If you last took stock of your team’s software development tools a few years ago, the list has changed more than you might think. Back then, the big decisions were which editor, which project tracker, and which version control host. Today, GitHub says more than 40% of the pull requests on its platform start with an AI agent’s first draft, up from 17% in March 2026. Your developers’ tools now write code, not just help them manage it.
For an independent software vendor (ISV), that raises the stakes on what you standardize, what you pay for, and what you allow near your source code. This guide pulls together the most current data we could find: the 2026 Stack Overflow Developer Survey (more than 30,000 respondents), the 2025 Stack Overflow Developer Survey (more than 49,000), a January 2026 JetBrains survey of more than 10,000 professional developers, and GitHub’s Octoverse 2026report. Here’s what developers use most in each category and what it means for your stack.
Code editors and integrated development environments
Visual Studio Code is still the most used editor by a wide margin. In the 2025 Stack Overflow survey, 75.9% of respondents said they used it, far ahead of Visual Studio (29%), Notepad++ (27.4%), IntelliJ IDEA (27.1%), and Vim (24.3%). The number to watch is further down the list. Cursor, an editor built around AI assistance, reached 17.9% of all respondents and 19.3% of professional developers, and Claude Code, a command-line coding agent, showed up in the same question at 9.7%.
The takeaway for ISVs is that the editor and the AI assistant are merging. When you pick or approve an integrated development environment (IDE), you’re increasingly picking an AI tool, too, with all the data handling questions that come with it. Language choice is shifting alongside. GitHub reports that TypeScript is now the top language by contributors on its platform, and our look at the highest-paid programming languages and why they pay covers what that means for hiring.
AI coding assistants and agents
This category didn’t exist in our 2020 roundup, and most developers now use it regularly. Among developers who use AI coding assistants or agents, the 2026 Stack Overflow survey found that 73% use them daily. The survey also found that 66% use coding agents and 63% use chatbots. Claude Code was the most used agent at 66%, followed by GitHub Copilot at 59%.
JetBrains measured work use a different way and got a different order. In its January 2026 survey, 29% of developers used GitHub Copilot at work, while Cursor and Claude Code were each at 18%. JetBrains also found that 90% of developers regularly use at least one AI tool at work for coding. GitHub’s own numbers show how quickly the work is shifting from autocomplete to agents. Among Copilot’s monthly active users, 55% now use its agentic features, up from 35% a year earlier.
Don’t read too much into the rankings, because the surveys ask different questions of different groups. The consistent message is that your developers are already using several of these tools, often more than one at a time. GitHub found that 61% of Copilot users who chose a model in September 2026 used more than one.
Version control and code hosting
Git is the default, and the main choice is where you host it. GitHub was the most used collaboration tool in the 2025 Stack Overflow survey at 81.1%, followed by Jira (46.4%), GitLab (35.6%), and Azure DevOps (16.6%). GitHub alone now counts more than 225 million developers and 1.3 billion merged pull requests in its latest reporting year.
For most ISVs, the hosting decision is really a platform decision. Your code host increasingly runs your pipelines, scans your dependencies, and hosts the AI agents that open pull requests. Choose it with that whole bundle in mind, and keep your repositories portable enough that you could move if pricing or terms change.
Continuous integration and continuous delivery
Automated build, test, and release pipelines are no longer optional for a software company of any size. GitHub reports 45.4 billion GitHub Actions jobs in the 12 months ending Sept. 30, 2026, which works out to about 1,440 jobs every second. Other widely used continuous integration and continuous delivery (CI/CD) options include GitLab CI/CD, Jenkins, Azure Pipelines, and CircleCI.
Agents raise the volume. When more of your pull requests start as machine-written drafts, your pipeline becomes the main place you catch problems before they reach customers. That makes test coverage, required reviews, and protected branches worth more than they’ve ever been. Our recent piece on why your build pipeline is now your product’s attack surface explains why pipeline credentials deserve the same care as production ones.
Containers, cloud, and infrastructure as code
Docker is the most used tool in its category. In the 2025 Stack Overflow survey, 71.1% of all respondents and 73.8% of professional developers used it. Package managers came next (npm at 56.8% and Pip at 40.9%), along with the major clouds: Amazon Web Services (AWS) at 43.3%, Microsoft Azure at 26.3%, and Google Cloud at 24.6%. Kubernetes reached 28.5%.
