients was quietly sliding. The metric had been drifting for two weeks in small steps, hard to see by eye. Albert plotted the trend and warned us. We had a conversation with the client before they left.
Albert runs on GPT-4o-mini under the hood, with a prompt tuned to read like a manager. It does not write "consider paying attention to". It writes "call Smith today, before it is too late". This matters; otherwise no one reads AI advice.
Four sections where all the work lives
Dashboard
The full table across every department and every metric. Marketing, sales, production, HR, finance — all visible at once. Filter by day, week, month. Each cell carries a colour dot (a traffic light: green, yellow, red). Open it in the morning, and in ten seconds you know what is fine, what is yellow, what is red.
Standup
The meeting view. Same departments, but only the problematic metrics. Green is hidden. Only the things that need talking about are on the screen. Next to each red metric, Albert's take. That is the whole standup script.
Charts
Six-month trends. Day-over-day, week-over-week, year-over-year. If a metric grew 20% over the month, you see one chart. If it sagged since last quarter, you see another. The naked eye does not catch trends like that without charts.
Decisions
A log of management decisions made at standups. Every meeting we decide something: launch a campaign, hire a manager, drop a vendor. All decisions land here with a status and an owner. Two weeks later you open it and check who did what. Without this, decisions get lost in chat and forgotten.
What changed in agency operations
First: standups got short. Five to ten minutes instead of an hour. The freed-up time goes into work, not talking. Fifty minutes × 26 people × 20 working days = 433 hours a month. That is a full-time employee we bought back from ourselves.
Second: transparency went up. Numbers in one place, visible to everyone. No one walks into a meeting with their own spreadsheet and a different version of revenue. Single source of truth.
Third: part of the micro-management disappeared. When an employee sees their metric and the plan, they pull themselves up. There is no need to ask "how is conversion going" when conversion is right there on both screens.
Fourth: Albert catches the small stuff. Slow-drifting trends we used to miss now surface. That is money we would have left on the table.
Who needs a system like this
- Teams of 10+ with 3 or more departments
- Agencies with many clients and monthly plan numbers
- Operations companies where one department's downtime breaks others
- Anyone whose Excel sheets keep breaking and not being filled
- Managers who spend 2+ hours a day on operational status
- Teams under 5, where everyone already knows what is going on
- If you do not know what your KPIs are in the first place
- Businesses without recurring metrics (one-off projects)
- If the leader is not willing to read numbers every day
- Anyone hoping it will "just work" without setup
How to set up something similar
If you want a system like this for your business, there are two paths.
Option 1. Join Pulsar7. The product is in closed beta, invite-only, free for now. Works if your business resembles an agency in structure and your metrics are standard. Sign up at pulsar7.ru.
Option 2. Commission your own plan-vs-actual SaaS, built around your business: your departments, your metrics, your formulas. If your case is unusual (a specific product, non-standard KPIs), the off-the-shelf Pulsar will not fit perfectly and a custom build is the right answer.
At AVAT, a project like this takes 5 working days at a fixed price. We went through the pain ourselves building Pulsar7, so we know the gotchas. See SaaS pricing or book a brief call.
The takeaway
A plan-vs-actual system only works when the team sees the numbers. Without visibility, no one fills it; without filling, no one trusts it.
The standup collapses to red metrics. Five minutes is enough when you only talk about what is broken.
An AI assistant that writes like a manager (direct, action-oriented) gets read. One that writes like a consultant gets ignored.
If your business is standard, off-the-shelf works. If it is unusual, a custom SaaS built in 5 days pays for itself in months.