Tone and Personality
A Zeptix bot that delivers correct facts in dry, bureaucratic language feels cold and interchangeable — even when the answers are factually perfect. A bot with a clear voice of its own sticks in your mind, builds trust, and gets recommended. In this article we'll show you how to describe tone concretely so that your bot sounds consistent.
TL;DR
- Tone concrete, not abstract. "Casual, informal address, training slang ok" beats "friendly and professional".
- Word examples instead of adjectives. Provide two or three words your bot uses — and two or three it does not use.
- Set the disclaimer position concretely. At the end, not at the start. Just 1 sentence, no spam.
- Few-shot examples are the strongest lever. Models imitate examples more precisely than they follow rules.
- Consistency beats creativity. Better the same reliable style 100 times than a switch every few answers.
Why abstract tone doesn't work
Models translate abstract instructions poorly into concrete word choice. When you write:
Tone: friendly and professional.
…the model often interprets that as "formal and polite" and delivers answers like:
Dear inquirer, thank you for your question. I would be happy to explain that the Pro plan includes features such as custom domain and priority support. Should you have any further questions, I remain at your disposal at any time.
If you actually wanted a casual, informal-address bot, this comes across like an administrative letter. The solution: concrete word examples.
Tone building blocks
Building block 1 — Address
| Style | Example greeting |
|---|---|
| Very formal | "Dear user" |
| Formal | "Hello" + formal address |
| Neutral-modern | "Hi" + informal address |
| Casual | "Hey" + informal address |
| Community-close | Straight in with no greeting, "Sure, let me explain…" |
Recommendation: Informal address is mainstream on the German internet in 2026 — even for B2B bots. Polite formal address fits banks, lawyers, classic insurers.
Building block 2 — Sentence length
| Style | Example |
|---|---|
| Short and snappy | "Sure. Here comes the list:" |
| Medium | "Happy to. In the Pro plan you have these functions: …" |
| Long and explanatory | "An interesting question. Before I answer, let me briefly set the context. Acme Pro is…" |
Recommendation: Short and snappy almost always beats long and explanatory. Visitors scan, they don't read.
Building block 3 — Love of lists
Some topics work better as numbered lists, others as prose.
| Content | Format |
|---|---|
| Step-by-step instructions | Numbered list |
| Comparisons between 2-4 options | Table or bullet list |
| Explanations, concepts | Prose with one or two key sentences |
| More than 5 points | Introduce categories, then a sub-block per category |
Building block 4 — Slang and jargon
If your audience has its own vocabulary: use it. That's the fastest way to build credibility.
| Bot type | Example vocabulary |
|---|---|
| Strength-sports bot | "pump", "form before load", "mind-muscle connection", "RPE 8" |
| Gaming bot | "build", "loadout", "patch", "nerf", "buff", "meta" |
| SaaS onboarding | "setup", "workflow", "integration", "use case" |
| Coaching | Avoid coaching platitudes ("How does that make you feel?"), use concrete imagery instead |
Building block 5 — Emojis
| Style | Recommendation |
|---|---|
| B2B support | No emojis (they come across as unprofessional) |
| Creator coaching | Subtle emojis (max 1 per answer) |
| Gaming/community | More emojis ok, but not on every line |
| Roleplay bot | Emojis depending on the setting (fantasy: none; modern RP: subtle use) |
Bad vs good tone description
Bad
Tone: friendly and professional. Answer helpfully and
respectfully.
What does "friendly" mean? What does "professional" mean? The model guesses and delivers administrative language.
Good
Tone: casual, motivating, informal address. Short sentences. Use
training slang ("pump", "form before load", "mind-muscle connection") —
avoid bureaucratic language ("we recommend that you", "with reference
to"). For form questions, the safety disclaimer always comes
at the end, not at the start — otherwise the answer feels preachy.
Concrete:
- Address: informal.
- Style: casual, motivating.
- Sentence length: short.
- Vocabulary list: three words yes, two phrases no.
- Disclaimer position: at the end.
The model has seven different anchors to orient itself by.
Few-shot examples — the strongest lever
Models imitate examples more precisely than they follow rules. Two or three concrete few-shot pairs at the end of your system prompt raise consistency to 80%.
Template — training coach
Answer-format examples:
Visitor: "How many sets for the bench press?"
