# Codeasy

> AI engineering company building production-grade systems since 2013.

Codeasy (legal entity: Loop d.o.o.) is an AI engineering company founded in 2013 in Split, Croatia. We design and ship end-to-end AI systems: computer vision, AI agents, voice AI, workflow automation, and full-stack delivery. A senior team, scaled to each project, shipping software since 2013 for clients in Norway, the United Kingdom, the United States and Croatia.

## Services

- Custom AI development (computer vision, AI agents, voice systems)
- Workflow automation and data pipelines
- Full-stack product delivery (architecture, UX, implementation, deployment)
- Full-stack web and mobile apps, back-end systems and APIs, UI/UX design
- Data engineering and MLOps
- AI strategy and audit

## Products

- [Codeasy Chat](https://chat.codeasy.com/): Secure AI assistant workspace for teams. Chat with uploaded documents, configure assistants per team, embed on your website.
- [Voxares](https://voxares.com/): SMS-first client communication automation for appointment-driven businesses. Reminders, confirmations, digital forms, reactivation campaigns.
- [Guide To Go](https://www.guidetogo.com/): GPS-triggered audioguide platform for tourism. Location-triggered audio, fully offline, multi-language tours.
- [Trebam Poklon](https://trebampoklon.com/): Curated gift marketplace connecting Croatian artisans with gift-seekers. 1,000+ handmade products.

## Case Studies

- Senzea (2025, ongoing; healthcare AI): the layer around Senzea's Masked-VAE model that turns raw predictions into ranked, explainable signals inside a clinical review workflow.
- EndemicAI (2025; computer vision): 3M+ artworks run through AWS Bedrock and Amazon Rekognition to detect duplicates and flag stolen art on Endemic, a curated NFT art marketplace on Ethereum.
- Elnav Aware Mate (2025; maritime computer vision): YOLO and MediaPipe pipeline that measures officer-of-the-watch eye closure and drowsiness (PERCLOS), spots distraction, and feeds alerts into the ship's bridge navigational watch alarm system (BWNAS).
- Elnav Helm (2024; voice AI): voice-command engine for ship bridges with speaker separation and command-chain state.
- Voice Of Norway (2021, Experio AS; tourism): GPS-triggered audio guides, AR scenes and outdoor games for museums and transport operators; back end, operator web app and mobile app, live across Norway; later grew into the Guide To Go product.
- Swiftly (2020, Qrypta Ltd.; telehealth): scheduling, video calls, SMS, intake forms and secure payment for UK GP and dental practices, used within the NHS; serverless on AWS.
- Hajduk User Account (2021, Hajduk; sports): user account backend and single sign-on for the football club from Split.
- OfflinePal / Enjoy.ly (2019, Enjoy AS; tourism): activity booking marketplace for Norway with Stripe and Vipps payments and an Organizer app for operators.
- PatientPal (2018, PatientPal Inc.; healthcare): appointments, patient intake forms and insurance payments, serverless on AWS with DynamoDB.
- Enswitch (2019, Integrics Ltd.; telecom): rebuilt telephony management interface with multi-language support, client plugins and custom reports.

## Industries

Healthcare, maritime, tourism, telecom, sports, art markets, ecology, e-commerce.

## Community

- [ŠK Codeasy](https://sk.codeasy.com): children's chess club in Vukovar sponsored by Codeasy; 60+ kids, FIDE-rated training, domestic and international tournaments.
- [POMOZI.hr](https://pomozi.hr): free, anonymous AI chat for psychological support in Croatian, 24/7, no registration and no tracking; built and run by Codeasy as a gift to the community; finalist at the 2024 Healthcomm Forum awards for best communication project in healthcare.

## Process

Discover (one call, then a written scope), Design (architecture, data flow and trade-offs on paper before code), Ship (short cycles, working software deployed to your environment and reviewed together), Iterate (measure the agreed metric, fix what shipped, extend what works).

## Open Data

- [Underwater Seafloor Litter Drone Image Database](https://codeasy.com/seafloor-litter-database): Labeled drone images of seafloor litter for training object detection models (5,405 images, 12,786 labels, 31 classes). About 500 GB in total, so it is shared on request rather than hosted for download.

## Contact

- Website: https://codeasy.com
- Email: info@codeasy.com
- LinkedIn: https://www.linkedin.com/company/codeasycom
- Instagram: https://www.instagram.com/codeasycom/
- Address: Tehnološki park Split, Dračevac 3D, 21000 Split, Croatia
- Press coverage: https://codeasy.com/media

## Links

- [Homepage](https://codeasy.com/)
- [Privacy Policy](https://codeasy.com/privacy-policy)
- [Legal](https://codeasy.com/legal)
- [Seafloor Litter Database](https://codeasy.com/seafloor-litter-database)
- [Full LLM context](https://codeasy.com/llms-full.txt)