# Founder Presence

> Personal brand content factory, inbox triage, calendar protection, and the personal-leverage loops that compound fastest.

Founders sell trust as much as product. Your inbox, your DMs, your social presence, your replies — they all telegraph who you are and how you operate. A good founder-presence stack means the right people hear from you, the wrong asks get politely declined, and your public-facing voice keeps shipping while you do the work that actually moves the business. The mistake most operators make: they try to be everywhere themselves. The fix isn't a VA who pretends to be you. The fix is an agent that knows your voice cold and either drafts in it or speaks in it for low-stakes channels, with you in the loop on anything that matters. A few principles: Drafts beat sends. Until your agent has months of voice training, it drafts and you click send. Trust earned, not assumed. Voice is per-channel. Your LinkedIn voice ≠ your DM voice ≠ your email voice. Each gets its own skill. Memory across all conversations is the unlock. The agent that remembers what you told someone six months ago is worth 10x the one that doesn't. Calendar protection is presence. Saying yes to the wrong meeting costs more than saying no the wrong way.

## 1. Personal Brand Content Factory

### Tip 1.1 — Ghostwriting agent in your voice

**What it does:** A content-generation loop built on your platform voice skills (see Marketing 1.4). The agent reads daily signal sources, drafts posts in your specific voice for each platform, and queues them for your morning approval. You spend 15 minutes a day on content and ship 5-7 posts a week. Why it wins: Founders who ship consistently on social compound an audience. Founders who try to write everything themselves ship sporadically. Ghostwriters cost $3-8k/month and never quite get your voice. An agent trained on 100+ of your own posts is closer than any ghostwriter to your real voice, costs cents per draft, and runs daily. Tools: Your platform voice skills (per Marketing Tip 1.4), your morning-pitch skill (Marketing Tip 1.6), an approval rail (Telegram, Slack, or your phone notifications). How to wire it:

## 1. Stand up the morning-pitch skill from the Marketing section so the agent has fresh ideas

every day. 2. For each platform you're active on, build a voice skill from 100+ of your own posts on that platform. 3. Daily cron: agent takes top 3 ideas from morning pitch, drafts one post per platform in the matching voice skill. 4. Sends drafts to your approval rail by 8am. Each draft includes the source idea (so you remember why it's interesting), the hook, the body, and the predicted engagement signal. 5. You approve, edit, or kill. Approved drafts post via Buffer / Typefully / your scheduler. Example prompt to your agent: Every morning at 7am, run morning-pitch from the Marketing section to get today's top 3 ideas. For each idea, pick the platform that fits best (LinkedIn for thinking-outloud, X for sharp takes, Threads for community). Draft one post per idea using the matching platform voice skill. Send all 3 to my Telegram by 8am with the source idea, predicted engagement, and approve/edit/kill buttons. On approve, post via my scheduler. On edit, learn from the edit (log diffs to voice-learnings.md ). On kill, log why so the morning pitch stops surfacing that pattern. Watch out for: Voice skills drift if you don't refresh them. Retrain quarterly with your most recent 100 posts. Agent-drafted posts that take you 10 minutes to edit are not saving time. If you're rewriting 70%+, the voice skill is broken. Don't auto-post. The cost of one bad agent-drafted post outweighs months of saved time. Skill file: voice-skill-template, linkedin-copywriting, content-engine

### Tip 1.2 — Cross-platform repurposing of your high-signal content

**What it does:** Anytime you publish a long-form artifact (podcast, video, essay, talk transcript), your agent automatically chops it into platform-native posts: LinkedIn carousel, X thread, Threads short, YouTube short, newsletter blurb. Each in the right voice for the platform. Why it wins: Most founders publish one thing and let it die. The same idea, repackaged correctly, lands 5-10x the reach. The repackaging is mechanical — pure agent work. Tools: Your platform voice skills, your scheduling tool, transcripts where applicable.

