PC Workman 1.6.8 - 80 Downloads, 800 Hours, and Why I'm Building for Microsoft Store

The Number That Changed Everything Eighteen downloads. On paper, that's nothing. Apps on the Microsoft Store get thousands in their first week. Open source projects on GitHub hit triple digits on launch day. But for PC_Workman - a system monitoring tool I built alone over eight months. 80 downloads meant something different. It meant 60 people: Saw a project from an unknown developerTook a chance on software that could access their CPU, GPU, and RAM dataTrusted code written by someone who learned PyQt5 while the app was being built Saw a project from an unknown developer Took a chance on software that could access their CPU, GPU, and RAM data Trusted code written by someone who learned PyQt5 while the app was being built That's not "just 60 downloads." That's 60 signals of trust. And trust is exactly what's broken in system monitoring software today. The Problem: Monitors That Show, Don't Explain I started building PC_Workman because existing tools frustrated me. MSI Afterburner shows "CPU: 87%" - but doesn't explain why. MSI Afterburner why HWMonitor logs temperature spikes, but doesn't help you prevent them. HWMonitor prevent Task Manager tells you what's consuming resources right now, but has no memory of what happened five minutes ago when your PC froze. Task Manager right now, The pattern: these tools are reactive, not proactive. reactive proactive They answer "what's happening?" but never "why is this happening?" or "how do I fix it?" I needed something different. Not just for monitoring, for understanding. understanding So I built it. Eight Months, Four Rebuilds, One Goal June 2025: Started with spaghetti code and enthusiasm. June 2025: September 2025: First rebuild. Realized I didn't understand threading. September 2025: November 2025: Second rebuild. CSV files hit 500MB, app lagged. November 2025: December 2025: Third rebuild. Lost warehouse job three days before Christmas. Restarted that night. December 2025: January 2026: Fourth rebuild. Shipped v1.6.3 - the first actual release. January 2026: February 2026: Fourth rebuild. Shipped for testers v1.6.8 - the first actual release. February 2026: March 2026: Roadmap is published for milestone 1.7.8. (And 2.0, Microsoft Store Publish goal) March 2026: 800+ hours between those dates. Not because I'm slow. Because I was learning while building. while Technical Evolution: CSV -> SQLite Technical Evolution: CSV -> SQLite The biggest architectural change: migrating from CSV logging to SQLite. Why it mattered: Why it mattered: In v1.5.x, PC_Workman wrote system stats to CSV files every minute: timestamp,cpu,gpu,ram 2025-12-01 14:23:45,34.2,67.8,45.1 2025-12-01 14:24:45,36.1,68.2,44.9 timestamp,cpu,gpu,ram 2025-12-01 14:23:45,34.2,67.8,45.1 2025-12-01 14:24:45,36.1,68.2,44.9 After a week: 10,080 rows. After a month: 43,200 rows. After three months: CSV files hitting 500MB. Reading that data to display historical charts? 3-5 second lag. 3-5 second lag. Users don't wait 5 seconds for a graph. Solution: Stats Engine v2 Solution: Stats Engine v2 Built a SQLite pipeline with tiered aggregation: Minute-level data: Kept for 7 daysHourly aggregates: Kept for 90 daysDaily summaries: Kept forever Minute-level data: Kept for 7 days Minute-level data: Hourly aggregates: Kept for 90 days Hourly aggregates: Daily summaries: Kept forever Daily summaries: Database size after a month: 15MB instead of 500MB. Chart load time: Under 200ms. Under 200ms. But the real win wasn't performance, it was intelligence. intelligence From Data to Insights From Data to Insights With SQLite came pattern detection. The engine now tracks: Which processes run at which times (recurring patterns over 7 days)CPU/GPU spikes correlated with specific applicationsTemperature trends during gaming vs. idleRAM usage patterns per application per day Which processes run at which times (recurring patterns over 7 days) Which processes CPU/GPU spikes correlated with specific applications CPU/GPU Temperature trends during gaming vs. idle Temperature RAM usage patterns per application per day RAM Example: PC_Workman noticed I always launched Battlefield around 8 PM. Started pre-optimizing system resources at 7:55 PM. Example PC_Workman noticed I always launched Battlefield around 8 PM. Started pre-optimizing system resources at 7:55 PM. That's not monitoring. That's learning. That's not monitoring. That's learning. What 60 Downloads Taught Me January 19, 2026: Released v1.6.3 with the first .exe installer. Download curve: Download curve: Day 1: 25 downloads (launch spike)Week 1: Average 10/dayMonth 1: Average 5-7/dayMonth 2: Settled at 1-2/day (organic) Day 1: 25 downloads (launch spike) Day 1: Week 1: Average 10/day Week 1: Month 1: Average 5-7/day Month 1: Month 2: Settled at 1-2/day (organic) Month 2: Total by mid-March: 60+ downloads. Here's what the numbers showed: 1. Launch Momentum Fades Fast 1. Launch Momentum Fades Fast 25 downloads on day one felt like success. Then 10. Then 5. Then 2. Pattern: Launch spike -> exponential decay -> organic plateau. Launch spike -> exponential decay -> organic plateau. Reality check: Initial excitement doesn't sustain growth. You need continuous discovery. 