Case studies

01 · Windows diagnostic decision support

System Guard AI

Translating Windows system signals into plain-English guidance without changing the machine.

Product question

How can PC diagnostics help a non-technical user decide what to check next—without taking control away?

Company context
Founder, Point Zero AI
My role
Product direction, diagnostic UX, validation and implementation review
Current status
v1.8.5 · Closed-test build
Development approach
AI-supported implementation under founder-directed requirements, testing and validation.
System Guard AI dashboard showing system status and a prioritised insight
The closed-test dashboard combines local system status with a prioritised insight and a safe next step.

01 · Problem

The problem at the centre

Windows exposes useful performance and stability signals, but interpreting them often requires technical knowledge. A user facing a slowdown, crash or black screen needs a sensible first step—not another wall of telemetry.

How can PC diagnostics help a non-technical user decide what to check next—without taking control away?
Target user

Windows users who need calm, understandable guidance before deciding whether to investigate further or ask for technical help.

Product direction

System Guard AI reads local diagnostic signals, identifies the most relevant issue, explains why it matters, shows supporting evidence and suggests safe manual checks. It never applies a fix on the user’s behalf.

Primary constraint

Read-only: no file deletion, process termination, setting changes, driver changes or automatic repair.

02 · Product Direction

A product boundary before a feature list

System Guard AI reads local diagnostic signals, identifies the most relevant issue, explains why it matters, shows supporting evidence and suggests safe manual checks. It never applies a fix on the user’s behalf.

Implemented

System snapshot

Reads CPU, RAM, disk, running-process and available hardware signals.

Tested

Slowdown diagnosis

Prioritises likely causes and identifies the most useful next check.

Tested

Plain-English reports

Structures findings as the main insight, why it matters, next step and supporting evidence.

Implemented

Guided Fix + re-check

Provides issue-specific manual guidance and compares what changed after the user acts.

Implemented

Monitoring

Supports live status updates and a user-triggered two-minute diagnostic session.

Implemented

Local history

Optionally stores baselines, saved reports, settings and a troubleshooting timeline on the device.

03 · Process and Contribution

How I directed the work

  1. 01

    Identified the trust problem around system utilities and set a read-only product boundary.

  2. 02

    Defined the diagnostic hierarchy: priority issue, explanation, evidence, action and re-check.

  3. 03

    Directed the desktop and CLI experience, including beginner and advanced reporting paths.

  4. 04

    Directed implementation and code review while retaining product direction, acceptance decisions and final verification.

  5. 05

    Reviewed the focused product test suite and validated the packaged Windows closed-test executable.

Implementation Approach

I used ChatGPT- and Codex-supported workflows to explore implementation, review code paths, investigate bugs and iterate on the desktop and CLI experience. I set the feature priorities and acceptance decisions.

Testing and Validation

The current 440-test suite checks diagnostic behaviour and safety boundaries. Early closed-test feedback and critical review helped move the experience beyond a raw diagnostic display toward explanation, Guided Fix, re-checks and monitoring.

Process Growth

The product evolved from presenting system data to helping non-technical Windows users understand what matters, why it matters and what they can safely do next.

04 · Testing and Validation

What supports the current status

Verified from the current local build and test suite.

440

automated tests passed

Executed locally against the authoritative source for this release.

1.8.5

current version

Confirmed in product version metadata.

Standalone

Windows build

Packaged as a Windows executable for closed testing.

Current status

A functioning closed-test diagnostic product with a tested, explicit safety boundary. The current outcome is product readiness for closed testing—not automated repair or production-scale adoption.

06 · System

How the system supports the experience

Simplified from current source structure and architecture records. Sensitive implementation detail is intentionally excluded.

System Guard AI architecture diagram
Source-backed architecture: local collection, diagnostic analysis, evidence-led explanation and user-controlled action.

07 · Key Decisions

Boundaries that shaped the product

  • Read-only: no file deletion, process termination, setting changes, driver changes or automatic repair.
  • Local-only: no system-data upload and no access to documents, photos, emails, passwords or browser history.
  • Diagnostic guidance is non-definitive and does not replace antivirus, Windows Security or professional repair.
  • The Windows executable is unsigned and may trigger SmartScreen.

08 · Lessons Learned

What I will carry forward

01

Trust is a product requirement. The most important feature is often a clear boundary around what the software will not do.

02

A diagnostic score is only useful when it leads to understandable evidence and a proportionate next step.

03

Progressive disclosure lets beginners act without preventing technical users from inspecting the detail.

Technology

Tools used in the build

Technologies used in the current implementation.

PythonCustomTkinterpsutilWindows event signalsLocal JSON storagePyInstallerpytest
Next case studyPlateProfitAI