On 3 May 2025 I asked my tech network on X:

“If you work in tech, do you use AI in your job?
Please vote and (if you can) reply with the tools you use.”

682 people voted.

Role & Usage Votes % of all voters
Developer — uses AI 572 83.9 %
Manager — uses AI 69 10.1 %
Developer — doesn’t use AI 28 4.1 %
Manager — doesn’t use AI 13 1.9 %

# Within each role

  • Developers using AI: 572 / 600 ⇒ 95 % (≈ 9 in 10)
  • Engineering managers using AI: 69 / 82 ⇒ 84 % (≈ 8 in 10)
Pie chart of poll results
Figure 1 — Visual snapshot of the vote split.

# The big picture

AI is no longer a “nice to have”. Nine out of ten engineers and eight out of ten engineering managers already rely on it daily. Only 6 % of voters remain AI‑free.

# What tools are hot

From 20 + replies, these names surfaced again and again:

Category Popular tools mentioned
Coding assistants GitHub Copilot, Cursor, Claude 3.5
General chat LLMs ChatGPT, Gemini, Grok
Low‑/no‑code & agents Flutterflow, Retool, n8n + AI agents
Docs & meeting notes GleanChat, NotebookLM
Code review & PR bots GitHub bots, Copilot‑for‑Emacs
Research / brainstorming ChatGPT Advanced Data, Gemini

# How people actually use AI

  • Rapid POCs and prototyping
  • Explaining unfamiliar code bases
  • Auto‑writing unit tests & configs
  • Drafting & polishing documents and RFCs
  • Summarising meetings and logs
  • Brainstorming solutions before coding
  • “Controlled” self‑paced learning (ask → dig → zoom‑out)

# Why a few still hold back

  1. Trust — fear of hallucinations or leaking code.
  2. Habit — they feel “fast enough” with their current workflow.

Both shrink once teams add guard‑rails (private models, review gates) and run small experiments.

# Two favourite real‑world use‑cases

# Engineering‑manager reply

1️⃣ Automating daily workflows in n8n (AI Agents + MCP)
2️⃣ Summarising any thread they’re tagged in
3️⃣ Polishing write‑ups
4️⃣ “Controlled learning” at their own pace
5️⃣ Vibe‑coding MVPs for fun

# Engineer reply

“Copilot writes the dull boilerplate, ChatGPT explains legacy code. Feels like I got a junior dev + a tutor on demand.”

These show why adoption is so high: AI erases drudge work and unlocks rapid learning — whether you manage people or write code.

# My take — opinion

I think AI will soon fade into the background, like IDE auto‑complete did years ago. The winning teams will be those that treat AI as a junior teammate — review its output, feed it context, let it handle the boring 80 %, and reserve human focus for the tricky 20 %.

# Key takeaway

If you’re still on the fence, start small:

  1. Pick one boring task (e.g. commit messages).
  2. Test a focused tool (Cursor or Copilot) for a week.
  3. Keep only what measurably saves time.

Chances are, you’ll join the 90 % before month‑end.

Got a different experience? Drop a comment below or ping me on X — I’d love to hear it!