2. 40% Were Repeat Visitors 2. 40% Were Repeat Visitors Traffic analysis revealed something unexpected: Four out of ten downloads came from people who: Saw the initial postClicked to GitHubLeftCame back days or weeks laterThen downloaded Saw the initial post Clicked to GitHub Left Came back days or weeks later Then downloaded Then Lesson: People don't need a perfect product. They need to see you're serious about building it. Lesson: The downloads that came back did so because: GitHub showed consistent commits (activity is seriousness!)Release notes were detailed (transparency in build in public is trust)Roadmap was public GitHub showed consistent commits (activity is seriousness!) Release notes were detailed (transparency in build in public is trust) build in public Roadmap was public 3. Build-in-Public Drove 40% of Traffic 3. Build-in-Public Drove 40% of Traffic Tracked referral sources: LinkedIn posts: 40%Sourceforge listing: 20%HackerNoon: 25%Direct GitHub: 5%Other: 10% LinkedIn posts: 40% LinkedIn posts: Sourceforge listing: 20% Sourceforge listing: HackerNoon: 25% HackerNoon: Direct GitHub: 5% Direct GitHub: Other: 10% Other: LinkedIn worked because I shared: Technical problems and solutionsActual code snippetsHonest struggles (bugs, rebuilds, failures)Incremental progress (commits, not just releases) Technical problems and solutions Actual code snippets Honest struggles (bugs, rebuilds, failures) Incremental progress (commits, not just releases) The worst-performing content? Generic announcements. "Released v1.6.3!" - 50 views. "Here's why I rebuilt the CSV logging system" + code screenshots - 600 views. 4. Trust Gaps Are Real 4. Trust Gaps Are Real Every download required the user to: Bypass SmartScreen: "Windows protected your PC"Grant system access: CPU/GPU monitoring needs permissionsTrust data privacy: No policy page, just MIT licenseBelieve it won't break things: No company backing, no reviews Bypass SmartScreen: "Windows protected your PC" Bypass SmartScreen: Grant system access: CPU/GPU monitoring needs permissions Grant system access: Trust data privacy: No policy page, just MIT license Trust data privacy: Believe it won't break things: No company backing, no reviews Believe it won't break things: 60 people cleared all four hurdles. That's not trivial. And it's exactly why Microsoft Store matters more than download counts. Why 80 Downloads Isn't the Goal, Microsoft Store Is Here's what changed my perspective: I'm not chasing 100 downloads, or 1,000, or viral GitHub stars. I'm chasing legitimacy. legitimacy And on Windows, legitimacy comes from one place: Microsoft Store. Microsoft Store. The Competition Gap The Competition Gap Check the Microsoft Store for "system monitoring." You'll find: Basic CPU metersTemperature widgetsGeneric task managers Basic CPU meters Temperature widgets Generic task managers You won't find: MSI AfterburnerHWMonitorCore TempGPU-ZAny serious monitoring tool MSI Afterburner HWMonitor Core Temp GPU-Z Any serious monitoring tool They're all distributed via: Direct downloads (executable files)Third-party sites (SourceForge, CNET)GitHub releases Direct downloads (executable files) Third-party sites (SourceForge, CNET) GitHub releases Why does this matter? Why does this matter? Because 95% of Windows users will never download an .exe from a random GitHub repo. They will click "Get" on the Microsoft Store. will Microsoft Store: Not Distribution, Validation Getting PC_Workman on the Microsoft Store isn't about reaching more users. It's about removing friction between software and trust. removing friction between software and trust. Here's what most people don't understand about desktop software distribution in 2026: When you download an .exe from GitHub: When you download an .exe from GitHub: Windows shows: "Windows protected your PC"You click "More info" -> "Run anyway"Antivirus scans it suspiciouslyNo update mechanism (you manually check for new versions)No accountability (developer could disappear tomorrow) Windows shows: "Windows protected your PC" You click "More info" -> "Run anyway" Antivirus scans it suspiciously No update mechanism (you manually check for new versions) No accountability (developer could disappear tomorrow) When you install from Microsoft Store: When you install from Microsoft Store: One click: "Get"Automatic updatesMicrosoft has verified the publisherSandboxed (can't access system files without permission)Reviews visible (social proof)Uninstall is clean (no registry residue) One click: "Get" Automatic updates Microsoft has verified the publisher Sandboxed (can't access system files without permission) Reviews visible (social proof) Uninstall is clean (no registry residue) For a solo developer with zero brand recognition, that difference is everything. everything. The Technical Requirements (And Why They Matter) The Technical Requirements (And Why They Matter) Publishing to Microsoft Store isn't just uploading a file. Microsoft enforces security standards that prove you're legitimate. 1. Code Signing Certificate ($200-400/year) 1. Code Signing Certificate ($200-400/year) Every executable must be digitally signed with a verified certificate. What this proves: What this proves: You're a real person/company (identity verified)The code hasn't been tampered with since signingYou're accountable (certificate ties to legal identity) You're a real person/company (identity verified) The code hasn't been tampered with since signing You're accountable (certificate ties to legal identity) Why it matters for users: Why it matters for users: No SmartScreen warningClear "Publisher: Marcin Firmuga" instead of "Unknown Publisher"Trust signal: "This person stands behind this code" No SmartScreen warning Clear "Publisher: Marcin Firmuga" instead of "Unknown Publisher" Trust signal: "This person stands behind this code" Cost: €200-400/year depending on certificate authority. For a free app with zero revenue, that's significant. But it's the price of legitimacy. 2. Privacy Policy (Required) 2. Privacy Policy (Required) Microsoft requires a public privacy policy before Store approval. For PC_Workman, this is actually good: For PC_Workman, this is actually good: Current state: No telemetryNo cloud syncNo user trackingAll data stored locally in /data/ No telemetry No cloud sync No user tracking All data stored locally in /data/ /data/ Privacy policy makes this explicit instead of assumed. explicit Users can read: "We don't collect anything. Your data never leaves your PC." That's not a burden. That's a competitive advantage. Competitors: Competitors: Many system monitors phone home (crash reports, usage stats)Some sell "anonymized" hardware dataPrivacy policies buried or vague Many system monitors phone home (crash reports, usage stats) Some sell "anonymized" hardware data Privacy policies buried or vague PC_Workman: PC_Workman: "We monitor your PC. We don't monitor you."One-sentence privacy policy possibleOpen source (code is auditable proof) "We monitor your PC. We don't monitor you." One-sentence privacy policy possible Open source (code is auditable proof) 3. MSIX Packaging (Sandboxing Challenge) 3. MSIX Packaging (Sandboxing Challenge) Store apps use MSIX format instead of traditional .exe installers. Key difference: Sandboxing. Key difference: Sandboxing. Traditional .exe: Can access any file on systemCan modify registry freelyCan run background servicesFull system permissions Can access any file on system Can modify registry freely Can run background services Full system permissions MSIX (Store app): MSIX (Store app): Limited file access (app folder + specific directories)Registry changes are virtualizedBackground tasks require explicit permissionRuns in AppContainer Limited file access (app folder + specific directories) Registry changes are virtualized Background tasks require explicit permission Runs in AppContainer Challenge for PC_Workman: Challenge for PC_Workman: Some features need system-level access: Reading all process CPU/RAM usageMonitoring GPU temperaturesControlling fan speedsManaging Windows services (TURBO BOOST mode) Reading all process CPU/RAM usage Monitoring GPU temperatures Controlling fan speeds Managing Windows services (TURBO BOOST mode) Solution: Solution: Request specific capabilities in app manifest: xml Microsoft reviews these requests. You must justify why each capability is necessary. Why this is good: Why this is good: Users see exactly what permissions the app needs. Not hidden in an installer. Not assumed. Explicit and reviewable. Explicit and reviewable. 