Skip to content
Stack Overflow Dev Survey 2026

AI Data

How are technologists using AI?
OverviewFull data

2.1 AI usage

AI tool usage

83% use at least one AI tool. Coding assistants (66%) and general chat tools (63%) dominate; agents/workflows reach 26%.

Question

Do you currently use AI tools or AI agents in your role at work?
All Respondents · n = 17,464AI coding assistants or codi…66%General-purpose AI chat tools63%AI agents or automated workf…26%Internal AI tools built by m…18%I don’t use AI tools17%17,464
RespondentsPercent
AI coding assistants or coding agents11,50965.9%
General-purpose AI chat tools10,92062.5%
AI agents or automated workflows4,58226.2%
Internal AI tools built by my company3,11517.8%
I don’t use AI tools3,00317.2%
All Respondents · n = 17,464 View and share

AI tool usage frequency

AI coding assistants have become habitual: 73% of their users use them daily. Internal company tools trail at 48% daily and 21% infrequently.

Question

How often do you use each of these AI tools or AI agents in your role at work?
All Respondents · n = 14,478 · Overall: Daily users - 77%, / Weekly = 18% / Monthly/Infreq. = 5%DailyWeeklyMonthly or InfrequentAI coding assistants or coding…73%20%6%General-purpose AI chat tools66%27%7%AI agents or automated workflo…63%26%10%Internal AI tools built by my…48%30%21%14,478Overall: Daily users - 77%, / Weekly = 18% / Monthly/Infreq. = 5%
DailyWeeklyMonthly or Infrequent
AI coding assistants or coding agents73.0%20.1%6.2%
General-purpose AI chat tools65.6%26.8%7.0%
AI agents or automated workflows63.1%25.5%10.3%
Internal AI tools built by my company47.7%30.2%20.7%
All Respondents · n = 14,478 · Overall: Daily users - 77%, / Weekly = 18% / Monthly/Infreq. = 5% View and share

AI tool usage daily

Daily use is substantial, not incidental: 55% use AI for at least two hours per day, including 31% at four or more hours.

Question

You indicated that you use AI tools and/or agents daily. How much of your workday do you use AI tools to help you work?
All Respondents · n = 10,9264+ hours/day31%1-2 hours/day25%2-4 hours/day24%<1 hour every day12%<1 hour and not every day7%None1%10,926
RespondentsPercent
4+ hours/day3,37830.9%
1-2 hours/day2,76525.3%
2-4 hours/day2,63624.1%
<1 hour every day1,33512.2%
<1 hour and not every day7637.0%
None490.5%
All Respondents · n = 10,926 View and share

Attitudes towards AI tools

Overall sentiment is 62% favorable. It closely tracks career maturity: 53% of early career users (1-5 years exp.) are favorable, compared with 69% of experienced users (16+ years exp.). This is not proof that work experience leads to favorable attitudes, but they are certainly related.

Question

How would you describe your attitude toward using AI tools or agents as part of your work?
All Respondents · n = 14,428Very favorable32%Somewhat favorable30%Neutral15%Somewhat unfavorable15%Very unfavorable7%Not sure1%14,428
RespondentsPercent
Very favorable4,56431.6%
Somewhat favorable4,38530.4%
Neutral2,20515.3%
Somewhat unfavorable2,11014.6%
Very unfavorable9946.9%
Not sure1701.2%
All Respondents · n = 14,428 View and share

2.2 Trust in AI and other trends

Trust in AI tools

Trust is conditional: 48% trust AI when they can easily verify it, while only 7% trust it with important decisions. Even among early career (1-5 years exp.) users, important-decision trust is just 4%.

Question

How much do you trust the output from AI tools or AI agents as part of your workflow?
All Respondents · n = 14,304When I can verify it48%Many tasks, not decisions16%Low-risk tasks only16%Don't trust it for most12%Many tasks and decisions7%Not sure or don't use AI1%14,304
RespondentsPercent
I trust it when I can easily verify the output6,87248.0%
I trust it for many tasks, but not important work decisions2,32916.3%
I trust it for low-risk work tasks only2,32616.3%
I do not trust it for most work tasks1,75412.3%
I trust it for many tasks, including important work decisions9496.6%
Not sure / I do not use AI tools740.5%
All Respondents · n = 14,304 View and share

AI workflow current and expected

Adoption centers on familiar-code generation (69%), debugging/refactoring (64%), straightforward questions (59%), and tests (58%). Communications tasks have the strongest resistance: 46% do not use AI there and do not plan to.

