Grok Bot Founder Sales Guide: Founder Mode Connects Early Customers and Product Feedback

Grok Bot's Founder Mode is a workflow configuration designed by Simon Lackowski for founders personally running early sales: one chief of staff and eight specialist bots. They prepare customer research, outreach, and meeting materials, then channel customer feedback into product improvements and follow-up communication.

Before hiring their first sales representative, founders still need to personally learn why customers buy, how to price their product, and how to build a repeatable playbook. The bots prepare the context and drafts in advance, while sending, contracts, pricing, and code merges still require human approval.

1. Initial Setup and the 8+1 Architecture: A 24/7 Collaborative Team for Founders

When trying to use AI at work, the most common mistake people make is cramming every request into a single chat window, turning the context into an unmanageable mess. In Founder Mode, the core design principle is that "each bot takes on only one specific job," organized and isolated across four color-coded folders: Pipeline, Customers, Product, and Marketing.

Running continuously on cloud virtual machines, Grok Bots keep working even when your laptop is closed. Every morning when you walk into the office, tasks prepared overnight are categorized and staged in their respective to-do bins. With memory capabilities, the bots can read past emails to understand outreach style, distilling a tone guide for all drafting bots to follow.

Initial Setup: Founder Mode Setup Pack and Tool Integration

The first step in deployment is installing the Founder Mode setup pack, which includes 1 digital Chief of Staff bot, 4 sidebar folders, 8 specialist bots, 16 paused routines, and 1 shared data folder:

• Roughly 20 onboarding questions: The system asks about company business and pitch priorities, founder tone samples, ICP profiles, target accounts, competitor lists, P0 issue definitions, connector choices, and which routines to activate.
• Activated only after schedule confirmation: All routines initially start in a paused state; only after the founder explicitly confirms the preferred schedule does the corresponding routine become active.
• Tool integration cadence: Prioritize connecting tools already in active use—Gmail and Calendar (drafting and read-only permissions only), Deal Tracker CSVs (until business volume outgrows spreadsheets), LinkedIn and X (drafting and read-only permissions only, keeping posting manual), and the call notes tool Granola. CRMs (such as HubSpot or Attio), GitHub for Product Bot engineering integration, Stripe or Mercury for cash reconciliation, and Webhooks like Zapier or Make for real-time automation can all be connected later as needed.

Mapping the Six-Stage Founder-Led Sales Journey

This architecture directly maps to the 6 critical business stages a founding team must navigate before hiring their first dedicated salesperson:

1. Define your ICP: Identify who derives real value from the product and where they are located, clearly distinguishing between the economic buyer paying the bills and the person who feels the pain every day.
2. Find design partners: Translate the ICP into a targeted shortlist of prospective accounts and seek warm paths in across your network and past interactions.
3. Prep for first meetings: Research target accounts in depth beforehand, rehearse communication, and plan key discovery frameworks.
4. Build your first playbook: Refine early face-to-face conversations into a repeatable, standardized process that can eventually be handed off to new sales reps.
5. Run pipeline and forecasting: Actively manage deals and clarify what factors are actually driving revenue growth.
6. Make your product better: Channel front-line customer and product feedback directly into priorities for new feature development.

The entire architecture comprises 1 Chief of Staff and 8 specialist bots:

1. Central Command: Founder Mode (Chief of Staff Bot)

This is the only bot the founder needs to interact with directly on a daily basis. It oversees the entire system:

• 8:00 AM daily: Consolidates overnight outputs from the other 8 bots into a single "Daily Action List."
• Every Monday: Reconciles figures against actual incoming revenue.
• Every Friday: Generates a weekly pipeline progress summary and Win/Loss Digest.
• Monthly: Drafts the monthly investor update letter (requiring founder approval) and proposes ICP refinements based on market feedback.
• Global circuit breaker: Any P0 critical customer issue or urgent bug detected by the system bypasses batch processing and is escalated here immediately.

