Leo (@LeoSparr) header
Leo (@LeoSparr) profile picture

Computational chemist at Lund University @cadenbotAI

Sweden
Joined September 2022
Photo by Leo (@LeoSparr)Photo by Leo (@LeoSparr)Photo by Leo (@LeoSparr)Photo by Leo (@LeoSparr)
Pinned Tweet
User feedback is the #1 bottleneck for SaaS right now.
64
I want to click founder-built stuff. Drop your URL.
3
31
If you don't have frequent user feedback built into your AI workflow in 2026... Seriously, what are you even doing?
1
1
2,281
SaaS founders 👀 What are you building? Drop your SaaS below
7
1
127
Your AI coding workflow is incomplete without real human feedback. Here’s how I optimized my SaaS for my users preferences, not my own: I used Claude and the Testbay MCP to get feedback from real people, spot what needed fixing, and make my product easier to use. 1. Install the Testbay MCP → testbay.dev/mcp 2. Ask Claude: “Help me get useful feedback on my website or iOS app. Create a Testbay post and help me choose the right questions.” 3. Publish it on Testbay and get feedback from testers. That’s it. Claude helps set up the test, and you get real user feedback to guide what to improve. Human feedback in the loop. That’s the missing piece
🤖 Made with AI
3
4
2,755
Founders, drop your SaaS/app. No description. No pitch. Free traffic.
10
7
173
Vibe coding is fun. But if you want to sell something, you can’t just build whatever you personally think is cool.
2
2
87
High quality feedback is hard to find for free. Just pay for it
1
1
29
25 reasons your app looks vibecoded 1. emojis 2. bento grids 3. radial orbs 4. lucide icons 5. soft corner radius 6. neon colors 7. em dashes 8. dot grids 9. pure white background 10. checkmark bullets 11. liquid glass 12. purple and black 13. "it's not x, it's y" 14. animated arrows 15. rainbow coloring 16. harsh gradients 17. no skeleton loaders 18. colored left stripe 19. drop shadows 20. sparkle icons 21. terminal window 22. no TOS 23. no privacy policy 24. hover animations 25. basic pastel colors Copy paste into a DESIGN.md file, works for all coding agents such as codex, claude code or antigravity.
3
179
This is what i have built the last 5 months: testflight.apple.com/join/xN… Caden, Ai for runners. Try it for free!
2
165
This is what i have built the last 5 months: testflight.apple.com/join/xN… Caden, Ai for runners. Try it for free!
31
This is what i have built the last 5 months: testflight.apple.com/join/xN… Caden, Ai for runners. Try it for free!
73
AI just pushed a particle physics calculation from 3 loops to 9. Claude worked through a frontier scattering-amplitude problem largely autonomously, with the result independently checked by a Stanford/SLAC physicist. anthropic.com/research/yes-c…
1
23
Heat seems to narrow the performance gap between men and women in ultramarathons. Across 1.35 million finishes, the speed gap fell from 0.73 km/h in cool conditions to 0.34 km/h at 34°C — and just 0.21 km/h at 90% humidity. dwrowland.substack.com/p/res…
13
Half marathon pace is funny. Too fast to relax. Too slow to get it over with quickly
13
Great use case for jev everyone!
Jev can't write a single word, which is exactly why it should pick what my agents read my Obsidian vaults hold 5k+ notes, one idea per note, each with a one-line summary in an index but Claude can't read all of that on every turn, and a keyword search misses the note that makes the same point in different words so Jev does the picking: > step 1: Jev reads the 28 vault descriptions and picks where the task belongs > step 2: it checks every note summary in that vault in one call, each judged alone > step 3: the top 5 come back with full text - Jev drops any that miss the task > step 4: my code loads the survivors into Claude's context (and nothing else) every pick comes from a list i define, so Jev can't invent a note the worst it can do is pick the wrong one... and each one carries a confidence score that tells my code when to fall back to Claude
30
One thing everybody should do if they really want to understand the fundumentals of ai is coding one simple llm yourself from scratch.
10