Summary
3 AI agents. Daily prospect lists in my inbox. Hours of manual LinkedIn research handed to automation.
I built three AI agents with Claude Code that research prospects and email me a ready list every morning.
Each agent covers one part of outbound prospecting: professional contacts, companies that may need content management, and potential client companies.
What changed: I stopped searching LinkedIn and building spreadsheets by hand. Now I review the lists, pick suitable prospects, and write personalized outreach.
Key result: A research workflow that runs on schedule, with no manual start each morning.
One prospect spreadsheet took about 6 hours to prepare. Other research tasks were expected to take around 10 hours each.
Every list meant searching LinkedIn, checking profiles, finding target companies, and organizing the results by hand.
The core problem was repetition, the same manual effort list after list.
Each agent serves one prospecting goal, so no agent runs a generic, catch-all search.
| Agent | Goal | Output |
|---|---|---|
| 1. Professional Contact Prospecting | Find people for direct LinkedIn outreach | 50 names with LinkedIn profile URLs |
| 2. Content Management Opportunity Research | Spot companies that may need content management services | Company details ready for assessment |
| 3. Outbound Company Prospecting | Build target lists for outbound campaigns | Potential client companies |
Why separate agents? A narrow objective keeps each list focused and easier to review.
The agents run on schedule and deliver results by email before my workday starts.
Automated: Steps 1 to 4, in under 15 minutes per list. Human: Steps 5 and 6. Review and judgment stay with me.
The agent starts automatically.
It finds contacts or companies that match its objective.
It organizes the findings into a structured list.
The list arrives by email every morning.
I check the profiles and companies, then select suitable prospects.
I send personalized messages through LinkedIn or the right business channel.
A list that took 6 to 10 hours by hand now takes under 15 minutes of automated runtime.
| Metric | Manual workflow | Automated workflow |
|---|---|---|
| Preparing one comparable list | 6 to 10 hours | Under 15 minutes of runtime |
| Initial agent build | Not applicable | About 30 minutes |
| Starting the daily research | Manual effort | Runs automatically |
| Receiving the results | After manual preparation | Email every morning |
| Final review and qualification | Required | Still required |
About 5 h 45 min to 9 h 45 min per comparable list, before review time.
This is the gap between manual preparation and automated runtime. It is a theoretical figure, not a measured net saving.
| Scenario | Manual preparation | Build + runtime | Time difference before review |
|---|---|---|---|
| 6-hour task | 6 hours | Under 45 minutes | Over 5 h 15 min |
| 10-hour task | 10 hours | Under 45 minutes | Over 9 h 15 min |
After the build, each new list costs only my review time.
Net time saved = Manual research time − review time − any other active work
Example: A list that would take 6 hours by hand and 30 minutes to review saves 5.5 hours.
Runtime is not counted as active work because the agent runs on its own.
Next step: Track review time and the share of prospects that meet the outreach criteria to confirm the real savings.
The system shifts my time from collecting data to qualifying prospects and starting conversations.
Human judgment stays in the loop. The agents cut the preparation work so my effort goes into deciding whom to contact.
The biggest win came from spotting a recurring task and putting it on a schedule.
Any repeated research task fits this model when three conditions hold:
I used Claude Code to build three agents that prepare prospect lists and email them to me every morning.
Less time building prospect lists. More time evaluating opportunities and starting the right conversations.