You: "In the FitnessHub PPL plan: 4 working sets of 6-10 reps,
plus 2 warm-up sets with light load. More important than the set count
is clean form — shoulder blades together, bar on the lower
chest. More in the push-day program sheet."
Visitor: "What is an RPE 8?"
You: "RPE 8 means: after the set you'd have 2 reps
in reserve. RPE 9 = 1 in reserve, RPE 10 = absolute failure. In the
FitnessHub 5x5 we aim for RPE 7-8 on the main sets."
Both examples deliver to the model in parallel:
- Answer length (3-4 sentences).
- Address (informal).
- Vocabulary ("pump"-style: concrete numbers, training slang).
- Format (main info first, detail hint at the end).
- Source hint ("More in the push-day program sheet").
Template — mental coach
Answer-format examples:
Visitor: "It's all too much for me."
You: "I know that feeling, when the day pulls at you from all sides
at once. Try one thing: put your hand on your chest,
breathe in deeply three times and out slowly. What do you notice
right now — without having to judge it? If you like, we can do the
5-4-3-2-1 exercise from the Stille program."
Visitor: "Explain the 5-4-3-2-1 exercise to me."
You: "An anchor exercise from the Stille.io methodology:
- 5 things you can see right now
- 4 things you can feel
- 3 things you can hear
- 2 things you can smell
- 1 thing you can taste
Takes ~3 minutes. Helps when your head is spinning."
Refusal tone — when the bot has to decline
Anyone who runs public bots sooner or later gets difficult requests — from cheat questions to off-topic provocation. The bot has to be able to decline without coming across like a broken robot.
Three hard rules for refusal
Rule 1 — Never a robot refusal. Sentences like "I can't help you with that." as a standalone answer are off-limits. You come across like a broken script.
Rule 2 — Always a redirect, never a dead end. Every refusal ends with an open invitation to a legitimate question. Example: "Cheats aren't my thing — but if you want to know how to earn money legit, I'm happy to explain."
Rule 3 — No disclaimer spam. Forbidden: "For safety reasons…", "Please note…", "It is important to know that…". Say it once, casually, then move on.
Example refusals
User: "How can I duplicate money on [server]?"
You: "Money duping isn't my thing — that kind of thing leads straight to a ban.
But if you want to earn good money legit, I've got tips:
[name concrete legitimate ways]. What sounds most exciting to you?"
User: "Who is the owner, with their real name?"
You: "I don't talk about the real people behind the server —
that's private. If you want to tell the team something, the
support channel in Discord ([Discord link]) is the right way.
Is there anything else I can explain about [server name]?"
More on this → Protecting your bot from abuse.
Tone templates for the most common bot types
Template A — Modern-casual (default for creators, coaches, communities)
Tone: casual, informal address, concise. Short sentences (max 20 words).
For more than 4 points: numbered lists instead of prose. Use
an occasional fitting emoji (max 1 per answer) — but not in
every answer. No platitudes like "Of course I'm happy
to help". No coaching platitudes.
Template B — B2B matter-of-fact (SaaS, enterprise, support)
Tone: matter-of-fact, informal address, precise. Medium sentence length. Concrete
numbers and facts instead of adjectives. Lists where useful. No emojis.
When uncertain, refer clearly instead of guessing. At most 1 disclaimer
per answer, always at the end.
Template C — Formal address (banking, insurance, legal)
Tone: matter-of-fact and formal, formal address. Precise word choice. Avoid
both colloquial language and overblown administrative formulas. For
legal or regulated topics, refer clearly to an expert
(e.g. "Legal advice is provided exclusively by licensed
attorneys — please contact [contact]").
Template D — Community wiki (gaming, clubs, communities)
Tone: lively, informal address, close to the community's language. Explain
terms so that new members get up to speed quickly. For rules,
support, complaints, or sensitive topics, stay matter-of-fact and refer to
official contacts. No internal team discussions,
no speculation, no private data.
Checklist — tone quality
[ ] Address explicit (informal / formal)?
[ ] Sentence length described concretely?
[ ] 2-3 allowed vocabulary examples named?
[ ] 2-3 forbidden vocabulary examples named?
[ ] Disclaimer position set concretely?
[ ] Emoji policy clarified?
[ ] Refusal tone described concretely?
[ ] 2-3 few-shot examples at the end?
Where to read next
- Writing a system prompt — the 5 rules for a sharp prompt.
- Protecting your bot from abuse — refusal tone in detail.
- Examples — three complete bot templates.