Tone — how your AI answers sound

Pick the voice your AI uses to draft answers. Live in topbar Settings (gear icon) → Tone. Six options today, and the picker survives across sessions — switch once, every subsequent answer follows it until you flip back.

The six tones

Tone	Sounds like	Best for
Default	Direct, confident, neutral. Fast and flexible.	Most interviews, when in doubt
Professional	Polished, measured, corporate vocabulary	Senior IC / management roles, big-firm screens
Concise & Direct	Bullets, short sentences, no filler	Engineering screens where rambling kills you
Highly Technical	Dense with jargon, precise terminology	Distributed systems, infra, ML/AI roles
Casual & Friendly	Warm, conversational, second-person	Startup culture-fit chats, founder calls
STAR (Behavioral)	Forces the answer into Situation → Task → Action → Result	Behavioral rounds ("tell me about a time...")

How tone changes the actual output

Same question — "Tell me about a project you led" — gets noticeably different shapes per tone:

• Default — A 4-5 sentence answer with one concrete example
• Concise & Direct — 3 bullet points, ~40 words total
• STAR — Explicit headers: Situation: / Task: / Action: / Result:
• Highly Technical — Heavy on architecture verbs, system metrics, named technologies
• Casual & Friendly — Opens with "Sure, happy to walk you through that..." vibe
• Professional — Reserved phrasing, hedges where appropriate, no contractions

When to switch mid-session

Most users pick a tone once at session start and leave it. Mid-session switching is rare but useful when:

• A behavioral question lands inside a technical round → flip to STAR for that question, flip back after
• Interviewer asks for a specific format ("can you walk me through that in bullets?") → flip to Concise & Direct for one answer
• Vibe shifts from formal to casual late in the call → match it with Casual & Friendly

The AI applies the current tone to whatever question fires next. Already-streamed answers keep their original tone.

Why so many tones (not just 3)

Earlier versions had only 3 tones (Conservative / Standard / Bold). Live testing showed users wanted format control (STAR), domain control (Technical), and audience control (Professional vs Casual) as separate axes — collapsing them into "Conservative/Standard/Bold" forced bad tradeoffs.

Gotchas

• Tone doesn't override the model. A Haiku-grade model with STAR tone still produces Haiku-grade reasoning. Combine the right tone with the right model — see Model picker.
• Tone doesn't fix factual errors. If the AI doesn't know your background, no tone will dress that up convincingly. Upload your resume first — see Upload a resume.

Still stuck?

Help & Feedback → Contact us with category "How do I…?" if you want a tone for a specific company / role and don't see it.

Video walkthrough

Coming soon.