4. Store Review Process (1-3 Weeks) 4. Store Review Process (1-3 Weeks) Microsoft reviews every submission: Automated security scanManual testing by reviewersPolicy compliance checkPrivacy policy verification Automated security scan Manual testing by reviewers Policy compliance check Privacy policy verification Possible rejection reasons: Possible rejection reasons: Crashes during testingAccessing more system resources than justifiedUnclear privacy policyIncomplete functionality Crashes during testing Accessing more system resources than justified Unclear privacy policy Incomplete functionality Benefit: Benefit: If Microsoft approves PC_Workman, it's a third-party validation that the software: third-party validation Does what it claimsDoesn't have obvious security issuesMeets minimum quality standards Does what it claims Doesn't have obvious security issues Meets minimum quality standards That's worth more than 1,000 downloads from random users. The Path to v2.0: Building for the Store Current state: v1.6.8 (February 2026) Target: v2.0 on Microsoft Store (Q3 2026) Between here and there: v1.7.9 roadmap. v1.7.9 roadmap. 16 Features in Research/Development 16 Features in Research/Development Not building randomly. Researching what actually works. Studied optimization tools: Advanced SystemCare, Razer Cortex, CCleaner, Wise Game Booster. Extracted 16 common features. Testing which deliver real performance gains. TURBO BOOST Mode (5 core features confirmed): TURBO BOOST Mode (5 core features confirmed): Feature 1. Auto RAM Optimizer Feature 1. Auto RAM Optimizer Trigger: RAM usage >75% for 30+ seconds Action: Clear standby memory (cached file data) Expected gain: 300-800 MB freed Status: Tested, works consistently Feature 2. Background Process Sleeper Feature 2. Background Process Sleeper Trigger: Process idle >5 minutes (zero CPU, no focus) Action: Suspend (not kill) via NtSuspendProcess Expected gain: 10-20% CPU reduction Status: Testing on different process types Feature 3. Non-Essential Service Killer Feature 3. Non-Essential Service Killer Target: Print Spooler, Windows Search, Xbox services (20-30 services) Action: Stop temporarily, restore on exit Expected gain: Noticeable on HDDs, moderate on SSDs Status: Safe service list finalized Feature 4. CPU Core Unparking Feature 4. CPU Core Unparking Problem: Windows parks cores to save power Solution: Unpark all cores, boost game priority Expected gain: 5-15% FPS on mid-tier CPUs Status: Registry tweak working, needs admin elevation Feature 5. High Performance Power Plan Feature 5. High Performance Power Plan Trigger: Fullscreen game detected Action: Auto-switch to High Performance plan Expected gain: 10-20% overall performance Status: Detection logic tested with 15+ games 11 more features in testing pipeline. 11 more features in testing pipeline. Not shipping until proven. No placebo toggles. Auto-Update System Auto-Update System Current problem: Users on v1.6.3 don't know v1.6.8 exists. Solution: Check GitHub releases API on startupShow dismissible banner if newer version availableClick -> opens release pageUser decides: update now or later Check GitHub releases API on startup Show dismissible banner if newer version available Click -> opens release page User decides: update now or later Not doing: Auto-download (security risk)Forced updates (annoying)Background updates (Store handles this anyway) Auto-download (security risk) Auto-download (security risk) Forced updates (annoying) Forced updates (annoying) Background updates (Store handles this anyway) Background updates (Store handles this anyway) This is a temporary bridge until Store version ships. temporary bridge Code Quality Pass Code Quality Pass v1.7.x focus: Clean, maintainable, auditable code. Why this matters for Store submission: Reviewers may examine source (it's public)Crashes during review = rejectionSloppy code = perception of sloppy security Reviewers may examine source (it's public) Crashes during review = rejection Sloppy code = perception of sloppy security Changes: Remove all commented-out code (500+ lines deleted)Add type hints (Python 3.9+ everywhere)Extract magic numbers to named constantsDocument complex logic with inline commentsBasic test coverage for critical paths Remove all commented-out code (500+ lines deleted) Add type hints (Python 3.9+ everywhere) Extract magic numbers to named constants Document complex logic with inline comments Basic test coverage for critical paths Not aiming for perfection. Aiming for professional. professional. What I Learned Building to 60 (That Applies to 1,000) 1. Transparency Compounds 1. Transparency Compounds Early on, I hid problems. Rebuild #3 happened because of bad architecture. I just released v1.4 like nothing was wrong. Later, I started sharing: "Rebuilt the stats engine because CSV files hit 500MB""Threading model was wrong, causing UI lag""Fan control crashed on AMD motherboards" "Rebuilt the stats engine because CSV files hit 500MB" "Threading model was wrong, causing UI lag" "Fan control crashed on AMD motherboards" Result: More engagement, more trust, more helpful feedback. Result: Lesson: Users prefer honest struggle over polished perfection. Lesson: 2. Metrics Lie, Signals Don't 2. Metrics Lie, Signals Don't 60 downloads = metric. 