2. Pipeline Folder: Deal Execution Group

• Sell Bot (Live Sales Meetings and Follow-up): Focuses on active, qualified opportunities. Ahead of meetings, it synthesizes calendar invites, Granola call recordings, and stakeholder maps into a 1-page "Pre-meeting Brief"; within minutes after a meeting ends, it automatically stages meeting notes, targeted follow-up emails in drafts, and suggests CRM stage updates. Core red line: proposals, pricing changes, and contract terms must never be decided autonomously.
• Outreach Bot (Outbound Outreach): Stages each day's outreach batch at 7:00 AM: cold emails arranged on what the author describes as a "5- to 6-day follow-up cadence," stopping after about 8 touches (the source does not provide a touch-by-touch interval schedule), along with paired LinkedIn messages drafted for each prospect. Before running, it checks the Deal Tracker to automatically exclude active deals or contacts marked do-not-contact; warm leads arrive from Social Bot. The entire batch of drafts awaits founder approval and is never auto-sent.

3. Customers Folder: Customer and Insight Group

• Inbox Bot (Support Inbox and Customer Care): Scans the support inbox twice daily, categorizes items from P0 to P4, drafts replies in the founder's voice, and appends everything to the shared feedback log. Borrowing principles from The Mom Test, it strictly separates superficial compliments from real historical behavior. P0 issues and bugs are immediately escalated to Founder Mode. When a requested feature ships, it drafts follow-up notifications for every customer who asked for it.
• Insights Bot (Demand Insights and Clustering): Conducts weekly cluster analysis across feedback logs, external signals, and meeting summaries. It ranks themes based on frequency, accounts and pipeline affected, and suggests ICP adjustments. Only top-ranked themes that clear preset thresholds are forwarded to Product Bot.

4. Product Folder: Product and Engineering Group

• Product Bot (Feature Specs and Engineering Drafts): Upon receiving top-ranked themes, it translates them into a concise technical specification (Spec), creates a GitHub Issue, and coordinates cloud coding agents to open a Draft PR (Pull Request). Core red line: it never merges code; merges always require human review and approval. Once code is merged, it drafts release notes and hands them to Inbox Bot to notify customers and close the loop.
• Stalk Bot (Competitor Radar): Monitors competitor activity using a dedicated research inbox, never the founder's primary email. Each scan summarizes competitor updates and reliably produces two deliverables: a counterpositioning move passed to Sell Bot, and a content angle handed to Content Bot. It never posts publicly and never direct-messages anyone.

5. Marketing Folder: Marketing and Distribution Group

• Content Bot (Content Marketing and Social Assets): Drafts long-form social posts, anonymized case studies, and weekly content calendars in the founder's voice. Distilled proof points are fed back to front-line sales as ammunition. All content must be published manually by the founder.
• Social Bot (Social Pain Point Radar): Scans trends, founder social accounts, and target profiles each weekday morning, drafts responses, and passes engaged users as warm leads to Outreach Bot. Insights Bot also recommends candidates for in-depth interviews by synthesizing recent meetings, power user activity, and evolving customer profiles.

2. Interactive Sandbox: Dual-Loop Flow and Red-Line Defenses

To provide an intuitive look at how these 9 bots collaborate like an orchestra, the interactive sandbox below illustrates the system's two primary loops, the detailed responsibilities and strict red lines of the 9 bots, and the three-way triage logic when a lead enters the system:

3. Interlocking Dual Loops: How a Social Complaint Becomes Production Code

In this system, the true leverage comes not from any individual bot, but from the two interconnected loops formed as data flows between them:

Loop 1: Opportunity Acquisition Loop

1. Social Bot scans for prospects openly discussing pain points, passing warm leads alongside high-value angles identified by Insights Bot over to Outreach Bot;
2. Outreach Bot cross-checks the Deal Tracker to filter out active deals and do-not-contact targets, identifies warm paths across your network, and drafts paired email and LinkedIn messages;
3. Founder Mode presents these in the Daily Action List, staged for sending upon the founder's manual approval;
4. Once a prospect responds, Sell Bot automatically generates pre-meeting briefs beforehand and drafts meeting notes, follow-up emails, and stage update recommendations minutes afterward.

