Generative AI. What a journey it’s been. This time two years ago, like many, I hadn’t heard of the term, let alone understood the possible capabilities. Two years later we’re in the midst of a generative AI revolution, with a plethora of tools available for all different use cases.
Wondering which AI tool will supercharge your Salesforce workflow? We’re putting ChatGPT, Google Gemini, Claud, Einstein for Developers, and GitHub Copilot to the test. Our in-depth analysis and comparison table will guide you toward the most effective AI solutions for your unique Salesforce challenges. This is based on the results of the recent Salesforce Ben developer survey, where we surveyed over 350 Salesforce Developers. You won’t be surprised to learn that AI was a big topic of discussion. You can download the full report based on the survey here.
ChatGPT
ChatGPT, developed by OpenAI, is a general-purpose conversational AI that’s versatile and adaptable. It can handle a wide range of queries, from general questions to specific programming problems, making it useful for explaining concepts, generating code snippets, and debugging. Its integration capabilities via API allow it to be embedded into various tools and platforms.
However, it lacks specialization in Salesforce development, so some advanced Salesforce-specific tasks might require more manual input. Additionally, its limited memory of previous interactions within a single session can pose challenges for continuous development work.
Google Gemini
Gemini, created by Google DeepMind, is an advanced AI with strong capabilities in both natural language understanding and generation. It excels at complex problem-solving and understanding nuanced questions, making it beneficial for intricate Salesforce configurations and custom development. With its rich context handling, Gemini can manage detailed and intricate questions effectively.
However, like ChatGPT, it’s not specifically designed for Salesforce development. Access to Gemini might also be more limited compared to other tools.
Claude
Claude, developed by Anthropic, is an AI that’s focused on safety and usability in language generation. It emphasizes generating safe and reliable responses and is designed to be user-friendly and intuitive. Claude’s ability to maintain better context over longer interactions is beneficial for development work.
Despite these strengths, Claude is not specifically tailored for Salesforce or any specific development environment and may need additional prompts or instructions to handle very specific Salesforce-related queries effectively.
Einstein for Developers
Einstein for Developers is a suite of AI tools integrated within the Salesforce product suite catalog, developed by Salesforce. It offers seamless integration with Salesforce, providing in-context assistance directly within the platform.
Designed specifically for Salesforce, it offers tailored insights and automation capabilities, including predictive analytics, natural language processing for Salesforce-specific tasks, and custom AI model creation. However, its scope is limited to the Salesforce ecosystem, making it less useful for tasks outside of Salesforce. There might also be a learning curve to fully leverage all of its features.
GitHub Copilot
GitHub Copilot, powered by OpenAI Codex and developed by GitHub, is an AI-powered code completion tool that provides real-time code suggestions and completions, significantly speeding up the coding process. It integrates well with popular IDEs like VS Code, making it easy to adopt. GitHub Copilot supports multiple programming languages and frameworks.
However, it’s not specifically tailored for Salesforce development, meaning that general code suggestions might not always align perfectly with Salesforce’s Apex language or Lightning framework, and the quality of suggestions can vary, especially for highly specialized code or complex logic.
One important thing to note about GitHub Copilot is that while all other tools on this list have free versions, GitHub Copilot does not.
Use Case Winners
So, we’ve given an introduction into what each of these tools are, but what are they good for? Here, we dive into various Salesforce-specific use cases and which tools are best suited to each one.
Apex Classes and Triggers
Of all the tools listed, Github Copilot is the ultimate winner when it comes to writing code. As GitHub Copilot is designed specifically for code generation and completion, it’s been trained on a massive dataset of code, including Apex, making it well-versed in the language’s syntax, conventions, and best practices.
GitHub Copilot also seamlessly integrates with Salesforce Extensions for Visual Studio Code, providing context-aware suggestions and autocompletion directly within your development environment. It can significantly boost your productivity by reducing the amount of boilerplate code you need to write and helping you discover new ways to solve problems.
Test Class
With the recent updates to Einstein for Developers, this tool is now the clear winner in terms of generating Apex test classes. When in VS code, simply click on the name of the class you need a Test Class written for, ⌘P on Mac or Ctrl+P on Windows, and select Einstein Generate Test Class.
Lightning Web Components and Aura Components
For Lightning Web Components (LWC) and Aura Components, GitHub Copilot is going to be the winner. As mentioned previously, it’s designed specifically for code generation and completion, making it the most suitable tool for writing code-heavy components like Aura and LWC.
This is where Github Copilot comes into its own, as it’s trained on such a vast dataset of code (which isn’t specific to Apex), it’ll handle writing JavaScript, CSS, and HTML without a problem.
Interestingly, the recent updates with Einstein for Developers give some enhanced capabilities for writing Aura and LWC, with the addition of inline auto-completion giving you recommendations while you type your code. It’s some way off GitHub Copilot, but the intention is clear for where Salesforce want the tool to get to.
Formula Fields and Validation Rules
For Formula Fields and Validation Rules in Salesforce, ChatGPT will be the best tool to use. You can explain your validation or formula requirements in plain English, and ChatGPT can help you translate that into the correct syntax and structure. You can also ask ChatGPT to provide examples of validation rules or formula fields based on your criteria, giving you a starting point or inspiration.
If you’re stuck on a particular issue, ChatGPT can help you troubleshoot by identifying potential errors or suggesting alternative approaches. This is particularly useful if you’re new to an organization and the documentation around Formula Fields and Validation Rules is poor. Simply copy and paste the formula into ChatGPT, and ask it what it does.
Google Gemini and Claude do a decent job with Apex Classes and Triggers, but as they’re general-purpose language models, they may not have as much specific knowledge about Salesforce formulas.
Salesforce Flow
ChatGPT or Google Gemini are currently the better tools to use for creating Salesforce Flows. They can assist in brainstorming and designing the flow’s structure and logic based on your requirements. You can also describe your desired outcome, and they can suggest different flow elements and their connections.
They can also help you generate clear documentation and comments within your flow to make it more understandable and maintainable. However, for any logic errors you need help troubleshooting, ChatGPT will be the better tool as it seems to have a better understanding of Salesforce in this context.
This, however, is very much likely to change in the not-so-distant future. Salesforce have recently put Einstein for Flow into beta, giving you the ability to use natural language prompts directly in Flow Builder to create your Flow.
Summary
In summary, ChatGPT, Gemini, and Claude offer broad capabilities and are excellent for general queries, explanations, and some level of code generation, but lack Salesforce-specific optimizations. Einstein for Devs is highly specialized for Salesforce, offering the best in-platform assistance and integration for Salesforce development tasks, but is confined to the Salesforce ecosystem. GitHub Copilot provides powerful code completion features but isn’t Salesforce-specific, which means it might require more manual adjustments for Salesforce code.
Each tool has its strengths and weaknesses, and the best choice depends on your specific needs, such as the level of Salesforce specialization required and the type of tasks you’re performing.