I built an AI that argues against your idea instead of validating it — would love brutal feedback
Most of us pitch ideas to ChatGPT and it tells us they're great. I got tired of that, so I built the opposite: 7 AI agents with opposing perspectives debate your idea in 3 rounds. One agent exists only to destroy your idea. A Judge delivers a final score.
I'm at $0 revenue, 0 customers, launched this week. Offering free sessions to the first 5 IH members who want to test it — just reply with your idea or a decision you're wrestling with.
councilia.com — honest feedback welcome.
Trending on Indie Hackers
The dedicated destroyer agent is a smart move. Curious how the judge scores - weighted toward feasibility or market demand?
Hello Indie Hackers! 👋
I'm excited to share that my latest micro-SaaS, SachCheck AI, just got approved and featured on the SideProjectors homepage!
The Problem:
In India, fake news in regional languages like Hindi spreads like wildfire. Most tools are built for English, leaving 600M+ Hindi speakers vulnerable.
The Solution:
SachCheck AI is a lightweight tool that uses the Google Fact Check API to verify claims instantly in Hindi.
Tech Stack:
I am now looking for a new owner to take this forward and scale it. You can see the live listing here: https://www.sideprojectors.com/project/sach-check-
Would love your feedback on the tool!
Love this angle — stress-testing ideas from the inside out. I built something complementary: DemandRadar scans HN/ProductHunt/IndieHackers daily to find if people are already complaining about the problem your idea solves. Real demand signals before you build. The two tools work well together — first check if the pain exists in the wild, then stress-test your solution with yours.
Multi-agent debates are great for reducing confirmation bias. The tricky part is keeping each agent committed to its role across all 3 rounds. A "destroyer" agent with a vague system prompt tends to soften and hedge instead of truly opposing.
Each agent needs a tight role, a clear objective, and explicit constraints that lock it to its perspective. When those are separate typed blocks rather than one mixed paragraph, the persona holds under pressure.
I've been building flompt for exactly this, a visual prompt builder that decomposes prompts into 12 semantic blocks and compiles to Claude-optimized XML. Open-source: github.com/Nyrok/flompt