**How to wire it:** 1. Define your "long-form drop" channel: a folder where you put new podcast episodes / essays / video links. 2. Agent watches the folder. New artifact → extracts transcript (Whisper if audio/video). 3. Agent reads the full piece, picks 3-5 of the strongest ideas. 4. For each idea, drafts the platform-native versions: LinkedIn post, X thread, short-form video script, newsletter pull-quote. 5. Stages everything in repurpose/<date>-<artifact-slug>/ . You approve in batch. Example prompt to your agent: Watch /long-form-inbox/. New file or URL → if audio/video, transcribe with Whisper. Read the full piece. Pick 3-5 strongest ideas. For each, draft: a LinkedIn post (using my LinkedIn voice), an X thread (5-8 tweets), a short-form video script (60s), a newsletter pull-quote. Stage in repurpose/<date>/. Ping me when ready, I'll batch-approve. Watch out for: Don't repurpose everything. Some podcasts have one good idea, not five. Force the agent to discard weak ideas. The same idea posted on three platforms within an hour reads as spam. Stagger 24-48h between platforms. Quote attribution: if you're quoting a guest on your podcast, make sure their name lands in the post. Skill file: content-repurposer, youtube-shorts-repurposer

## 2. Inbox Triage

### Tip 2.1 — Priority signal + response drafts

**What it does:** Your agent reads every inbound email, classifies into priority tiers, and for high-priority messages, drafts a response in your voice. You open your inbox to find 90% of the noise filtered and the top 10% with drafts waiting. Why it wins: Inbox is where founder time goes to die. A good triage agent buys you 1-2 hours a day. The compounding effect over a year is the difference between burnout and breathing room. Tools: Gmail or your mail API, your email voice skill, your CRM and calendar for context. How to wire it:

## 1. Define priority tiers: P0 = urgent + important (customers in trouble, time-bound

business asks), P1 = important not urgent (warm intros, partner asks), P2 = routine
(newsletters, recurring), P3 = noise (spam, mass marketing).
1. Hourly cron: agent reads new emails, classifies each.
2. P0 → ping you immediately on Telegram with a one-line summary.
3. P1 → draft a response in your voice, stage in Gmail drafts.
4. P2 → auto-archive to a "review later" label.
5. P3 → auto-trash with reversible rule. Example prompt to your agent: Hourly during work hours: read new emails. Classify each P0-P3. P0: ping me on Telegram with subject, sender, one-line summary, and a draft response. P1: draft a response in my voice and save to Gmail drafts. P2: auto-archive to label reviewlater . P3: auto-trash. Every Friday at 5pm: send me the week's classification stats so I can correct anything you've consistently mis-classified. Watch out for: The first 2 weeks, force the agent to NOT auto-archive anything — just label P2/P3. You correct, it learns. A misclassified P0 is catastrophic. Build a soft floor: anyone in your CRM at "client" status defaults to at least P1. The agent will be tempted to draft long replies. Tell it: "default to short. 3 sentences max unless I tell you otherwise." Skill file: lead-status, email-followups, outreach-drafter

### Tip 2.2 — Snooze and follow-up tracking

**What it does:** Your agent watches emails you've sent and snoozes them: if no reply in N days, it surfaces the thread back to you with a draft nudge. Different N per recipient type — clients get 3 days, prospects get 5, vendors get 7. Why it wins: "I'll follow up Tuesday" is the most common reason deals die. The agent never forgets Tuesday. Tools: Gmail API for read access to your sent folder, your CRM for recipient classification. How to wire it: 1. Per email you send, agent logs: thread ID, recipient, expected response window (based on recipient type or your explicit tag). 2. Cron daily: check each logged thread. If no reply within window, surface it.

## 3. For each surfaced thread, agent drafts a nudge in your voice — short, references the

previous message, opens a small new value. Example prompt to your agent: Watch my Gmail sent folder. For each sent message, log: thread ID, recipient, classification (client / prospect / partner / vendor / personal), expected response window from response-windows.json . Daily at 11am, check threads past their window with no reply. For each, draft a nudge in my voice: short (3 sentences), reference previous message, add small new value (link, observation, question). Stage as a draft reply. Ping me with a daily list. Watch out for: Some replies don't need following up. Tag those with a label and the agent should respect it. The nudge should never sound like "just bumping this." That makes it worse. Force "new value or new specific question." Don't nudge personal contacts on a corporate cadence. Your aunt doesn't need a 5-day follow-up. Skill file: stale-lead-blitz, email-followups

