January 31, 2024
Martha J. Lindeman, Ph.D.
Several reviewers have compared and contrasted Microsoft’s GitHub Copilot and two GitHub Copilot alternatives: Amazon’s AWS CodeWhisperer and the Tabnine Copilot. These were identified by TechTarget as the best copilot alternatives for 2024.
This post compares these three with a free GitHub Copilot alternative, Pieces for Developers. Pieces started as a one-of-a-kind code-snippet manager in 2022 and has grown into and beyond the set of features usually found in enterprise-level copilots.
Definition of a Copilot
As companies release new copilots, it gets more difficult to define the new copilot’s purpose and its expected users. Some are identified as ‘copilots for developers,’ but they use the same trade name for multiple pricing tiers with different features.
An effective copilot for developers should have the following features:
- Real-time code suggestions and explanations, automated code completion as you type, and generation of code from a prose description. The code is adapted to your coding style and current context. Also, users have the ability to set custom context for their conversations based on code snippets, web pages, and entire repositories.
- Integrated with popular IDEs, browsers, and productivity tools, and supporting numerous programming languages while updated to integrate with new tools as they are introduced into developers’ workflows.
- Upgrades with new features and potential for uses beyond testing, AI code review, debugging, and refactoring.
- The option to choose which AI models to use with at least some models completely local to the device.
Feature Comparisons
When looking for a GitHub Copilot alternative, there are four key feature sets to use to compare different copilots: (1) Coding, (2) Security, (3) Workflow, and (4) Vendor. Let's dive into some of these key feature sets to understand what tools like GitHub Copilot have to offer.
Coding Features
Four crucial coding-related features were used to compare the four copilots:
- A unique coding feature of Pieces for Developers is the ability to OCR a screenshot and convert its contents to working code.
- All four copilot alternatives have the ability to explain code. As discussed in the Workflow section, Pieces’ additional features make its explanations more personalized and sometimes deeper than those of any other copilot alternative.
- All four alternatives have the ability to generate code suggestions. CodeWhisperer in particular utilizes Amazon Q, the AI expert assistant that recommends AWS services, and it (a) provides step-by-step instructions for working with Amazon services, (b) troubleshoots Amazon connectivity issues, and (c) helps implement new features in CodeCatalyst.
- Core functionality for a copilot includes support for different languages, IDEs, and models. Pieces for Developers has the most available IDEs and it is the only tool that is integrated with web browsers such as Chrome and Firefox. All Pieces products use the same centralized on-device repository, which makes it easier for developers to move between a variety of integrated tools for a seamless workflow. With Pieces, all workflow activity, conversations, and saved materials persist in the desktop application, browser, and other tools.
Security Features
There are four key security features to compare in a copilot alternative:
- Can the copilot be air-gapped (run without an internet connection): CodeWhisperer and GitHub require internet access to run at all. In contrast, Tabnine and Pieces allow the user to control whether to run locally. Of the four copilot AI alternatives, Pieces for Developers is the only full-featured system that can be run on an air-gapped machine.
- On Row 2, only Tabnine and Pieces have SOC 2 certification.
- How the copilot protects developers against incorporating public code with an inappropriate (e.g., copyleft) license into their code.
- Protection from malicious code when making auto-complete suggestions.
Workflow Features
Workflow is not typically included in the feature set of a developer copilot. However, Pieces for Developers has built its fundamental functionality with a focus on the developer’s workflow. With the goal to minimize the cost of context switching, developers can move between IDEs and browsers while Pieces saves, enriches, and provides retrieval from the developer’s on-device repository.
The two features in the Workflow table distinguish Pieces from all other copilots for developers; in Pieces, writing code is viewed as part of the larger moment-to-moment workflow of the developer. The AI works to anticipate what could be needed later or would be helpful if suggested now.
Pieces allows the developer to maintain a continual train of thought during or in spite of (a) frequent changes between tabs for IDEs and browsers, (b) distractions that break the developer’s train of thought, (c) the need to communicate with other people, (d) the need to retrieve code written in the past, (e) searching to find code written by someone else, and (f) any other distractions that break the train-of-thought.