40% repeat visitors = signal (they checked, thought about it, came back). 14 GitHub stars = metric. 11 watchers actively tracking development = signal (they care about updates). Lesson: Optimize for signals, not metrics. Lesson: 3. Infrastructure > Features (At First) 3. Infrastructure > Features (At First) Most of v1.7.x isn't features. It's: CONTRIBUTING.md (shows project is welcoming)Issue templates (makes bug reports useful)Automated release notes (saves time)Public roadmap (shows direction) CONTRIBUTING.md (shows project is welcoming) Issue templates (makes bug reports useful) Automated release notes (saves time) Public roadmap (shows direction) These don't add functionality. They add credibility. credibility. For a solo dev, credibility unlocks contributors, feedback, and trust. 4. The Store Isn't the Finish Line 4. The Store Isn't the Finish Line Microsoft Store approval won't suddenly make PC_Workman popular. But it will: Remove trust barriers (no SmartScreen, verified publisher)Enable organic discovery (1B+ Windows users search Store)Force quality standards (sandboxing, review process)Provide update infrastructure (automatic, seamless) Remove trust barriers (no SmartScreen, verified publisher) Enable organic discovery (1B+ Windows users search Store) Force quality standards (sandboxing, review process) Provide update infrastructure (automatic, seamless) The real goal: The real goal: Create a feedback loop where: Store for easier to trustEasier to trust for more downloadsMore downloads for more feedbackMore feedback for better productBetter product for easier to recommendRecommendations for more organic growth Store for easier to trust for Easier to trust for more downloads for More downloads for more feedback for More feedback for better product for Better product for easier to recommend for Recommendations for more organic growth for 80 downloads -> 200 -> 500 -> Microsoft featuring in "Utilities" category. 80 downloads -> 200 -> 500 -> Microsoft featuring in "Utilities" category. That's the path. Store is step one. What's Next v1.7.0 ships late March (auto-update check, UI polish). v1.7.9 ships June (TURBO BOOST complete, code freeze). v2.0 ships Q3 2026 (Microsoft Store, MSIX packaging, local AI). Between now and then: Weekly roadmap updates (GitHub Discussions)Build-in-public posts (LinkedIn, Twitter)Technical deep-dives (DEV.to, Medium)Community feedback integration (Issues, Discussions) Weekly roadmap updates (GitHub Discussions) Build-in-public posts (LinkedIn, Twitter) Technical deep-dives (DEV.to, Medium) Community feedback integration (Issues, Discussions) 800 hours so far. Probably 400 more until Store submission. Built on constraints. Optimized for reality. Shipped despite imperfection. Not because I'm talented. Because I'm stubborn. If You're Building Something Similar Don't wait for perfect. Don't wait for perfect. Ship what you have. Improve it publicly. Don't chase vanity metrics. Don't chase vanity metrics. 60 engaged users > 1,000 passive downloads. Don't skip infrastructure. Don't skip infrastructure. CONTRIBUTING.md matters more than you think. Don't underestimate legitimacy. Don't underestimate legitimacy. Microsoft Store approval = third-party validation worth more than self-promotion. Don't build alone if you don't have to. Don't build alone if you don't have to. I built solo because I had no choice. You might. Open source invites collaboration. Use it. Links PC_Workman on GitHub: https://github.com/HuckleR2003/PC_Workman_HCK PC_Workman on GitHub: https://github.com/HuckleR2003/PC_Workman_HCK Development roadmap: https://github.com/users/HuckleR2003/projects/3 Development roadmap: https://github.com/users/HuckleR2003/projects/3 All my links (Twitter, LinkedIn, Ko-fi, etc.): https://linktr.ee/marcin_firmuga All my links (Twitter, LinkedIn, Ko-fi, etc.): https://linktr.ee/marcin_firmuga Follow the journey: Follow the journey: Twitter: @hck_lab (daily updates) @hck_lab