The AI Sniper Workflow

January 2026

Most B2B outreach today is a volume game: buy a massive list, plug it into an automated sequencer, and hope for a 1% reply rate. While this works for scale, it often fails for high-value deals where context and accuracy matter more than numbers.

I've been using a CLI-based AI agent to handle my outreach. It’s not a "bot" that blasts emails; it’s a research partner that helps me verify, draft, and send emails one-by-one from my own inbox.

The Workflow: Research, "Jam," and Send

Here is how this interactive workflow looks, using a B2B SaaS company—let’s say a PropTech CRM called "EstateFlow"—as an example.

1. Real-Time Verification

Instead of relying on a static CSV, I have the CLI verify the target live.

  • The Prompt: "Read row 10 (EstateFlow). Find their Head of Sales. Check LinkedIn to see if they recently mentioned expanding their regional teams."
  • The Result: The CLI finds that Sarah K. is the Head of Sales and she recently spoke at a conference about "The Future of Digital Real Estate in India." I now have a specific, relevant hook.

2. Contextual "Jamming"

I don't use a fixed template. I "jam" with the AI to refine the message.

  • Drafting: I tell the AI: "Sarah spoke about digital real estate. Pitch her our localized UGC video service for property walk-throughs. Make it professional but direct."
  • Refining: The AI drafts a long message. I tell it: "Too wordy. Cut the intro. Focus on the fact that Indian buyers trust UGC 3x more than professional shoots."
  • Final Result: A 3-sentence email that hits exactly what Sarah cares about.

3. Human-Triggered Execution

No email leaves the terminal without an explicit "Send" command. I review the final payload in the terminal window, and once I'm satisfied, I trigger the send.


Technical Setup: Connecting the CLI to Google Workspace

If you use Google Workspace (business email), you can't just use your password in a script. You need an App Password.

Step 1: Google Admin Settings

As the administrator, you must allow these permissions:

  1. Go to admin.google.com -> Security -> Authentication -> 2-Step Verification.
  2. Enable "Allow users to turn on 2-Step Verification."
  3. Search for "Less secure apps" in the Admin Console and ensure users are allowed to manage their access.

Step 2: Generate an App Password

  1. Go to your Google Account Security settings.
  2. Turn ON 2-Step Verification for your account.
  3. Search for "App passwords" in the search bar.
  4. Create a new one named "CLI Outreach" and copy the 16-character code.

Step 3: The Sending Script

The CLI agent writes a Python script that uses smtplib to connect to Gmail's servers. Store your 16-character code in a local .env file (and add it to .gitignore) so your credentials stay on your machine, not in your code.


Is This Better Than a Bulk Tool?

The Pros:

  • Better Deliverability: Since you are sending one-by-one from your actual, warmed-up business inbox, you avoid the "bulk sender" flags that often hit mass-outreach tools.
  • Deep Customization: You can pivot your strategy mid-batch based on what the AI finds on a founder's Twitter or LinkedIn.
  • No Monthly Fees: You aren't paying $100/mo for a sequencer; you're using a simple script.

The Cons (Where it Doesn't Work):

  • Not for Volume: This is not for sending 1,000 emails a day. Google has daily limits (usually 500-2,000), and sending too many too fast will get your account flagged as spam.
  • No Analytics: You won't get "Open Rates" or "Click Tracking" unless you manually insert tracking pixels into your HTML body.
  • Manual Effort: You have to stay in the terminal and approve each send.

This is a High-Touch strategy. Use it for the 20-30 partners that would actually move the needle for your business, not for thousands of cold leads.