Grow without hiring. Do more with the team you already have.

This system handles an RFP from inbox to finished draft. It picks up RFPs from Gmail or a shared Drive folder, filters out duplicates, and reads every document to extract the client, deadline, reference number, and each question with its format limits. Each question is checked against your library of past RFP responses, case studies, certifications, and policies, and when it finds a match, it drafts the answer for you, preferring answers from the same industry. Risky topics like pricing, legal, and compliance are flagged for a human, and so is any question with no matching history. Flagged questions go to Slack as a numbered list. Your team replies with the answers, the system logs each one into the RFP record, and it saves them to the library so the same question never needs a human again. The drafts are filled into the client's original document and your own branded template. The full system runs on 6 connected n8n workflows: knowledge base ingestion, email intake, RFP parsing, answer drafting, document assembly, and a Slack feedback loop.

You add search keywords to a Google Sheet, such as “plumbers in Austin”, and the system does the rest. For each pending keyword, it pulls local businesses from Google Maps, then visits each business’s website and checks the homepage and four common contact and about pages for emails. It filters out junk and duplicate email addresses, including businesses already in your list. Valid emails are saved to a lead sheet, businesses with no website go to a separate list, and unreachable sites are logged so nothing disappears silently.

A scheduled sender then works through the list at your desired time. Gemini then writes a short, plain-text email tailored to the lead’s business type, with a varied subject line and a low-pressure close, and Gmail sends it. After each send the lead is marked as emailed, so no one gets contacted twice. You get a notification when a batch finishes and when the list runs out.

You upload your old lead list, and the system goes to work. It cleans the list, sorts leads by service interest, budget, and how long they’ve been cold, and chooses text or email for each one. It sends AI-written messages that mention what each person originally asked about, then follows up every 5 days, up to 5 touches. When a lead replies by text or email, the AI reads the full conversation, replies naturally, and sends the booking link when they’re ready. Opt-outs like “STOP” are honored immediately. Pricing questions, unhappy leads, and requests for a person trigger an alert to your team with the full context. When someone books, the lead is marked as booked and your team gets an email. All messages are logged, so you can see the whole history for every lead.

Drop call recordings into a Google Drive folder organized by client and rep, and the system does the rest. Every 10 minutes it checks for new recordings, transcribes each one with speaker labels (rep vs. customer), and pulls that client’s playbook. The AI then grades the call against that playbook: whether the rep used the required opening, asked the required discovery questions, explained pricing correctly, gave the required disclaimers, and attempted to close. It also scores five sales skills from 0 to 10 and identifies the objections raised and how well the rep handled each one. It flags the outcome (won, lost, or follow-up needed) and, for lost calls, the likely reasons why, with evidence from the transcript. Each call gets an overall score out of 100 (40% script compliance, 60% sales skills) and is saved with strengths and improvements.

Once a week, each rep gets a personal Slack summary of their week with strengths, weak spots, common objections, and one focus for next week, and the manager gets a team report with outcomes, average score, top objections, and coaching opportunities. Managers update each client’s playbook by editing a Google Sheet and marking the row “Need to Update”, with no code needed, and the system syncs it within a minute.

The system checks your inbox every minute. It ignores promotions, newsletters, your own messages, and automated mail like bounces and out-of-office replies, so only real replies reach the AI. It also skips anyone already in your lead list. For each real reply, AI reads the email and pulls out the sender’s name, role, company size, industry, order volume, pain points, decision-making power, and tools they already use. It never guesses: anything not stated stays blank. Then it sorts the reply into one of four outcomes:

• Qualified: the key details are present and the interest is genuine. The lead is added to your tracker and your team gets a Slack alert with the details.
• Needs info: genuine interest, but a key detail is missing. Your team gets a Slack alert showing what’s missing, so a person follows up instead of the lead being rejected.
• Not a fit: a clear mismatch gets a polite decline email.
• Ignored: unsubscribes, flat rejections, vendor pitches, and spam get no reply at all.

Three more workflows keep leads from slipping. One checks the Gmail thread for each lead to see whether your team has replied, and marks them as responded. One counts how many days each lead has been waiting and posts a Slack list of everyone still waiting. A third posts a summary of new leads and leads still pending.

An AI assistant monitors your inbox and answers every incoming question using only your own knowledge base: it searches it first and never fills gaps with guesses. It remembers each person’s earlier messages, matches their tone, and replies as a complete, polite email. When the answer isn’t in the knowledge base, it tells the customer it’s checking with the team, alerts your team in Slack with the question and a link to the email, and logs the question in a sheet. If that customer writes again while waiting, they get a holding reply instead of a guess, and the team gets another nudge. Hostile messages get a firm boundary reply.

Here’s how it learns: your team answers the customer directly from Gmail, then reacts with a ✅ on the Slack alert. The system finds your reply, uses AI to rewrite it into a clean, standalone answer with no names or greetings, and adds it to the knowledge base document. Within the hour, the knowledge base is re-indexed, and the assistant resumes answering that customer. The next person who asks the same question gets the answer automatically, so the questions your team has to handle shrink over time.

You keep a content calendar in Google Sheets. Each run takes the next pending topic and researches it on the web. A random one of five proven copywriting frameworks is chosen, and AI writes a separate post for LinkedIn, Facebook, and Instagram, each in that platform’s voice. AI then designs a custom infographic-style image for each platform in your brand colors, adds your logo, and also turns the Facebook and Instagram images into Stories. Everything is scheduled through Buffer, the calendar is updated, and you get an email with every caption and image link.

If you want a change after it’s scheduled, just text a Telegram bot in plain English: “Rewrite my LinkedIn caption” or “New image for the Instagram post.” The bot works out which platform and which post you mean, regenerates the caption, the image, or both, and updates the scheduled post. It keeps previous versions, so “revert my Facebook caption” rolls it back. If your request is unclear, mixes a revision with a reversion, or doesn’t match a scheduled post, it tells you what to fix. When it finishes, it sends one summary of what changed on each platform.