Exactly what we can see on your computer
and the report you get

What exactly you get, and when

The agent running on the computer sends a snapshot every 15 minutes. The report you receive is built from the latest snapshot and from every check that has run since the previous report, and it is emailed to you by a technician — not automatically by the machine. The subject line is a plain 'Computer health report' with the machine name and date, and above the list there is a 'Summary and recommendations' block the technician wrote when there is something worth saying.

The report has a fixed structure, meant to be read top to bottom:

• A large health score at the top with a single status word: Healthy / Needs attention / Needs urgent care, and a sentence explaining what the score means.
• Summary and recommendations — a paragraph the technician wrote, when there is one.
• Detailed check report — every check that ran, grouped into six sections. Above it a summary line reads '47 checks ran — 41 healthy, 6 need attention', and right after it the sentence that matters most to the reader: 'Nothing here needs any action from you', or '2 items need action from you'.
• What is taking up disk space — shown only when the disk is nearly full.
• Computer details — operating system, memory load, free space and uptime.

Between reports you have your own customer area with passwordless sign-in (a passkey), showing all your computers, who is online now, each one's score and how many alerts are open. The full explanation of how monitoring works is on the 24/7 computer monitoring page.

The six sections — with a real example from each

The checks are not shown as one flat list. Fifty rows in a row is a wall of text, so they are grouped into six sections in the order a person actually reads: what protects me, what might break, is it fast, is it maintained, is it connected. Each section has its own status line — 'All healthy' or '3 need attention · 11 checks'.

The names were deliberately changed too. The checks are originally named after tools, and on a real report 23 of 30 rows carried an English tool name. A customer does not know what wmic is, so every check has a second, plain-English name: 'Hard-disk health' instead of chkdsk, 'System file integrity' instead of sfc.

🔐 Security and protection

Disk encryption, real-time antivirus, firewall, the prompt before changes to the machine, suspicious sign-in attempts in the last 24 hours, and a layer of hardening checks on top. A real row: 'Protection against the antivirus being switched off — off: malware could disable the antivirus before it acts'. Or: 'Office document macros — all macros enabled: an infected Word or Excel file runs code the moment it opens. This is the most common infection route in email'. Full detail on the computer security audit page.

🖥️ Hardware

Hard-disk wear, free space, memory, battery health, CPU temperature. Example: 'Hard-disk wear — not healthy: the disk reports bad sectors. Back up now and plan a replacement'. This is the single row in the report worth 35 points of the score, because a disk starting to die is data loss. More on the hardware check page.

⚡ Stability and crashes

Blue screens, application crashes, unexpected shutdowns and stability over time. Crashes are counted over a 7-day window but the context is kept: 'No crashes in the last 7 days — the system is stable (there were 3 in 30 days, but none in the last 7)'. That way a machine that crashed three weeks ago and settled does not keep paying for it forever. Full explanation: crash analysis.

🚀 Performance

Memory pressure, programs that load the machine, speed and responsiveness. Rows here are almost always informative rather than a 'fault' — they explain why the computer feels slow, which is exactly the subject of a slow computer.

🔧 Maintenance and updates

Pending updates, system file integrity, device drivers, background services and the system event log. Example: 'System updates — needs attention: 12 updates pending, including security fixes. Installing is recommended'. What happens from there is covered in updates and maintenance.

🌐 Network and internet

Internet connectivity and what is connected to the machine. This is where latency and packet-loss measurements appear, along with what is physically attached — network cards, USB devices and printers. More: network check.

At the end of every row that needs something there is a single ownership line: '← we will handle this' or '← action needed from you'. A report that shows eighteen warnings without saying who does what is not a readable report, only a stressful one.

How the score out of 100 is built — and what it does not mean

The score starts at 100, and each finding subtracts points by the check's weight times the severity of the finding: a severe finding removes 100% of the weight, a needs-attention finding removes 55%, and a healthy or informational check removes nothing. The weights are not arbitrary — they rank real risk:

The colour follows the number: 85 and up is green, 60-84 amber, below 60 red. A sentence always sits next to it, because '71 out of 100' alone does not tell a reader whether that is good or a reason to pick up the phone.

A check that could not run says so — it does not score zero

This was the biggest distortion in the report, and it was fixed. The memory integrity test is the second heaviest check (30 points), but it requires a dedicated restart and so cannot run remotely — and a perfectly healthy computer used to lose 16.5 of 29 points for a check that never ran. Today such a row is written explicitly: 'This check requires a restart and has not run yet — we will run it at the next opportunity. This does not indicate a fault', at weight zero. The same rule applies to a check unsupported on that hardware, such as reading CPU temperature on some Apple machines: it is marked 'checked' and does not hurt the score. An honest silence beats a false 'off'.