Question

Which parts of your workflow are you currently integrating into AI or using AI agents to accomplish? Will they be more or less integrated over the next year? Please select one for each scenario.
All Respondents · n = 12,547Currently use AI or AI agents for thisExpect to use AI More or AI agents more for thisExpect to use AI or AI agents less for thisDo not use AI or AI agents for this and do not plan toThis is not part of my workflowData, logs and metrics42%38%6%16%13%Complex questions50%33%11%18%4%Straightforward questions59%29%8%17%4%Architecture and design47%28%8%19%12%Dev and test environments28%29%7%29%18%Debugging and refactoring64%36%8%9%3%Deploying and operating20%23%7%36%22%Security issues35%38%7%15%19%Code in a familiar area69%31%8%9%2%Code in an unfamiliar area56%34%11%12%4%Visual and audio assets18%19%5%24%43%Product or UX design30%25%5%15%35%Code and design review41%32%9%23%10%Internal knowledge50%36%6%16%8%Onboarding to a codebase50%35%6%13%12%Documentation51%36%7%16%7%Emails and messages24%21%8%46%11%Writing tests58%35%6%9%10%Tickets and plans28%28%8%31%17%12,547
Currently use AI or AI agents for thisExpect to use AI More or AI agents more for thisExpect to use AI or AI agents less for thisDo not use AI or AI agents for this and do not plan toThis is not part of my workflow
Analyzing data, logs, metrics, or query results41.9%37.5%5.7%16.3%13.0%
Answering complex or ambiguous technical questions50.0%32.7%11.4%17.9%4.4%
Answering straightforward technical questions59.4%29.0%7.5%16.5%4.2%
Architecture or technical design exploration47.4%28.2%8.2%19.2%12.1%
Creating or configuring development/test environments28.1%28.8%6.7%28.7%17.9%
Debugging, troubleshooting, or refactoring code63.8%36.1%8.2%8.8%2.7%
Deploying, operating, or troubleshooting production systems19.9%23.3%7.4%36.0%22.0%
Finding or fixing security issues34.6%37.9%6.6%14.7%19.1%
Generating code in a familiar area69.3%31.2%7.7%9.0%2.3%
Generating code in an unfamiliar area56.0%33.8%11.0%12.3%4.3%
Generating visual, audio, or multimedia elements17.5%18.8%5.3%24.1%42.9%
Product, UX, or visual design30.3%25.4%5.0%15.1%34.6%
Reviewing code, pull requests, or design decisions40.5%32.4%8.7%22.6%10.1%
Searching, summarizing, or interpreting internal knowledge49.7%36.1%5.7%15.8%7.9%
Understanding or onboarding to a new codebase49.7%35.1%5.7%12.9%11.6%
Writing documentation, comments, or knowledge base content51.1%35.9%7.0%16.3%6.7%
Writing emails, messages, summaries, or other communication23.5%20.8%8.2%45.9%10.9%
Writing or improving tests58.1%35.0%5.6%8.5%10.4%
Writing tickets, requirements, plans, or project updates27.8%28.0%7.5%30.5%17.2%
All Respondents · n = 12,547 View and share

Next steps after receiving an AI prompt response

Most developers validate AI output by running it locally (76%), comparing it with the codebase (64%), or checking tests/security (53%). Only 10% use answers as-is.

Question

When you receive an AI-generated answer for a development task, what do you usually do before using it? Select all that apply.
All Respondents · n = 13,162Run it locally77%Compare against codebase con…64%Inspect tests/security impli…53%Read official docs38%Search Stack Overflow12%Use as-is10%Ask a coworker7%I do not use AI tools.2%13,162
RespondentsPercent
Run it locally10,06776.5%
Compare against codebase context8,41163.9%
Inspect tests/security implications6,97753.0%
Read official docs4,99838.0%
Search Stack Overflow1,60112.2%
Use as-is1,2659.6%
Ask a coworker9076.9%
I do not use AI tools.2281.7%
All Respondents · n = 13,162 View and share

How good are AI tools at complex tasks

Only 16% say AI handles complex tasks very well; 39% say “good, but not great.” Daily users are much more positive, while 45% of infrequent users rate performance bad or very poor.

