Quick read. Claude is thoughtful and will disagree with you; ChatGPT is friendly and agreeable; Gemini is efficient and brief; Grok is irreverent and opinionated; Copilot is professional and task-focused; Perplexity is neutral and citation-led. Pick the temperament that fits the job — then fine-tune it with a system prompt.
The personality matrix
Editorial assessment based on published personality research and industry analyses. Higher warmth/verbosity is not "better" — it is a fit decision.
| AI | Natural tone | Pushback | Warmth | Verbosity | Best for |
|---|---|---|---|---|---|
| Claude | Thoughtful, slightly formal, curious | Will disagree, explains why | High | Medium-high | Analysis, nuanced work |
| ChatGPT | Friendly, confident, service-oriented | Mostly agreeable | Medium | High | General tasks, brainstorming |
| Gemini | Efficient, punchy, corporate | Low | Low-medium | Low | Quick answers, Google tasks |
| Grok | Irreverent, opinionated, edgy | Will provoke | Medium | Variable | Creative, informal |
| Copilot | Professional, task-focused | Very low | Low | Medium | Office tasks |
| Perplexity | Neutral, citation-heavy, journalistic | Very low | Low | Low | Research |
Why tone is a business decision
When you deploy an AI customer service agent, its temperament is part of your brand. A warm, reassuring tone suits consumer support; a precise, formal tone suits legal or financial contexts; an efficient, low-verbosity tone suits internal tools where people want answers, not conversation. One provider famously built a campaign letting users speak to AI "breeds" with distinct personalities — the point being that personality customisation drives performance, it isn't decoration.
The most empathetic AI
For warmth and empathy — acknowledging feeling, softening difficult messages, coaching tone — Claude consistently rates highest, with ChatGPT close behind. Gemini, Copilot and Perplexity are deliberately more functional. If your use case is emotionally sensitive (HR, healthcare front-desk, wellbeing), lead with empathy in your model choice. See the most empathetic AI.
How to change any AI's tone
Tone is controllable. A system prompt — a standing instruction applied to every reply — is the single most effective lever. Drop one of these in front of any major model:
| Want… | System prompt |
|---|---|
| Concise | "Answer in three sentences or fewer. No preamble." |
| Warm | "Be warm and encouraging. Acknowledge the person before answering." |
| Formal | "Use professional, formal British English. No contractions or slang." |
| Direct | "Be blunt and direct. Lead with the answer, skip hedging." |
| Empathetic | "Respond with empathy. Validate the concern before giving guidance." |
For the full guide, see how to change AI tone and AI tone for business.
Decision matrix
| If you want… | Choose | Why |
|---|---|---|
| A thinking partner that challenges you | Claude | Highest pushback + warmth |
| An upbeat all-round assistant | ChatGPT | Friendly, agreeable, versatile |
| Fast, no-fluff answers | Gemini | Lowest verbosity |
| Personality and edge | Grok | Irreverent, opinionated |
| Cited, neutral research | Perplexity | Journalistic, source-led |
What changed in June 2026
- Tone customisation moved from novelty to a standard part of customer-service AI deployment.
- Claude retained its lead on warmth and willingness to push back in published assessments.
- System-prompt steering became reliable enough across all major models to treat tone as a configurable setting, not a fixed trait.
Matching an AI to a customer-facing role? Pair this with the customer service guide and the match engine.