## 3. "Speak As Me" Replies

### Tip 3.1 — High-signal DM replies in your voice

**What it does:** For inbound DMs on platforms you're not heavily active on (LinkedIn DMs, X DMs, Instagram messages), the agent classifies, prioritizes, and either drafts a reply for your approval or — for trusted contact patterns — sends directly in your voice. Why it wins: Most founders ignore DMs because the volume is too much. But buried in the noise is a deal, a recruit, a press hit. The agent reads everything and surfaces only what matters. Tools: Platform APIs or scrapers, your platform-specific voice skills, your CRM. How to wire it: 1. Per platform, define what auto-send is allowed: nothing, or only replies to people in your CRM, or only short acknowledgments. 2. Agent reads all inbound DMs hourly. 3. Classifies: opportunity (deal/intro), engagement (fan/peer), ask (favor), pitch (sales), noise.

4. Opportunities and asks above a threshold → draft for you. 5. Engagement from known contacts → draft a 1-line warm reply. 6. Pitch and noise → auto-archive. 7. Trusted patterns (e.g., "thanks for the message, can you book a 15-min via this link?") can be auto-sent with rate limits. Example prompt to your agent: Hourly: read all inbound DMs across LinkedIn, X, Instagram. Classify each: opportunity / engagement / ask / pitch / noise. For opportunity and ask + sender in CRM: draft a reply in my platform voice and stage. For engagement from known contacts: draft a 1-line warm reply. For pitch and noise: auto-archive. Don't auto-send anything yet. After 30 days of training, propose specific auto-send rules. Watch out for: Voice clones on DMs sound robotic faster than on long-form. Keep them short and conversational. A high-signal DM that gets a generic reply destroys the relationship. Bias toward "human, please review." Some platforms aggressively penalize automated DM activity. Stay well under their thresholds. Skill file: voice-skill-template, outreach-drafter, whatsapp

### Tip 3.2 — Conversation memory across all channels

**What it does:** The agent maintains a unified memory across every channel you talk to people on — email, DMs, WhatsApp, Slack, in-person notes. When anyone messages you, the agent can answer "who is this, what's our history, what did I promise them last time" in 5 seconds. Why it wins: This is the founder presence super-power. The CEO who remembers "you mentioned your daughter's school last time we spoke" is the CEO who builds relationships. Manual memory caps out at ~150 close contacts. With an agent, the cap is your storage budget. Tools: Each channel's archive, an embedding-based search index, your CRM. How to wire it: 1. Index every conversation thread you have across all channels. Per contact, build a rolling summary.

## 2. On every new inbound message, agent loads the contact's full history summary plus the

last 5 specific exchanges. 3. Surfaces a 3-line context note: who they are, what you last talked about, anything you promised or pending. 4. The triage from Tip 2.1 and the DM drafter from Tip 3.1 both load this context before drafting. 5. Quarterly: agent generates a relationship-pulse report — who haven't you spoken to in 90 days that you should, who's gone cold on you, who's heating up. Example prompt to your agent: Index every message across Gmail, LinkedIn, X DM, Instagram DM, WhatsApp, Telegram, Slack. Per contact, maintain a rolling summary at contacts/<slug>.md : who they are, our relationship, what we've discussed, anything pending or promised. On every new inbound message, surface a 3-line context note to me before I read the message. Every Friday: relationship-pulse report — top 10 people I haven't spoken to in 90 days who are warm contacts, top 10 going cold, top 10 heating up. Watch out for: Privacy. Be deliberate about which channels get indexed. Don't index your spouse's messages. "Pending" promises rot fast. The agent should chase you for unkept promises older than 30 days. Cross-channel deduplication. If the same person emailed and DM'd, build one summary, not two. Skill file: lead-status, _auto-memory-search

## 4. Calendar Protection

### Tip 4.1 — Saying no to low-value asks in your voice

**What it does:** For meeting requests that don't clear a value bar, your agent drafts a polite decline — in your voice, specific to why this ask isn't a fit, offering a smaller alternative when appropriate (async note, content link, "ping me in 6 months when X is true"). Why it wins: Saying no is hard. Most founders say yes to too many meetings, then resent it. An agent that drafts the decline reduces the friction to almost zero. The decline can also be more thoughtful than what you'd write under time pressure. Tools: Your email voice skill, a "what's worth my time" rubric, your CRM.

