top of page


When Andrej Karpathy tweeted (X’d?), "English is the hottest new programming language", he wasn't kidding. Over the past few years, we've seen a paradigm shift in the interaction between humans and machines. And at Hannibal AI, this has radically transformed how we approach, design, and implement our solutions. Here's our journey through re-learning English as a Programming Language.

At their core, programming languages are just a way to formally define the logic we want a machine to follow. Traditionally, this has been best accomplished with strict syntax and patterns that can be converted directly into machine code. With prompt engineering, we must still accurately and clearly define our logic, but without reliance on strict syntactical standards.

Chains of English

In essence, the AI coding process requires scripting the flow in Python, deploying servers using JavaScript languages, and curating prompts in English. Conventionally, the use of English was predominantly confined to client communications, documentation, and task tracking systems like Jira. But as AI's role in development amplifies, English takes center stage running the core functionalities of AI applications.

Quality prompts have become our primary language for churning out consistent and dependable results. This evolution mandates our engineers to not only be adept at programming but also to be exceptional writers. They need to craft prompts that epitomize clarity, conciseness, and precision. And as with all coding, attaining perfection often demands multiple iterations.

This simplified chain from OrthoScribe is just used to set the context for the model, prior to generating any results.

  • Request 2 models to transcribe a patient-doctor conversation

  • Checking the transcription for errors against each other

  • Prompting it to correct unclear or inaccurate sections

  • Validating the updated transcription

  • Adding historical context from the patient’s medical record

  • Adding medically relevant information based on the interaction and medical history

This refinement process allows us to efficiently guide the model from a rough initial transcription to a reliable source of truth for our models. Chaining in this way enables iteration through results through essentially natural conversation.

‘Stack’ it on Top

While English may be the "hottest" new programming language, it certainly doesn't replace the necessity of understanding and mastering supporting languages. In the dynamic realm of technology, there's a delicate dance between innovation and foundational knowledge. At Hannibal AI, we recognize the value of both.

Taking a concept from prompting to production involves multiple layers of technical expertise:

  • JavaScript (Web Stuff): Every interactive element you see on modern websites and applications likely has JavaScript at its heart. It brings to life the designs and interactive functionalities that make web experiences intuitive and engaging.

  • Databases (Storage Stuff): Whether it's the latest user sign-up or a transaction record, databases play a pivotal role in storing, managing, and retrieving this data efficiently. They are the unsung heroes that ensure data integrity and availability.

  • Python (Data Stuff): Recognized for its simplicity and power, Python has become the go-to for a vast range of applications, especially in data analysis, machine learning, and AI. It’s the bridge between raw data and meaningful insights.

  • DevOps (Delivery Stuff): An oversimplified description would be that DevOps ensures everything works seamlessly together. It’s the amalgamation of practices, tools, and philosophies that increases an organization's ability to deliver applications and services swiftly and consistently.

Understanding and leveraging English as a programming language offers a fresh perspective, making complex AI-driven processes more accessible. However, the intricacies of bringing an AI concept into a fully functional, user-facing product still heavily rely on the robustness and efficiency of traditional programming languages and practices.

English Programming

At Hannibal AI, English programming has become deeply integrated into our development process. Our team leverages it in concert with traditional code to efficiently build customized AI solutions. While still an emerging technique, we see the use of natural language as a profoundly promising paradigm shift.

As AI capabilities rapidly evolve, English ‘programming’ will likely become a standard part of how humans collaborate with machines. The future of our industry will require broader skill sets, with a heavy emphasis on the lost art of writing. Hannibal AI looks forward to exploring this brave new frontier, and perhaps blazing a trail for others to follow.


Group 1171275591 1.png


bottom of page