Indicode.io
In December 2020, I co-founded Indicode.io with Nick & Eryl to improve manufacturing quality control through deep learning. Our goal was to create a defect detection system that could work with manufacturers’ existing hardware while leveraging deep learning to identify issues.
We began by experimenting with autoencoder models trained on synthetic data, eventually moving to CNN-based anomaly detection. Our early results caught TOSFA’s attention, leading to pre-seed funding.
Over the next few months, we built out a web platform and connected with several manufacturers who were intrigued by our approach.
Where We Stumbled
The manufacturing industry proved more challenging than we anticipated:
Long Sales Cycles: What we thought would be 3-month sales cycles stretched into 6+ months of conversations and demonstrations.
Trust Deficit: Manufacturers were hesitant to work with an unknown startup, even when our technology showed promise.
Misaligned Incentives: Production managers had little motivation to change their existing processes. The potential disruption outweighed the benefits.
PoC Paralysis: We found ourselves caught in endless proof-of-concept projects that never converted to implementations. We should have focused on securing just one paying customer, even at a loss.
Key Lessons Learned
Looking back, several valuable lessons emerged from this experience:
Sell Before Building: We spent too much time perfecting our platform before securing solid customer commitments. A better approach would have been to focus on sales and customer validation before extensive development.
Direct Outreach is Crucial: While we had some success with inbound interest, we should have been more aggressive with direct outreach to potential clients. Building relationships early in the sales cycle could have helped overcome the trust deficit we faced.
Be Willing to Take Initial Losses: One of our biggest mistakes was not being willing to take a loss on an initial client to build credibility. Having a reference customer, even at unfavorable terms, would have been invaluable for future sales.
Avoid Pre-optimization: We often found ourselves solving problems we thought customers would have, rather than addressing their actual, validated needs. This led to wasted development effort on features that weren’t critical to securing initial sales.
Technical Growth
The project taught me a lot about:
- Building production Django apps
- Deploying ML models on-premise
- Creating pitch decks and financial projections
- B2B sales in manufacturing
Tech Stack
Frontend & UI
Backend & Storage
Infrastructure
Conclusion
Indicode didn’t work out, but it changed how I think about building startups, especially in manufacturing. This was my first realisation that the “build it and they will come” mentality does not work. Good tech isn’t enough - you need market understanding and customer relationships from the start.