NemoClaw Knowledge Wiki

Tools that developers use. Prompt Engineer channel

5 min read

https://www.youtube.com/watch?v=c0-DV23Bf64 This video provides a summary and comparison of two recent developer surveys focused on AI tool and tech stack adoption:

  1. The Pragmatic Engineer 2025 Survey: What’s in your tech stack?
  2. Artificial Analysis AI Adoption Survey – H1 2025

Both surveys offer interesting, and sometimes contradictory, insights into current developer trends.

I. Survey Demographics & Scope:

  • Pragmatic Engineer Survey: Received ~3,000 responses from tech professionals (2,997 after cleaning). Roles: Predominantly Senior/Software Engineers (~1,300), followed by Staff+ Engineers and Engineering Managers/Tech Leads. A smaller percentage were C-level, Directors, or ML/AI/Data Science roles. Experience: Mostly experienced developers with 5+ years of professional experience, with the largest group (24.2%) having 6-10 years. Primary Focus: Primarily Backend Development (60.7%), with Frontend/Mobile Development at 16.2%. Company Size: Responses distributed across Tiny (1-50 people), Small (51-200), Mid-sized (201-1,000), Large (1,000-10,000), and Huge (10K+ people) companies.
  • Artificial Analysis Survey: Collected responses from 1,000+ developers, product managers, and executives. Roles: Majorly Engineering & Technical roles (450+ respondents). Company Size: More inclined towards Large enterprises and Individuals.

II. AI Tool Adoption & Popularity:

  • Overall AI Tool Usage (Pragmatic Engineer): A significant 85.3% of respondents mentioned using at least one AI tool. Only 4.3% reported not using AI tools (due to reasons like not finding them useful, company policies, or ethical concerns).
  • Most Mentioned AI Tools (Pragmatic Engineer): GitHub Copilot (surprisingly high, given last year ChatGPT was more mentioned) ChatGPT Cursor Claude Gemini Note: Gemini’s usage was lower than expected, appearing somewhat static compared to its model releases.
  • Demand for AI Coding Tools (Artificial Analysis): GitHub Copilot (53%) Cursor (49%) Claude Code (29%) Gemini Code Assist (25%) This shows a very similar trend for the top coding tools across both surveys.
  • Year-over-Year Trends (Pragmatic Engineer, 2024 vs. 2025): GitHub Copilot: Mentions increased (but overall usage is lower than last year’s mentions, which the speaker notes is confusing). ChatGPT: Significant decrease in mentions. Cursor & Claude: Showed an upward trend in mentions. Gemini & Perplexity: Appear relatively static or slightly downward. Zed & Windsurf: Show an upward trend from a very low base.

III. AI Tool Usage by Company Size (Pragmatic Engineer):

  • GitHub Copilot: More widely used in medium to huge companies.
  • ChatGPT, Cursor, Claude: More popular in tiny, small, and medium-sized companies.
  • Gemini: Usage appears consistently low across all company sizes, leading to the speculation that company size is irrelevant for its adoption. Hypotheses for Gemini’s flat adoption: Bundled with Google Workspace, so users might access it without actively seeking it. Usage could be higher among Android developers (as Android Studio often has specific Google integrations).
  • Vendor Lock-in: The speaker suggests vendor lock-in might play a critical role in larger enterprises favouring specific tools, while smaller companies are more nimble and experiment with alternatives.

IV. General Tech Stack & Tools:

  • Programming Languages (Pragmatic Engineer): Most Used: TypeScript, Python, JavaScript, Java. Most Loved: Python (expected, due to AI/ML popularity), TypeScript, Ruby on Rails + Ruby, JavaScript, Java.
  • Most-Loved Tools (excluding AI coding tools, from Pragmatic Engineer): VS Code JetBrains IDEs (IntelliJ, PyCharm & others) Cursor Linear Neovim & Vim GitHub Copilot ChatGPT Claude Zed Expo Docker Emacs
  • Most-Disliked Tools (Pragmatic Engineer): JIRA Microsoft Teams Confluence Jenkins Azure AWS Bitbucket Xcode GitHub Actions Windows Reasons for Dislike: Slow, Complex/Complicated, Buggy, Bloated, Painful, Annoying, Cumbersome, Expensive, Confusing, Lacks Features.
  • Version Control: Overall: GitHub (49.6%), GitLab (17.1%), Bitbucket (10.4%). Among those mentioning a Git vendor: GitHub (62.6%), GitLab (22.8%), Bitbucket (14.5%).
  • CD: GitHub Actions and Jenkins are the top two.
  • Cloud Providers: AWS (54.7%) leads, followed by Azure (19.6%) and GCP (17.5%).

V. Language Model & Inference Provider Insights (Artificial Analysis):

  • Demand for LLM Families: OpenAI (GPT/o) (84%) Google Gemini (80%) Anthropic Claude (67%) DeepSeek (53%) Meta Llama (42%) The average number of LLM families used/considered increased from ~2.8 in 2024 to ~4.7 in 2025, indicating increased diversification.
  • Model Market Share Change (over the past year): Google Gemini: Huge surge from 31% to 80% (+49%). OpenAI: Maintained its lead (+1%). DeepSeek: Surged (+53%). Anthropic Claude: Gained share (+21%). xAI Grok: Substantial increase (+31%). Meta Llama & Mistral: Experienced a fall in demand.
  • Demand for Inference Providers: First-party APIs from model labs and hyperscalers lead. OpenAI (80%) Google (68%) Anthropic (50%) Groq (42%) - surprisingly high, given its recent emergence. Azure & Amazon: Their share fell in inference services, while Groq and Cerebras gained share.

VI. Key Takeaways from Artificial Analysis Report:

  1. AI in Production: 45% are using AI in production, while an additional 50% are prototyping or exploring uses with AI.
  2. Engineering & R&D: Is the clear frontrunner use case (66% considering AI for this purpose).
  3. Model Share Shift: Google, xAI, DeepSeek gain share, while Meta and Mistral lose share.
  4. Diversification of AI Use: Companies are increasingly diversifying their AI use, with the average number of LLMs used/considered increasing from ~2.8 in 2024 to ~4.7 in 2025.
  5. Build vs. Buy: 32% favor building, 27% buying, and 25% a hybrid approach for AI solutions.
  6. Openness to Chinese Models: 55% would be willing to use LLMs from China-based AI labs, if hosted outside of China.

Conclusion: The surveys highlight a dynamic landscape in AI tool and model adoption among developers. GitHub Copilot remains a dominant force for coding assistance, while the broader LLM market sees significant shifts, with Google Gemini experiencing remarkable growth. Companies are diversifying their AI model usage, and there’s a growing openness to models from regions like China. The findings also underscore common frustrations with traditional project management and collaboration tools.

Graph View