Question

How well do the AI tools or AI agents you use in your development workflow handle complex tasks?
All Respondents · n = 14,341Very well16%Good, not great39%Neither good or bad14%Bad at handling complex tasks16%Very poor8%Don't use AI / don't know7%14,341
RespondentsPercent
Very well at handling complex tasks2,28816.0%
Good, but not great at handling complex tasks5,56438.8%
Neither good or bad at handling complex tasks2,06714.4%
Bad at handling complex tasks2,29916.0%
Very poor at handling complex tasks1,1418.0%
I don't use AI tools for complex tasks / I don't know9826.9%
All Respondents · n = 14,341 View and share

AI tool count

Three-quarters use two to five AI products weekly. Tool proliferation exists, but only 5% use six or more.

Question

In a typical week, about how many distinct AI tools or AI-enabled products do you personally use for work?

Count a tool if you intentionally use its AI capability. This may include chatbots, AI search or answer tools, AI coding assistants, AI features in IDEs, autonomous AI agents, internal AI assistants, local model runtimes, or AI tools built into workplace software.

Do not count AI features you do not knowingly use. Do not count multiple models inside the same product unless you use them through separate tools or interfaces.

All Respondents · n = 11,9190 tools1%1 tool20%2 tools34%3-5 tools41%6-10 tools4%11+ tools1%11,919
RespondentsPercent
0 tools750.6%
1 tool2,35119.7%
2 tools4,09734.4%
3-5 tools4,82340.5%
6-10 tools4503.8%
11+ tools1231.0%
All Respondents · n = 11,919 View and share

2.3 AI agent tools

AI agent data storage tools

Mem0 leads at 20%, but no product exceeds one-fifth of the sample, the agent-memory market remains fragmented.

Question

Have you used any of the following tools for AI agent memory or vector databases in the past year? Do you want to use any of the following in the next 6 months?
Used · n = 1,392Mem020%LangMem15%Memary13%Graphiti12%Supermemory12%Zep11%Letta11%Cognee10%Hyperspell9%1,392
RespondentsPercent
Mem027920.0%
LangMem21215.2%
Memary17612.6%
Graphiti16211.6%
Supermemory16211.6%
Zep15110.8%
Letta15010.8%
Cognee1339.6%
Hyperspell1299.3%
Used · n = 1,392 View and share

AI agent orchestration tools

LangChain leads at 30%, followed by OpenAI Agents SDK at 24%. “Terminal Use” at 19% suggests many developers favor direct, lightweight orchestration over dedicated frameworks.

Question

Have you used any of the following tools for AI agent orchestration or agent frameworks in the past year? Do you want to use any of the following in the next 6 months?
Used · n = 4,400LangChain30%OpenAI Agents SDK24%Terminal Use19%LangGraph19%Llama Stack17%Google ADK14%Hermes Agent10%Vercel AI SDK9%Pydantic AI9%CrewAI6%Semantic Kernel5%AutoGen5%Strands4%DSPy4%Mastra3%Dify3%Smolagents3%Atomic Agents3%Instructor2%Agno2%21st.dev2%GripTape2%4,400
RespondentsPercent
LangChain1,32930.2%
OpenAI Agents SDK1,04823.8%
Terminal Use83218.9%
LangGraph82518.8%
Llama Stack76717.4%
Google ADK61714.0%
Hermes Agent44910.2%
Vercel AI SDK3959.0%
Pydantic AI3908.9%
CrewAI2425.5%
Semantic Kernel2185.0%
AutoGen2004.5%
Strands1784.0%
DSPy1593.6%
Mastra1323.0%
Dify1262.9%
Smolagents1162.6%
Atomic Agents1152.6%
Instructor1062.4%
Agno1052.4%
21st.dev1002.3%
GripTape912.1%
Used · n = 4,400 View and share

AI agent observability and security

Sentry leads at 26%, ahead of LangSmith at 17%. Established monitoring products may have an advantage as AI observability converges with conventional application monitoring.

