Henry Coder
First AI Smart Contracts Translator From Solidity to Ralph
First AI translator for smart contracts from Solidity to Ralph (custom language on Alephium). This cuts down time of writing new smart contracts by 87%.
87%
Downtime creating SC
+1k
Active users
Maud Bannwart
Chief Operating Officer - Alephium
"With Alephium, we’ve had the opportunity to work with the talented Blockchain Collab team for several years. They’ve made a tremendous impact on our ecosystem, most recently with Henry Coder, a tool that makes developer onboarding easier than ever."
The Challenge
First AI translator for smart contracts from Solidity to Ralph (custom language on Alephium). This cuts down time of writing new smart contracts by 87%. Henry Coder allows faster smart contract creation while maintaining accuracy and security.
This is the first tool of this kind on Alephium. Connecting AI and custom smart contract language was a great challenge. We solved it and now builders have an easier life building new dApps on Alephium.
Development process and time
We’ve begun the project in a time where agentic systems were taking their baby steps - think Cursor was starting to gain attention on Twitter. This and lack of high-end high-quality models was a severe hurdle. We understood from the get-go that the translation context must be refined and fine tuned for the specific LLM that we will use for the approach. This was not made any easier by context rot which resulted in the LLM often forgetting principles of Ralph. In the end we have pushed through and resolved this by strategically planning the context and the information density of provided Ralph documentation. On top of that, the most important ideas were repeated multiple times with multiple examples to guarantee successful translations in many different configurations.
The second phase of the project drastically scaled the capabilities of an AI Agent, not just a simple chat based translation with context injection. The agent is now able to dynamically fetch the necessary libraries for the translation context. While the agentic process is run by small LLM, the main translation process has been kept sturdy with a SOTA LLM as the brains of the operations.
In the hands of a seasoned Alephium developer, Henry cuts development times of a Ralph contracts by 90%
Time to deliver:
MVP: 1 month
Phase II with an agentic approach: 1 month
Technologies and Integrations
TypeScript with Nuxt.js frontend (Relationship between Nuxt and Vue is the same as Next and React) integrated with a Python backend (pydantic + OpenAI). Without mature external frameworks at the time we’ve built a baseline translation engine which relied on Deepseek V3/R1 (decided based on contract complexity).
In Phase 2 the focus has shifted to Langchain and Langgraph as the agentic backend with SOTA LLMs as the translation brains.
visit website
Back to our work
Henry Coder
First AI Smart Contracts Translator From Solidity to Ralph
First AI translator for smart contracts from Solidity to Ralph (custom language on Alephium). This cuts down time of writing new smart contracts by 87%.
87%
Downtime creating SC
+1k
Active users
Maud Bannwart
Chief Operating Officer - Alephium
"With Alephium, we’ve had the opportunity to work with the talented Blockchain Collab team for several years. They’ve made a tremendous impact on our ecosystem, most recently with Henry Coder, a tool that makes developer onboarding easier than ever."
The Challenge
First AI translator for smart contracts from Solidity to Ralph (custom language on Alephium). This cuts down time of writing new smart contracts by 87%. Henry Coder allows faster smart contract creation while maintaining accuracy and security.
This is the first tool of this kind on Alephium. Connecting AI and custom smart contract language was a great challenge. We solved it and now builders have an easier life building new dApps on Alephium.
Development process and time
We’ve begun the project in a time where agentic systems were taking their baby steps - think Cursor was starting to gain attention on Twitter. This and lack of high-end high-quality models was a severe hurdle. We understood from the get-go that the translation context must be refined and fine tuned for the specific LLM that we will use for the approach. This was not made any easier by context rot which resulted in the LLM often forgetting principles of Ralph. In the end we have pushed through and resolved this by strategically planning the context and the information density of provided Ralph documentation. On top of that, the most important ideas were repeated multiple times with multiple examples to guarantee successful translations in many different configurations.
The second phase of the project drastically scaled the capabilities of an AI Agent, not just a simple chat based translation with context injection. The agent is now able to dynamically fetch the necessary libraries for the translation context. While the agentic process is run by small LLM, the main translation process has been kept sturdy with a SOTA LLM as the brains of the operations.
In the hands of a seasoned Alephium developer, Henry cuts development times of a Ralph contracts by 90%
Time to deliver:
MVP: 1 month
Phase II with an agentic approach: 1 month
Technologies and Integrations
TypeScript with Nuxt.js frontend (Relationship between Nuxt and Vue is the same as Next and React) integrated with a Python backend (pydantic + OpenAI). Without mature external frameworks at the time we’ve built a baseline translation engine which relied on Deepseek V3/R1 (decided based on contract complexity).
In Phase 2 the focus has shifted to Langchain and Langgraph as the agentic backend with SOTA LLMs as the translation brains.
visit website
Back to our work
Henry Coder
First AI Smart Contracts Translator From Solidity to Ralph
First AI translator for smart contracts from Solidity to Ralph (custom language on Alephium). This cuts down time of writing new smart contracts by 87%.
87%
Downtime creating SC
+1k
Active users
Maud Bannwart
Chief Operating Officer - Alephium
"With Alephium, we’ve had the opportunity to work with the talented Blockchain Collab team for several years. They’ve made a tremendous impact on our ecosystem, most recently with Henry Coder, a tool that makes developer onboarding easier than ever."
The Challenge
First AI translator for smart contracts from Solidity to Ralph (custom language on Alephium). This cuts down time of writing new smart contracts by 87%. Henry Coder allows faster smart contract creation while maintaining accuracy and security.
This is the first tool of this kind on Alephium. Connecting AI and custom smart contract language was a great challenge. We solved it and now builders have an easier life building new dApps on Alephium.
Development process and time
We’ve begun the project in a time where agentic systems were taking their baby steps - think Cursor was starting to gain attention on Twitter. This and lack of high-end high-quality models was a severe hurdle. We understood from the get-go that the translation context must be refined and fine tuned for the specific LLM that we will use for the approach. This was not made any easier by context rot which resulted in the LLM often forgetting principles of Ralph. In the end we have pushed through and resolved this by strategically planning the context and the information density of provided Ralph documentation. On top of that, the most important ideas were repeated multiple times with multiple examples to guarantee successful translations in many different configurations.
The second phase of the project drastically scaled the capabilities of an AI Agent, not just a simple chat based translation with context injection. The agent is now able to dynamically fetch the necessary libraries for the translation context. While the agentic process is run by small LLM, the main translation process has been kept sturdy with a SOTA LLM as the brains of the operations.
In the hands of a seasoned Alephium developer, Henry cuts development times of a Ralph contracts by 90%
Time to deliver:
MVP: 1 month
Phase II with an agentic approach: 1 month
Technologies and Integrations
TypeScript with Nuxt.js frontend (Relationship between Nuxt and Vue is the same as Next and React) integrated with a Python backend (pydantic + OpenAI). Without mature external frameworks at the time we’ve built a baseline translation engine which relied on Deepseek V3/R1 (decided based on contract complexity).
In Phase 2 the focus has shifted to Langchain and Langgraph as the agentic backend with SOTA LLMs as the translation brains.
visit website
Back to our work