The developer’s on-device repository contains code and related materials from the previous workflow events. These have been enriched with important metadata such as a title with an explanation, tags, the original source of the code or material (e.g., a website), the people who have received or sent related documents, and more. This enrichment is not available in other copilots.
Vendor Features
As with all products and services, the attentiveness of the company offering the service is equally as important as the offering itself. Several things emphasize the company’s intense focus on remaining close to its customers to learn their needs and satisfy them quickly.
Some copilots are identified as ‘copilots for developers,’ but they use the same trade name for multiple pricing tiers with different features. Also, some copilots offer free plans for special categories of individuals, such as open-source maintainers, teachers, and students.
GitHub Copilot has three pricing tiers for different sets of features: Individual, Business, and Enterprise. Tabnine also has three tiers: a basic code-completion air-gapped free tier, a Pro tier, and an Enterprise tier. CodeWhisper has a limited free tier and a Professional Tier.
In contrast, Pieces for Developers, with the same core features as the other GitHub Copilot competitors and much more, is still 100% free for individuals. There is a private version for Enterprises that will be available publicly later this year.
Pieces for Developers is the only full-featured free alternative to GitHub Copilot.
Conclusion
AWS CodeWhisperer: The Amazon copilot is optimized for AWS APIs including Amazon Elastic Compute Cloud (Amazon EC2), AWS Lambda, and Amazon Simple Storage Service (Amazon S3). It recommends code snippets that meet AWS best practices, and it has single sign-on (SSO) with AWS IAM Identity Center integration. Its high-tier users can receive suggestions and define users for specific repositories.
CodeWhisperer is the best choice if, and only if, the code is being written for the Amazon platform and services. CodeWhisperer does not have the features to be the best choice outside of Amazon-specific work.
Tabnine: The Tabnine copilot is for when no information is to be saved or transmitted and the developer’s environment is a currently supported environment: VS Code, Eclipse, Visual Studio 2022 (Windows), or JetBrains. The future for relating to other developer tools is unclear.
GitHub Copilot: GitHub Copilot is a general developer copilot but does not stand out amongst the competition aside from being available in a very commonly used developer platform. With its recent release of Microsoft’s Copilot, GitHub’s parent company has indicated a focus on large enterprises rather than the concerns and problems of individual developers.
Pieces for Developers: Pieces for Developers was built with the singular goal to improve a developer’s workflow–to help developers be more effective and efficient at their work. With its on-device machine-learning algorithms and blended AI leveraging the user’s choice of LLM, Pieces provides integrated and contextual support at all points of the development process. With a single click Pieces Copilot can improve a snippet’s readability, explain an entire repository, save a useful piece of code, translate code to a preferred language, or generate a contextualized snippet based on your unique project.
The following list provides more details on Pieces’ four unique features.
1. Context-Aware Intelligence: You can generate, iterate, and curate code wherever you are in IDE, browser, or Desktop. Every interaction you have in any module of the Pieces suite helps tailor the responses of your personal AI engine on your device. Retrieval-Augmented Generation (RAG) technology helps make the responses more precisely match your prompts. Also, you can set specific contexts for each conversation, including personal repositories, code snippets, website URLs, and videos.
2. Air-Gapped and On-Device Functionality: Pieces has local AI capabilities for strict enterprise-grade security and compliance requirements. With dynamic LLM runtime options, you can choose whether to run in-cloud or completely on-device.
3. Multimodal Interactions: You can input images of code or text, website URLs, file inputs, and even video, making the code instantly available for question-answering.
4. Team Communications: Pieces Copilot can understand the unique skill sets of your teammates, and it provides Related People suggestions based on the context of your query. You may want to find help with a specific project, do pair programming with an expert in your organization, share materials, or find out lessons learned from others on the team. These abilities can make relationships within a large team more comfortable (especially in remote or hybrid work environments).
Read why developers gave Pieces for Developers five-star reviews.
https://www.producthunt.com/products/pieces-for-developers
In summary, Pieces for Developers is the best alternative to GitHub Copilot and it is 100% free for individuals. Not sure yet which copilot to pick? Check out our article on GitHub Copilot vs ChatGPT vs Tabnine and how they stack up with Pieces.
