Every AI app has the same silent problem: you're paying 3–5× more than you need to for LLM calls. Decomp.ai audits your pipeline in 60 seconds, finds which calls can be replaced (sentiment checks, formatting, repeated queries) and gives you before/after code diffs plus one-click prompts to paste into Cursor or Claude Code. Free scan. Pay 5% of your estimated monthly saving to unlock fixes (S$4.99–S$9.99 max). Zero if we find nothing. Use DECOMPLAUNCH, Free during launch week.
Hey Product Hunt! 👋
I built Decomp.ai after noticing that most AI apps I've seen — including my own — are quietly burning money on LLM calls that didn't need to be LLM calls.
Things like:
Classifying sentiment with GPT-4 (a regex would do it)
Reformatting JSON outputs (that's just code)
Calling the API on every request when the answer doesn't change
The audit takes under a minute. You paste in your pipeline, it finds the waste, shows you real before/after code (blurred until you unlock), and gives you a prompt you can paste straight into Cursor to apply the fix.
The pricing model is the part I'm most proud of: you only pay if we find savings, and you pay 5% of what we estimate you'll save per month. Max S$9.99. If we find nothing useful, it's free.
Use DECOMPLAUNCH for free access during launch week — I want early users to get full value before I start charging.
Would love feedback on the audit quality. Drop your pipeline in and let me know what you think!
Report
No reviews yetBe the first to leave a review for Decomp.ai — Agent Cost Optimisation