Question

Have you used any of the following tools for AI agent observability, prompts, or evals in the past year? Do you want to use any of the following in the next 6 months?
Used · n = 2,277Sentry26%LangSmith17%MLflow15%Langfuse15%Datadog LLM12%Weights &amp; Biases11%Arize AI6%Braintrust6%DeepEval6%Respan6%Promptfoo5%Agenta5%Ragas5%Galileo AI5%Portkey5%LangWatch4%Helicone4%Humanloop4%PromptLayer4%Patronus AI4%HoneyHive4%Confident AI4%Traceloop4%Chamber4%Opik4%Future AGI4%Sentrial4%Athina AI3%Lunary3%Moda3%Ashr3%Parea AI3%2,277
RespondentsPercent
Sentry59226.0%
LangSmith39317.3%
MLflow34715.2%
Langfuse34515.2%
Datadog LLM27111.9%
Weights &amp; Biases24410.7%
Arize AI1335.8%
Braintrust1325.8%
DeepEval1315.8%
Respan1275.6%
Promptfoo1215.3%
Agenta1145.0%
Ragas1114.9%
Galileo AI1104.8%
Portkey1094.8%
LangWatch1014.4%
Helicone954.2%
Humanloop934.1%
PromptLayer924.0%
Patronus AI914.0%
HoneyHive904.0%
Confident AI863.8%
Traceloop863.8%
Chamber853.7%
Opik843.7%
Future AGI803.5%
Sentrial793.5%
Athina AI783.4%
Lunary773.4%
Moda763.3%
Ashr703.1%
Parea AI693.0%
Used · n = 2,277 View and share

AI coding agents

Claude Code (66%) and GitHub Copilot (59%) are the clear leaders.

Question

Have you used any of the following coding agents or assistants in the past year? Do you want to use any of the following in the next 6 months?
Used · n = 12,255Claude Code66%GitHub Copilot59%OpenAI Codex30%Cursor26%Google Antigravity16%Gemini Code Assist15%Open Code14%JetBrains AI13%Windsurf7%Warp5%Cline5%Pi4%Amazon Q Developer4%Continue4%Mistral Vibe3%Kilo Code3%Devin3%Google Jules3%Replit Agent2%Tabnine2%Aider2%Codium (Qodo)2%BLACKBOX AI2%Supermaven1%Trae1%Augment Code1%Sourcegraph Cody1%Emdash1%Approxima1%ECA1%Sweep1%Kavia AI1%Syntropy1%12,255
RespondentsPercent
Claude Code8,03065.5%
GitHub Copilot7,19658.7%
OpenAI Codex3,62029.5%
Cursor3,12725.5%
Google Antigravity1,95616.0%
Gemini Code Assist1,78914.6%
Open Code1,66213.6%
JetBrains AI1,56912.8%
Windsurf7966.5%
Warp6415.2%
Cline5514.5%
Pi5354.4%
Amazon Q Developer4763.9%
Continue4663.8%
Mistral Vibe3933.2%
Kilo Code3923.2%
Devin3602.9%
Google Jules3072.5%
Replit Agent2852.3%
Tabnine2572.1%
Aider2562.1%
Codium (Qodo)2061.7%
BLACKBOX AI2021.6%
Supermaven1581.3%
Trae1541.3%
Augment Code1451.2%
Sourcegraph Cody1140.9%
Emdash850.7%
Approxima690.6%
ECA680.6%
Sweep680.6%
Kavia AI630.5%
Syntropy630.5%
Used · n = 12,255 View and share

AI-no-code agents

Lovable leads at 38%, with Replit and v0 near 28%.

