What I Actually Did
I didn't fire anyone. I had a 5-person team doing work that kept us stuck at ₦3.2M/month revenue. Three of them were spending 70% of their time on tasks that didn't require human judgment — responding to initial inquiries, collecting client information, formatting reports, scheduling calls, and sending follow-up emails.
I built AI agents to handle those tasks. The three team members didn't lose their jobs — they shifted to exclusively high-value work: closing deals, managing complex client relationships, and building strategy. The result: revenue went from ₦3.2M to ₦16M/month in 37 days because the team could suddenly handle 5x the clients.
Agent 1: Lead Qualification Agent
Before: A team member spent 3 hours/day responding to website inquiries, asking qualifying questions, and deciding which leads to pursue. They handled about 15 leads/day.
After: An n8n workflow + Groq agent that:
- Receives the inquiry via webhook
- Sends a personalised email within 2 minutes asking 4 qualifying questions
- Analyses the response using a scoring rubric
- Scores the lead (Hot/Warm/Cold) and adds to the CRM
- For Hot leads: automatically schedules a call and notifies the sales team
- For Warm leads: adds to a nurture sequence
- For Cold leads: sends a polite "not right now" email
Result: 80+ leads processed per day with zero human time. Hot lead response time dropped from 4 hours to 2 minutes. Close rate on Hot leads increased 22% because speed of response matters.
Agent 2: Client Onboarding Agent
Before: A team member spent 5 hours/onboarding collecting business info, setting up accounts, configuring dashboards, and sending welcome materials. Average onboarding: 4 days.
After: An n8n workflow that:
- Sends a Typeform link when a deal is closed
- Collects all business information in one form
- Creates their Supabase account and configures permissions
- Generates their dashboard from a template
- Sends the welcome packet via email
- Schedules the kick-off call
- Sends a Slack notification to the delivery team
Result: Onboarding time dropped from 4 days to 45 minutes of actual human review. Client satisfaction on the onboarding experience went from 3.2/5 to 4.7/5 because nothing gets missed.
Agent 3: Report Generation Agent
Before: A team member spent 6 hours/week pulling data from 4 different sources, formatting it into a report, and emailing it to clients. Each report was slightly different in quality.
After: A scheduled n8n workflow that:
- Pulls data from Supabase, Google Analytics, and social media APIs every Friday at 5pm
- Sends raw data to Groq with a report template
- Groq writes the narrative analysis
- n8n formats it into a branded PDF
- Emails it to the client automatically
Result: Reports are consistent, never late, and actually better because the AI catches patterns humans miss when they're rushing. Zero human hours per week.
The 37-Day Revenue Jump
Days 1–7: Built and tested the three agents. Ran them in parallel with the human processes to verify accuracy.
Days 8–14: Switched fully to agent-driven processes. Freed 70% of three team members' time.
Days 15–30: Redirected that capacity to sales and client growth. The team went from handling 12 active clients to handling 45.
Days 31–37: Revenue caught up to the new capacity. ₦3.2M → ₦16M.
The agents didn't create the revenue. The humans created the revenue. The agents just removed the ceiling on how much human capacity was available for revenue-generating work.
What I'd Tell Someone Building This
Start with the most painful, repetitive task your team does. Not the most complex one — the one that causes the most groaning when it comes up. Build one agent for that. Get it running perfectly. Then move to the next one. Don't try to automate everything at once. Each agent takes about a week to get right if you're focused.