What the score does not mean

It is not a comparison to other computers, it is not a promise that nothing will break tomorrow, and it is not a security certificate. It is a weighted summary of what these checks found on this machine today. A computer scoring 100 whose user opens an infected attachment tomorrow will be an infected computer scoring 100 until the next snapshot.

What exactly the agent collects

The agent collects system and configuration state. That is not a vague definition — it is a closed list, and this is the list:

What the agent never collects — explicitly, not by implication

This list is no less important than the previous one, so it is written in the same detail:

• Personal files. Not documents, not spreadsheets, not PDFs, not photos and not videos. Not contents and not file names.
• Email and messages. Not the mailbox, not WhatsApp messages, not chats.
• Browsing. Not browsing history, not sites you visited, not searches, not cookies.
• Passwords. Not passwords saved in the browser and not system passwords. The only check that touches this looks at whether the protection mechanism for saved passwords is active — not what they are.
• Microphone, camera and screen. The system cannot turn on a camera, record audio, capture the screen or log keystrokes.
• Remote control of your desktop. The agent only runs commands from a closed, predefined list (checks, installing an update, starting a service). There is no open 'run anything' exposed to the network.

There is a separate module in the system that does deal with privacy — it answers one question: which apps hold sensitive permissions on the machine (camera, microphone, location, full disk access, 'all sites'). It is built so it can never answer 'who uses the computer' or 'what did they do': only fields from an approved list are stored, and every string is scrubbed of personal identifiers — home-folder paths, email addresses and user IDs — before it is written. It has no field for a user name, a file name, a URL or content.

'The disk is 94% full' is not 'I know what is inside'

This is the distinction everyone actually wants to hear, so we will say it directly. The system sees metrics about the computer, not content inside it. It knows 6% of space is left and that the disk has been running four years — and it has no way to know whether what fills that space is contracts, wedding photos or old backups. Those are two entirely different capabilities, and only the first one exists here.

There is one point where we do go down to folder resolution, and we will state it plainly so it is never a surprise: when a drive drops below 20% free or below 25 GB, once a day the agent measures the size of the folders at the root of that drive, and of the folders one level inside the largest of them. What is kept is the folder name and its size in gigabytes, a row like '74 GB · C:\Users' — not the names of the files inside it, and certainly not their contents. The report also hides hidden system folders and anything under 1 GB, because that is noise. Such a path can include the computer's user name, and that is all it includes.

Who can see your data

Three groups, and that is all:

• You. Through the customer area, signing in with a passkey — your computers only, their scores, the open alerts and the messages the technician sent.
• The technician who handles you. Through the internal dashboard, which requires a login.
• No one else. The data is not sold, not used for advertising and not passed to any third party.

And so this does not stay a promise: every administrator write action is recorded in an audit log — who did it, when, on which machine, from which IP address. Sending a report, running a remote check, requesting system logs from the machine, deleting a device — all of it is logged. That mechanism exists so that even we have no quiet way to do something.

How long is data kept?

Snapshots are kept for 60 days. That is not an arbitrary number: without history there is no trend, and without a trend you can only react to a fault after it has already happened. A disk filling at 1.5% a week looks fine in any single snapshot and very dangerous over six weeks, and the same is true of a slow memory leak or sustained CPU load.

History is not kept at the same density for the whole period: the most recent days are kept in full — every 15 minutes — and older history is thinned to one snapshot per day per machine. The trend is preserved, and old data is not kept forever 'just in case'. System logs pulled from the machine are deleted after 30 days.

Your control over all of it

Four practical points:

1. A restart is a recommendation, never forced. Some checks and updates require a restart. We write it to you in the report on an explicit line and wait — you choose when. No one restarts your computer in the middle of your work.
2. The collection level of the privacy module is adjustable, per customer or per single machine: 'local only' — no privacy data is sent to the server and the scan stays on the machine; 'anonymous report' — only risk scores and numbers are sent, with no app names (this is the default); 'full diagnostics' — app names and their permissions are sent, and still no user names, file names or content. The level is enforced twice: the agent limits itself, and the server checks again on receipt.
3. The agent can be removed completely, with a single command that deletes both the scheduled task and the files from the machine.
4. You can request a report at any moment, not only wait for the next one — and ask what a particular row means. If a sentence in the report is unclear, that is a wording bug on our side, not a gap in your knowledge. Pricing is on the pricing page, and more questions on the FAQ.

FAQ

Read next

• 24/7 computer monitoring — what happens between reports
• Computer security audit — the first section of the report, in depth
• Hardware check — how a dying disk is spotted
• FAQ — anything still open