Question

Have you used any of the following no-code agent builder tools in the past year? Do you want to use any of the following in the next 6 months?
Used · n = 2,219Lovable38%Replit28%v028%Bolt.new17%Base4411%Langflow8%Stack AI7%Flowise4%Voiceflow4%Create4%Glue4%Rivet4%Botpress4%Relevance AI3%Sparkles3%Wordware3%Marblism3%Pickaxe3%2,219
RespondentsPercent
Lovable85038.3%
Replit62928.3%
v061127.5%
Bolt.new38117.2%
Base4425011.3%
Langflow1747.8%
Stack AI1597.2%
Flowise924.1%
Voiceflow924.1%
Create914.1%
Glue853.8%
Rivet783.5%
Botpress773.5%
Relevance AI723.2%
Sparkles683.1%
Wordware673.0%
Marblism652.9%
Pickaxe612.7%
Used · n = 2,219 View and share

AI agent automation tools

Make (39%) and n8n (33%) lead Zapier (18%). Developer-oriented automation appears to be shifting toward flexible, composable platforms.

Question

Have you used any of the following workflow automation tools in the past year? Do you want to use any of the following in the next 6 months?
Used · n = 4,173Make39%n8n33%Zapier19%Temporal6%Prefect3%Trigger.dev2%Inngest2%Gumloop2%Relay.app2%DBOS2%Activepieces2%Bubble Lab2%Cofia2%4,173
RespondentsPercent
Make1,62538.9%
n8n1,36432.7%
Zapier77118.5%
Temporal2496.0%
Prefect1253.0%
Trigger.dev1002.4%
Inngest962.3%
Gumloop932.2%
Relay.app922.2%
DBOS892.1%
Activepieces822.0%
Bubble Lab791.9%
Cofia691.7%
Used · n = 4,173 View and share

2.4 AI at work

AI impact in different work functions

AI is considered useful or very useful for implementation (75%), debugging (73%), documentation (66%), and review (64%).

Question

When using AI tools or AI agents at work, how useful has AI proven itself for the task or activity? Select one answer for each.
All Respondents · n = 12,734Not applicableNo strong opinionMore impractical than useful, not usefulUsefulVery usefulArchitecture exploration13%19%9%34%25%Communication19%24%23%22%13%Debugging/refactoring5%14%9%38%35%Design17%19%14%31%19%Documentation10%15%10%30%35%Implementation4%13%9%40%34%Review10%17%9%37%27%12,734
Not applicableNo strong opinionMore impractical than useful, not usefulUsefulVery useful
Architecture exploration12.6%18.9%9.4%34.4%24.7%
Communication18.8%23.6%22.6%21.8%13.2%
Debugging/refactoring4.8%13.7%8.8%38.1%34.5%
Design17.2%19.3%14.3%30.7%18.5%
Documentation10.1%14.6%9.6%30.4%35.3%
Implementation3.9%12.5%9.0%40.3%34.3%
Review10.0%16.7%8.8%37.2%27.3%
All Respondents · n = 12,734 View and share

Satisfaction with AI tools

AI most improves satisfaction with debugging (66%), documentation (61%), and implementation (60%). Implementation is also most polarizing: 20% report decreased satisfaction.

Question

When using AI tools or AI agents at work, which of the following have increased or decreased your usual satisfaction with completing the task? Select one answer for each.
All Respondents · n = 12,125Greatly decreasedSomewhat decreasedNo changeSomewhat improvedGreatly improvedArchitecture exploration5%5%36%32%23%Communication10%8%50%19%14%Debugging/refactoring4%6%23%36%30%Design6%7%39%29%19%Documentation5%5%30%28%33%Implementation8%12%20%34%27%Review5%7%33%32%23%12,125
Greatly decreasedSomewhat decreasedNo changeSomewhat improvedGreatly improved
Architecture exploration4.5%4.8%35.8%31.9%22.9%
Communication10.0%7.9%49.5%19.2%13.5%
Debugging/refactoring4.3%6.2%23.4%35.7%30.4%
Design5.8%7.3%39.4%28.7%18.9%
Documentation4.6%5.2%29.6%27.9%32.7%
Implementation8.2%11.7%19.9%33.6%26.5%
Review5.3%6.6%32.8%32.3%23.0%
All Respondents · n = 12,125 View and share

Did you switch AI tools in the past year?

Two-thirds switched tools during the year. Output quality is the largest reason at 25%, well ahead of integration (7%), cost (6%), security (2%), or model access (2%).

