🧵Thread: I Accidentally Built a More Secure RDP MFA System Than Cisco Duo I tried adding MFA to a Windows RDP connection… and ended up building something more secure than Cisco Duo itself. Here’s the brutal truth nobody talks about 👇

That's how the system work

Google's antigravity experiment isn't about defying physics. It's about reframing constraints. When you hit a wall in your code, sometimes the solution isn't pushing harder. It's questioning the wall itself. Most breakthroughs come from asking better questions, not finding

AI coding assistants are incredible, but they can't replace good CLI tools. AI helps you write the initial implementation. CLI tools handle the repetitive execution. Example: I use Claude to help structure my JSON-to-TS parser logic. But I use the CLI to actually run it 50

Revenue: $7/month. Validation: priceless. First FaceInsightAI subscriber just signed up. The spreadsheet finally has a number that isn't zero.

Beginners want clarity, intermediates want confidence, and experts want shortcuts.

Built my first CLI tool today. Automates something I was doing manually 10+ times a week. The code is simple. The time saved is not. 85 lines that turned 15 minutes of repetitive work into 30 seconds of automation. Sometimes the best tools solve the smallest problems.

The real AI opportunity isn't replacing developers. It's automating the stuff nobody wanted to build software for in the first place. Legal docs. Expense reports. Meeting notes. All the boring work that costs $200/hour but nobody wants to pay for software. That's the play.

Revenue: $7/month. Validation: priceless. First FaceInsightAI subscriber just signed up. The spreadsheet finally has a number that isn't zero.

Pattern I keep seeing: smart people make the same mistakes, just faster. Hire too quickly. Build features nobody asked for. Optimize before validating. The difference isn't avoiding mistakes. It's recognizing them in week 2 instead of month 6. Speed of learning beats speed of

Pattern I keep seeing: smart people make the same mistakes, just faster. Hire too quickly. Build features nobody asked for. Optimize before validating. The difference isn't avoiding mistakes. It's recognizing them in week 2 instead of month 6. Speed of learning beats speed of

Claude Code now lets you drop custom rules in .claude/rules/ as markdown files. Code style. Testing conventions. Security requirements. Auto-loaded with the same priority as .cursorrules. The interesting part? Three-tier config system - project rules, local user preferences, or

The best build-in-public content comes from people who forgot they were supposed to be creating content. They just shipped something useful and shared it because it solved a real problem. The rest is performance anxiety disguised as transparency.

optimized a slow database query this week that was taking 8 seconds to return results. added a composite index on the join columns and moved the filtering logic before the join. query time dropped to 200ms. database performance is 90% about understanding how your queries

The best build-in-public content comes from people who forgot they were supposed to be creating content. They just shipped something useful and shared it because it solved a real problem. The rest is performance anxiety disguised as transparency.

73% of ChatGPT usage is non-work tasks like shopping and meal planning. models score 45% on these consumer workflows. the infrastructure spending focuses on reasoning and coding, but most users just want reliable product recommendations without hallucinated prices. gap between

Implementing pricing tiers today. The technical part took 20 minutes. The strategic part took 3 days. Here's what I learned: Your pricing page isn't a feature list. It's a decision tree. Each tier needs to answer one question: "Is this obviously for me?" If a user has to

The full-stack developer toolkit just shifted. 2020: monolithic frameworks, manual deployments, REST everywhere 2025: edge-first architecture, serverless by default, tRPC for type safety Your deployment pipeline matters more than your framework choice now. Infrastructure

Most SaaS companies overthink pricing. Users don't compare your tiers. They compare your price to their current pain level. If the pain is high enough, they'll pay. If it's not, your clever tier names won't matter. Price for the problem you solve, not the features you ship.

checkout:

generic skincare advice: retinol drink water wear sunscreen personalized skincare: analyzing YOUR skin's moisture levels, sensitivity patterns, and environmental factors to recommend the exact products that work for your biology. FaceInsight builds routines for your skin,