> ## Documentation Index > Fetch the complete documentation index at: https://docs.guhan.ai/llms.txt > Use this file to discover all available pages before exploring further. # Personalization > How Guhan writes a unique message per prospect. Every message Guhan sends is written fresh for that specific prospect. No template variables, no merge tags — actual writing, per person. ## What Guhan reads before writing For each prospect, before drafting: 1. **The prospect's LinkedIn profile** — name, title, current company, headline, past roles 2. **Their recent activity** — the post that triggered them, or their most recent LinkedIn post (if the step is set to fetch it) 3. **Your brand summary** — how you talk, what you sell, what you promise 4. **Your product** — the product they're being pitched 5. **The step context** — is this a first invite? A day-7 follow-up? A response to their reply? ## Your brand voice controls the tone The **Brand & Voice** page is where you tell Guhan how to sound: * **Preferred tone** — professional, casual, friendly, direct, playful, etc. * **Preferred language** — English, Spanish, French, German, etc. * **Phrases you always want** — a signature phrase, a value prop you always mention * **Phrases you never want** — banned words, no-fluff filters, "we're excited to..." off-limits Guhan follows these strictly. Change them once → every future message is written to match. ## Message templates give you starting points Instead of writing a blank brief every step, you can **start from a template**: * Guhan ships 10 default templates covering common cold outreach shapes (cold connect, signal-hook, break-up, follow-up, etc.) * You can clone + customize any template * Templates work across multiple channels — one "Signal hook — funding" template can be sent as a LinkedIn DM, email, or WhatsApp When you pick a template on a step, Guhan uses the template's brief as the intent + still personalizes per prospect. ## The 6-tier personalization ladder Guhan writes with as much personalization as possible, in this order: 1. **Explicit signal** — "Congrats on the Series A" 2. **Recent LinkedIn post** — "Saw your post on attribution" 3. **Research brief hook** — a specific angle Guhan researched 4. **Structured profile facts** — "As Head of RevOps at Acme..." 5. **Company hook** — "Companies in fintech typically..." 6. **ICP-fit prose** — the generic-but-still-warm fallback Higher tiers get used when available. First-touch invites usually land at tier 4-5 (short + focused). Follow-ups drop to tier 3 or higher (Guhan has more context to work with). ## LinkedIn post fetch (opt-in) For DM / email / WhatsApp steps, you can opt in to **fetch the prospect's recent LinkedIn post** at send time. Guhan reads their latest post and uses it as the opener. Cost: 1 credit per fetch (only when the toggle is on). Skipped silently if the prospect has no recent post. ## What Guhan won't do * **Won't invent facts** — if the prospect's profile doesn't mention their company raised, Guhan won't say it did * **Won't overclaim familiarity** — no "I've been following your work for years" unless you literally told it to * **Won't blast the same wording twice** — each message is generated fresh, even if the prospect is enrolled in two Outreach Agents ## What you can override * **Sender identity** — the message signs off as you, from your brand + your role * **Every draft** — you can turn on **Require approval** on any step; nothing sends until you approve the draft in the Inbox ## Related Set up how Guhan should sound. Save reusable message shapes.