**How to wire it:** 1. Define your "yes" rubric: criteria that justify a meeting (existing client, qualified prospect, strategic partner, fundraising-related, etc.). 2. Default decline: thank-you, brief reason, offer an alternative (a relevant content link, an async question they can email, a future-condition "happy to chat once X"). 3. Agent classifies inbound meeting requests (from Tip 2.1). 4. For requests that fail the rubric, drafts a decline in your voice. 5. You scan and send. Example prompt to your agent: Watch for meeting requests in my email. For each, check against yes-rubric.md : existing client / qualified prospect / strategic partner / fundraising / press / strong personal warmth. If it clears, route to my normal triage. If it doesn't, draft a polite decline in my voice: warm thank-you, brief reason (don't lie — find a real one), and one alternative (content link, async question, future-condition). Stage in drafts. Watch out for: Don't auto-send declines. People remember a clumsy no. "I'm too busy" is the worst possible decline reason. Force the agent to find a real, specific one. Some declines are openers. "Not now, but here's a content piece on this" can become a relationship. Skill file: email-followups, outreach-drafter

### Tip 4.2 — Calendar audit + buyback

**What it does:** Weekly, your agent audits your calendar: time spent in meetings, time spent on deep work blocks, recurring meetings that have outlived their purpose. Flags candidates to cancel, shorten, or move to async. Why it wins: Most calendars accrue meetings the way attics accrue stuff. Nobody questions a recurring meeting until they hate it. The agent questions it weekly. Tools: Your calendar API, optionally your meeting transcripts for the "was this useful" signal. How to wire it: 1. Weekly Sunday cron: agent pulls last week's calendar. 2. Computes: % time in meetings, longest deep work block, recurring meetings that ran > 30 min over their average length, recurring meetings you've shifted/rescheduled twice

(= you don't want to be there). 3. Output: 5-line report + 3 candidate calendar moves for next week ("cancel X, shorten Y to 30 min, move Z to async with a Loom"). 4. You approve. Agent sends the polite reschedule/cancel notes. Example prompt to your agent: Every Sunday at 7pm: audit last week's calendar. Compute % time in meetings, longest deep work block, recurring meetings that ran long, recurring meetings I've rescheduled twice. Output: 5-line summary plus 3 candidate moves for next week (cancel / shorten / async). For each candidate, draft the polite note to the other attendees. Stage everything. On approval, send. Watch out for: Some recurring meetings exist for political reasons. The agent shouldn't propose canceling those. Async replacements are great until they're not. Don't push everything async — some things genuinely need a face. Weekly retros are the easiest recurring meeting to bloat. They're also the easiest to fix with a written template + 15-min sync. Skill file: calendar-event, meeting-prep

### How it all stacks

Founder presence is the layer between your time and the world's demands on it. Get the
inbox triage right and you reclaim hours. Get the voice skills right and your output sounds
like you while you're elsewhere. Get the memory layer right and every interaction starts five
steps ahead.
Install order:
1. Inbox triage (Tip 2.1). Highest immediate ROI. Reclaims hours the day you turn it on.
2. Snooze and follow-ups (Tip 2.2). Pair with 2.1. Catches the threads you'd forget.
3. Email voice skill (Marketing 1.4 pattern, applied to email). Without it, drafts sound generic.
4. Conversation memory (Tip 3.2). Compounds value over months as the corpus grows.
5. Ghostwriting + repurposing (Tip 1.1, 1.2). Install once your platform voice skills are solid.
6. DM autopilot (Tip 3.1). Install only after inbox is dialed.

7. Calendar audit (Tip 4.2) and decline drafter (Tip 4.1). Once you have data on your calendar to audit. The compounding piece is voice + memory. Both get better the longer they run. Start them early even if you only use a fraction of the output.

### Learning & R&D