At Lyvena, we recently completed a project that exemplifies our applied AI delivery methodology: Kiews, a no-code platform that lets users integrate NEAR Protocol functionality into their applications without writing a single line of blockchain code.
Project Snapshot
Kiews enables drag-and-drop NEAR Protocol integration — smart contract deployment, token transactions, and NFT minting — through a visual interface powered by AI.
The Problem
NEAR Protocol is one of the most developer-friendly blockchain platforms, but it still requires Rust or JavaScript proficiency to build meaningful integrations. For businesses exploring blockchain capabilities, the cost of hiring specialized developers or training existing teams creates a barrier to entry.
We saw an opportunity: use AI to abstract away the blockchain complexity. If a user could describe what they wanted in natural language, and the system could generate, test, and deploy the necessary NEAR Protocol code, that would lower the barrier by an order of magnitude.
Our Delivery Methodology
We applied our Audit, Design, Pilot, Scale framework to the Kiews project:
Phase 1: Audit
We spent two weeks mapping the NEAR Protocol integration landscape: common integration patterns, pain points, API surfaces, and security requirements. This phase identified that the highest-impact target was smart contract deployment and token transactions — the operations that most businesses needed but found hardest to implement.
Phase 2: Design
The design phase focused on two questions: how do we make AI generate reliable blockchain code, and how do we make the interface intuitive for non-developers? The solution was a two-layer architecture:
- A visual workflow builder — users drag and drop blocks representing NEAR operations (deploy contract, transfer tokens, mint NFT) and connect them into pipelines.
- An AI code generator — trained on NEAR Protocol documentation and thousands of example implementations, the AI translates visual workflows into optimized Rust or JavaScript code.
Phase 3: Pilot
We ran a four-week pilot with five businesses that were exploring NEAR Protocol integration. During the pilot, we measured code generation accuracy, user satisfaction, and deployment success rates. The results were encouraging: 87% of AI-generated contracts compiled on first try, and users reported a 70% reduction in integration time compared to traditional development.
Phase 4: Scale
Based on pilot feedback, we expanded Kiews to support additional NEAR features: cross-contract calls, storage management, access control, and event indexing. The visual workflow builder grew to include conditional logic, loops, and error handling.
Technical Architecture
Kiews is built on a modern AI stack:
- Frontend: Next.js with a drag-and-drop workflow interface
- AI engine: Fine-tuned Llama 3 on NEAR Protocol code and documentation
- Backend: FastAPI with WebSocket support for real-time code generation
- Blockchain layer: NEAR Protocol SDK for Rust and JavaScript
- Testing: Automated NEAR sandbox for compiling and testing generated contracts
87%
First-try compile rate
70%
Faster integration time
5
Pilot participants
“Kiews doesn't replace blockchain developers. It removes the friction of getting started — and that changes the adoption calculus entirely.”
Kiews is currently in private beta, and we're expanding the platform to support additional blockchains. The lesson from this project is clear: applied AI delivery works best when it removes a specific, measurable friction — in this case, the cost and complexity of first-time blockchain integration.