Loop 2: Product Self-Evolution Loop

1. Meeting summaries and customer support emails feed into the shared feedback log;
2. Insights Bot clusters and ranks items weekly by frequency, accounts and pipeline affected, formulating ICP adjustment proposals;
3. Once high-priority themes cross preset thresholds, they are routed to Product Bot, where cloud agents write specs, create GitHub Issues, and open Draft PRs;
4. Code merges require rigorous human review; once merged, Product Bot compiles release notes;
5. Inbox Bot drafts individualized update notifications to all customers who requested the feature, while Content Bot refines real-world case studies to arm the sales front lines.

Across both loops, every node hands work downstream, but every exit is guarded by safety boundaries:

1. Collision filtering at input: Outreach Bot checks the Deal Tracker beforehand, never contacting accounts in active negotiations or marked do-not-contact;
2. Authenticity tagging during processing: Inbox Bot applies principles from The Mom Test to strictly differentiate superficial compliments from past actual behaviors—rather than dismissing unpaid feedback out of hand—keeping product development from being misled by polite chatter;
3. Human oversight at output: All outbound emails, social posts, code merges, investor update drafts, CRM entries, and final invoice confirmations require personal review and approval by the founder.

The Author's Sample Week: From 8 Draft Pairs to Friday Review

On the Monday list, Outreach Bot has prepared 8 pairs of email and LinkedIn drafts: leads that fit the ICP, show clear signals, and have contact paths move to KEEP; those matching the profile but lacking a "why now" signal move to HOLD; active opportunities, existing customers, or do-not-contact entries move to DQ with no outreach drafted. Midweek pre-meeting materials are staged ahead of time, with debriefs and follow-up drafts ready minutes after calls end; unverified claims in security responses are explicitly flagged. Friday synthesizes affected accounts and pipeline, stalled next steps, and commitments for the coming week. These 8 pairs represent the author's illustrative sample, not a default daily quota.

4. Core Prompt Breakdown: The 7:00 AM Monday "Find Design Partners" Routine

In the Founder Mode setup pack, the author provides a prompt for the "Find Design Partners and Lead Scanning" routine that runs automatically every Monday at 7:00 AM. This is a standardized template you can migrate directly into your own agents or workflows.

To help you adapt this for your own business, here is the prompt faithfully presented and its key constraints broken down in detail:

Every Monday at 7:00 AM my local time, find companies matching my ICP (Ideal Customer Profile) that have publicly discussed [specific pain point] or posted [specific job opening] within the past week.

Before the first run, ask me questions to confirm my ICP details (industry, team size, company stage, geography), who the economic buyer paying the bills is, who the daily user experiencing the pain is, what business signals count as valid, and which companies are on an absolute do-not-contact blacklist. Save these answers and enforce them strictly across all subsequent runs.

For every matching target identified:
1. Rank them by relevance and include direct web links to the signal sources with explanations.
2. Cross-check my Deal Tracker, work email, and calendar. Never surface existing customers or active prospects as new leads.
3. Identify warm paths through my past network and explicitly name the mutual connection.
4. Draft a concise initial outreach message or email in my voice: the opening sentence must directly address the specific signal or pain point observed, with no immediate product pitch.

Strictly categorize all screened results into three tiers, explaining each determination in one sentence:
- KEEP (matches ICP, clear signal, viable warm path; generate drafts)
- HOLD (matches ICP but lacks a compelling "why now" timing signal; hold for observation)
- DQ (disqualified: active deal, existing customer, or on do-not-contact list)
Maintain a running log of researched accounts and report only net-new incremental leads each time. If no qualified new leads are found this week, return a single-line summary note.

All generated email drafts must be placed in Gmail Drafts. Never invoke sending APIs directly, never enroll contacts into automated outreach sequences, and never modify my CRM data without my explicit permission.

