- B2B consultants and agencies who want a few qualified conversations every day instead of a giant cold list.
- Founders and operators who need a lightweight outbound system before buying another prospecting tool.
- Codex users who want to turn research, scoring, and outreach prep into reusable skills and automations.
Guides The Rundown AI
Find 5 Extra Prospects a Day With This Agentic Framework
In this guide, you will learn how to build a Codex prospecting system that finds five qualified prospects a day, ranks them, and drafts the next outreach step.
Introduction
In this guide, you will learn how to build a Codex prospecting system that finds five qualified prospects a day, ranks them, and drafts the next outreach step. Instead of a giant spreadsheet you never use, you get a daily report with fit reasons, source links, and a running prospect log.
Step-by-step 6 steps
Instructions
-
Step 1 of 6 Set Up the Project Folder
Create an empty folder and open it in Codex.
Ask Codex to create the files it will maintain every day:
Set up this prospecting project folder: - reports/ - prospects.csv - burned_sources.csv - source_queue.md Do not prospect yet.
Prospecting Project Files Use reports/ for the daily Markdown reports, prospects.csv for accepted prospects, burned_sources.csv for sources that are getting stale, and source_queue.md for fresh places to search next.
Pro tip: Codex will not keep a prospecting project organized automatically unless you tell it exactly which files to maintain.
-
Step 2 of 6 Build the ICP Brief
Before Codex researches anything, have it interview you.
Use this prompt:
Here is my company, offer, and URL: - company: [COMPANY] - offer: [OFFER] - URL: [URL] Interview me about my ICP and sales process before finding prospects. When the interview is done, create a reusable prospecting brief with: - ideal company profile - ideal buyer roles - buying signals - pain points - hard disqualifiers - region or market focus
Prospecting Brief This is where the rest of the system gets sharper. If the ICP is vague, the daily reports will be vague too.
Pro tip: Add your company URL if Codex does not already know your business. It gives the interview better context and saves you from re-explaining the offer.
-
Step 3 of 6 Run a Dry Prospecting Pass
Now run the prospecting workflow once before turning it into a skill.
Use this prompt:
Use my ICP brief to find 5 qualified prospects today. Inspect sources first. Build a longlist. Reject weak matches. Return the best 5. Create a dated report in reports/. Append accepted leads to prospects.csv. Update burned_sources.csv.The best output should explain why each company fits, what public source supports the fit, and what buying signal made the lead worth keeping.
In the recording, Codex used public job posts and company pages to look for RevOps pain. That worked because the offer was a Revenue Operations Sprint, and job posts often reveal CRM, reporting, routing, and handoff problems.
Pro tip: Do the dry run before creating the skill. There is no point automating a list that is noisy or missing the real buying signal.
-
Step 4 of 6 Turn the Finder Into a Skill
Once the dry run looks useful, ask Codex to turn the process into a reusable skill.
Use this:
Create a Daily Prospect Finder skill from the dry run. The skill should: - use the ICP brief - inspect sources before recommending leads - return exactly 5 qualified prospects - reject weak matches - write a dated report in reports/ - append accepted leads to prospects.csv - update burned_sources.csvThen schedule it to run every weekday.
Keep this automation focused. It should find and qualify prospects. It should not draft outreach yet.
-
Step 5 of 6 Rank the Final Five and Draft Outreach
Create a second skill for the follow-up work.
Use this:
Use today's final 5 prospects to prepare the next outreach step. Return: - ranked priority order - why each lead is ranked there - one personalization angle per lead - short outreach for the top 1 to 2 leads - anything that needs human verification Do not invent private facts, recent events, or metrics.
Outreach Draft This second skill turns the report into a next action. You are not just collecting names. You are deciding who to contact first and what to say.
Schedule this after the finder runs. In the recording, the finder ran first, then the ranking and outreach prep ran after it.
-
Step 6 of 6 Add a Source Refresh Automation
Prospecting systems get stale when they keep searching the same public lists.
Add a weekly automation that refreshes the source queue:
Once a week, review source_queue.md, burned_sources.csv, and the current ICP. Find fresh public sources or source categories to test next week. Update source_queue.md. Do not run prospecting in this automation.This is what keeps the system alive. The daily finder searches. The weekly refresher keeps the search surface from getting exhausted.
You're done
That's the whole workflow
Mark it complete to save your progress. Background and answers to common questions follow below.
Keep going
Keep learning with Rundown Pro
A guide shows you one workflow. Pro opens the full course library, so you can learn the whole system behind it.
- Every on-demand course
- Certifications on completion
- Group office hours and member perks
$29/mo, cancel anytime
Overview Back to the steps
Background and context
Who This Is For
What You Will Build
You will build a small prospecting system inside Codex.
The system will:
- define your ICP from an interview
- find five qualified prospects from public sources
- write a dated prospecting report
- keep a running prospects.csv
- track stale or overused sources in burned_sources.csv
- rank the final five and draft short outreach for the best leads
What You'll Need
- The Codex desktop app
- A paid OpenAI account
Going Further
Once the reports are useful, ask Codex to package them into a PDF or email them to you each morning.
If you want to connect this to your sales workflow, keep a human approval step in the middle. Codex can draft outreach or prepare CRM notes, but you should review the lead, the source, and the personalization before anything gets sent.