Infrastructure as code tools, such as Terraform and its open source fork OpenTofu, let you define those cloud environments in files your team can review and version like any other code. If you’re standardizing developer machines at the same time, our updated guide to the best Linux distros for developers covers the operating system side.
Testing tools
Test frameworks are where developers remain most cautious about AI. In the 2025 Stack Overflow survey, 44.1% of respondents who answered the question said they don’t plan to use AI for testing code, compared with 28.9% for writing code. Common choices include Playwright, Cypress, and Selenium for browser testing, and language-specific frameworks such as Jest, pytest, and JUnit for unit tests.
That caution is healthy, but it also creates an imbalance. If AI speeds up code writing while testing stays manual, your test suite becomes the bottleneck. Invest in automated tests now, and consider AI help for generating test cases that a person reviews, rather than for deciding what passes.
Observability and monitoring
Production is the area developers least want to hand to AI. In the 2025 survey, 75.8% of respondents said they don’t plan to use AI for deployment and monitoring, and the 2026 survey found that only 20% use AI to deploy, operate, or troubleshoot production systems. Many teams pair a logging and metrics platform with OpenTelemetry, the open standard for collecting traces, metrics, and logs, so they can change vendors without re-instrumenting their code.
For smaller ISVs, the practical goal is simple. You should know about an outage before your customers call you. If you don’t have that today, it belongs ahead of most other tool purchases.
Collaboration and project management
Jira remains the most common project tracker, used by 46.4% of all respondents and 52.1% of professional developers in the 2025 Stack Overflow survey. Confluence (32.8%) and Notion (16.5%) are the main documentation tools, and 34.8% of respondents keep documentation in plain Markdown files. That last figure matters more than it used to. AI agents work best with written context, and the 2026 survey found that developers trust AI output most when it draws on project goals and requirements (85%) and on code and documentation (73%).
Chat tools such as Slack and Microsoft Teams round out the stack. The more of your decisions live in searchable writing instead of meetings, the more useful both your new hires and your AI tools will be.
Security tooling
Dependency and secret scanning have become standard parts of the development workflow. GitHub reports that developers merged more than 30 million Dependabot pull requests (automated dependency updates) in its latest reporting year, about double the previous year’s pace, and its push protection feature ran 5.8 billion scans for exposed secrets. Use of AI-assisted code review and GitHub Advanced Security each grew more than 90% year over year.
Whatever platform you use, your stack should cover software composition analysis (SCA) for open source dependencies, secret scanning, and static application security testing (SAST) for your own code. AI-generated code makes this more pressing, not less. We’ve covered how your AI coding assistant can create security debt at machine speed, and the fix starts with scanning everything that gets merged.
Set rules for AI coding tools before someone asks
Odds are your developers already use AI tools, whether or not you’ve approved them. Most of them share your concerns. In the 2025 Stack Overflow survey, 81% of respondents agreed they have concerns about data security and privacy when using AI agents, and in 2026, 93% said AI output needs source attribution before they’ll trust it. Organizations are starting to respond. Among 2026 respondents, 38% said their organization sets usage limits on AI tools, 28% review spending data, and 25% limit who can use AI.
Start by publishing a short list of approved AI coding tools and the plans or settings your company pays for. Next, decide what code and data those tools may see, and check each vendor’s terms on retention and model training. Then require human review of AI-generated changes before they merge, which your pipeline can enforce. Finally, keep records of those approvals and reviews. If you’re working toward System and Organization Controls 2 (SOC 2) or answering customer security questionnaires, auditors and buyers will want to see how you control changes to your code, and AI tools are now part of that answer.
Review your software development tools before 2027 budgets lock
If you only do one thing this quarter, inventory the tools your developers actually use, including the AI subscriptions they’ve expensed or signed up for on their own. Compare that list with what you’d choose to standardize, and close the gaps before renewals and 2027 budgets are set. The ISVs that get their tool stack under control now will ship faster, spend less on overlapping subscriptions, and have a clean answer when a customer asks how AI touches their code.