Five Engineering-Grade Safeguards in This Prompt:

1. Adaptive cold start and persistent memory: Before the first run, ask me questions... Save these answers. This ensures the prompt is not hardcoded, forcing the agent at startup to extract unstructured business insights from the founder's mind and establish a persistent baseline.
2. Grounding outreach in behavioral signals: Blind outreach often defaults to targeting job titles. This prompt strictly dictates that outreach must be anchored in publicly discussed pain points or specific hiring signals; the timing signal is what makes the message worth reading, not the recipient's title.
3. Multi-source cross-deduplication to prevent embarrassing collisions: Cross-check my Deal Tracker, work email, and calendar. Nothing is more mortifying for an early-stage startup than having an automated tool blast a generic cold email to an active enterprise deal. Multi-source data cross-validation is a non-negotiable prerequisite for deploying agents into production workflows.
4. Strict KEEP / HOLD / DQ three-way triage: Prevents the agent from stuffing every borderline lead into your inbox to appear productive. Leads without clear "act now" triggers go to HOLD; those with existing touchpoints are disqualified (DQ) immediately.
5. Explicitly defining human-in-the-loop checkpoints: The strict prohibitions at the prompt's conclusion—no auto-sending, no automated sequences, and no CRM alterations without permission—form the architecture's security perimeter. The bots solely research, draft, and stage; outbound sends, pre-merge code reviews, investor updates, CRM entries, and final invoice confirmations must all be personally approved by the founder.

💡 Tool Migration Tips: The original setup relies on common Western SaaS tools like Gmail, Granola, GitHub, and CSV spreadsheets. If adapting this to alternative enterprise environments:

• Email and calendar -> Lark / Feishu, DingTalk, or WeCom;
• Deal Tracker -> Notion database, Lark Base, or local Excel spreadsheets;
• Meeting notes Granola -> Lark Minutes, Tongyi Tingwu, or a local Whisper transcription pipeline;
• Code hosting GitHub -> GitLab, Gitee, or internal code repositories. You can adopt the same role specialization, signal triggers, draft staging, and human approval gates, verifying that your chosen platforms support the requisite connectors and actions.

5. Five Survival Rules for Founder-Led Sales

At the end of this workflow, the author shares 5 pragmatic yet crucial rules drawn from his journey from front-line sales to startup leadership. These insights transcend the tools themselves and dictate early sales success:

1. Never Auto-Send

The bot's job is to research, draft, and stage. Without approval, never send or publish anything, and never merge code automatically; CRM record updates and final invoice confirmations similarly require sign-off. Every message a customer reads must be personally reviewed by the founder first.

2. Warm Intro Over a Cold Template

When Social Bot spots a relevant connection path, Outreach Bot uses it as the entry point, weaving in genuine conversational context rather than relying on cold outreach templates.

3. Signal Beats Title

Outreach should be triggered by what the account is currently doing and the specific pain points they are experiencing, not because someone holds a VP title. Genuine behavioral signals are what earn attention.

4. Feedback I Give Twice Becomes a Saved Instruction

When reviewing drafts produced by agents, if you find yourself making the same edit twice—whether on terminology, phrasing taboos, or competitor positioning—the underlying system prompt is missing a rule. Don't just fix the draft; immediately have the Chief of Staff bot turn that correction into a permanent Skill or Settings instruction to prevent recurring mistakes.

5. Know What to Keep

Early discovery, pricing negotiations, and the critical call where your first design partner commits to a "yes" must be handled exclusively by the founder.

These challenging face-to-face conversations are your only primary source for discovering who truly values your product and how it should be sold.

When is the right time to make your first sales hire? Not when you feel exhausted by sales, but when the playbook executed by Sell Bot and Insights Bot has matured into a repeatable rhythm, and your own calendar is so packed with meetings that you have become the company's sole bottleneck in closing deals. That is the moment to hand the wheel to a dedicated sales team.

Source Template Links

• Founder Mode setup pack

The link above is provided by the original source. After installation, connecting your own accounts, data, and tools is required; this review did not install the pack or verify post-login template states.