Question

In the past year, have you switched from one AI tool to another for work? If yes, what was the primary reason?
All Respondents · n = 13,144No34%Better output quality25%My organization required it11%Better integration7%Yes — lower cost6%Better agent support4%Another reason4%The old tool became limited3%Better security or privacy2%A preferred model2%For a specific feature1%Don't use AI for work1%13,144
RespondentsPercent
No4,42333.7%
Yes — better output quality3,30525.1%
Yes — my organization required or recommended a different tool1,39410.6%
Yes — better integration with my workflow or development environment9757.4%
Yes — lower cost8356.4%
Yes — better support for agents, automation, or coding workflows4873.7%
Yes — another reason (write in)4663.6%
Yes — the previous tool became too expensive, limited, or unavailable4333.3%
Yes — better security, privacy, or compliance2832.2%
Yes — access to a preferred model or provider2802.1%
Yes — for a specific feature1791.4%
Not applicable — I do not use AI tools for work840.6%
All Respondents · n = 13,144 View and share

AI memory retention

54% want personal control of AI memory; only 15% prefer centralized organizational control. Another 14% reject retained memory entirely.

Question

When AI tools remember things about your work, who should control that memory?
All Respondents · n = 12,825Just me54%My organization centrally15%I do not want AI to retain m…14%My team11%Not sure6%12,825
RespondentsPercent
Just me6,87753.6%
My organization centrally1,89114.7%
I do not want AI to retain memory1,82114.2%
My team1,42011.1%
Not sure8166.4%
All Respondents · n = 12,825 View and share

AI at work

Organizational maturity is uneven. AI is optional and individual-led for 30%; only 24% report approved tools and guidance.

Question

Which of the following best describes how AI tools or agents are used at your organization today?
All Respondents · n = 13,167Optional, up to individuals30%Approved tools and guidance24%Encouraged, but varies16%Expected for certain work12%Measured in performance10%We don't use AI at work6%Not sure3%13,167
RespondentsPercent
AI use is mostly optional and left up to individuals3,91529.7%
My organization provides approved AI tools and guidance for how to use them3,15424.0%
Some teams actively encourage AI use, but practices vary across the organization2,09715.9%
My organization expects some employees to use AI tools for certain types of work1,50711.5%
My organization measures or evaluates AI use as part of employee or team performance1,34410.2%
My organization does not currently use AI tools or agents for work8046.1%
Not sure3462.6%
All Respondents · n = 13,167 View and share

AI avoidance

59% intentionally avoid AI for some reason. Skill loss/job replacement (17%) exceeds ethical (15%), privacy/security (13%), environmental (8%), or training concerns (3%).

Question

Do you try to avoid using AI tools or agents at work or school? Select the option below that most accurately reflects your reasoning.
All Respondents · n = 13,857I don't avoid AI41%Losing skills to AI17%Morals/ethics15%Privacy or security13%Environmental concerns8%Other3%Lack of training3%13,857
RespondentsPercent
I do not avoid using AI at work or school5,72841.3%
Concern over losing job skills or training an AI to replace you2,36717.1%
Moral or ethical concerns2,10815.2%
Issues with privacy or security1,80213.0%
Environmental concerns1,1298.2%
Other, please specify:3692.7%
Difficulty or lack of training3542.6%
All Respondents · n = 13,857 View and share

Vetting new AI tools

Half of organizations have standardized on one or a few approved tools, while 30% are decentralized or actively experimenting. Only 4% are building internal tools/models.

Question

Which of the following best describes how your organization currently approaches AI tools or models?
All Respondents · n = 12,637A few approved tools33%Mostly one approved tool19%Teams choose their own17%Trying several before decidi…13%Not really using AI10%Not sure5%Building internally4%12,637
RespondentsPercent
We use a few approved tools or models4,13832.8%
We mostly use one approved tool or model2,34518.6%
Different teams choose their own tools or models2,18617.3%
We are trying multiple tools or models before deciding what to use longer-term1,57612.5%
We are not using AI tools or models in a meaningful way1,1979.5%
Not sure6495.1%
We are building internal tools or models5464.3%
All Respondents · n = 12,637 View and share
See the AI highlights

What’s next?