Z. Ahmed
I have spent ten years in growth marketing. Today I run my own company, Your Growth System, with a team of AI agents. Eight of them work every day. They plan, write, build, check each other's work and watch for problems. Nothing goes live until I say yes. Here is what I have built.
Now: Director of Growth Marketing at Veritas Automata. Before: Chartis.io, Bailey Brand Consulting, BeMo.
I call on original specialists too: Mark (growth strategy), Vera (blocks machine-made writing), Iris (design), Gus (claims), and a CEO twin.
Each agent has one job. A different agent checks the work. I make the final call on what goes live, any money spent and any price.
Agents write the first draft. Other agents check it. Nothing ships until I say yes.
An AI voice receptionist answers the calls an owner cannot take and writes down what each caller needs. In one test call it answered, handled an interruption, booked a visit on the calendar and sent a confirmation email. That was one call. I keep testing.
Launching now. One test call so far.
With Phone Catcher on and calls forwarded to it, a missed caller is saved as a lead and the owner gets an email with a text ready to send from their own phone.
Built. Not yet tested with a live customer call.
Add a quote you sent. AI drafts up to three follow-up emails, for day 2, day 6 and day 12. Each one waits for the owner's OK, then goes out on its day.
Built. Follow-up sends not yet tested with a customer.
One list for every lead and customer, with a next step, a due date and dated notes. Bring a list in with a CSV file. Take it out the same way.
Live in the product.
Eight agents build and check my software. One writes code. One reviews it and can say stop. One watches system health. One looks for new methods and what they cost. Nothing goes live until I say yes.
Live. This is how I work today.
A blog helper researches what people ask, writes the post, and runs it past a checker named Vera. She blocks writing that sounds machine-made. A person approves it before it goes out.
I study which hooks work in the first two seconds of a short video. I keep a library of fill-in-the-blank hooks and use one for each new script.
The detail that mattersI take apart short videos that did well and write down why. The notes go back into the library.
Every Monday, research sweeps run. One checks what people are searching for and asking in five industries. One tracks what competitors did that week. A third weekly job checks what is new in AI agent methods and what it would cost. The findings go into one shared record. The strategy agents read the newest ones before they plan, so research reaches the work with no one carrying it over by hand.
My agents hand each other tasks on a shared list. One asks, another does the work, a third checks it. At night they stop before anything customers can see. I sign off in the morning.
Each new client gets a branded intake. They talk, upload files and answer questions that change as they go. The agents turn it into notes the team can use. Then the client gets an approval board where they check the work before it goes out. I have done this for two client businesses.
Each system I build comes with a written playbook, so someone new can take it over. Those are the playbooks I have been writing since my first marketing job.
Research lands on Monday. The agents read the newest findings before they start work.
Research feeds strategy. Strategy feeds production. I check the results and feed them back into strategy.
Chief carried the work and my yes between Guard and Growth. The fixed page was live 24 minutes after the rollback.
Green checks missed the crash. Reading the real page found it, and the fix was read there too before I called it done.
On copy changes, Gus checks claims, Vera checks for machine-made writing, and Iris checks design.
The note carries the job across each restart. The fix went live within the hour of my screenshot.
The note is a file. The job's history lives on a shared task list.
Most checks find nothing and stop on purpose. The one that finds something does the work, and a second timer proves it.
A different agent checked the work before it reached me. I got one clear pick, and the video waits until I say go.
Only the kind of job that passed moved to the lighter model. The whole test took eleven minutes.
Everything I have built is written up in the Letters, including the parts that broke.
What worked in my phone test, what I still need to check, and how that changes the way I explain the offer.
How I split planning, drafting and review across our AI setup, and what marketers can borrow from it.
A model can only repeat the business facts we leave in its instructions.
Before the agents, the same job with people: multi-client budgets across Google, Meta, LinkedIn and programmatic, KPI frameworks, CRM architecture, and being the escalation point between clients and leadership. The systems work because that discipline is real.
An AI growth system for service businesses: a plan to follow, a monthly Autopilot with a customer list, follow-ups and phone tools, and a done-for-you build. The Agent OS is the white-label version for agencies. Built with agents as the engineering team, checked by other agents, approved by me.
I moved from associate-level work into director-level growth work, with a focus on growth strategy, demand generation, CRM architecture, and AI-assisted research and reporting.
Multi-client digital strategy across Google, Meta, LinkedIn and programmatic. Client spend management, KPI frameworks, and the escalation point between clients and leadership.
Brand-side digital: paid budgets and A/B roadmaps across eight platforms for the firm and its clients.
PPC, funnel testing, attribution (Hyros), demand-gen automation, and the first SOPs I ever wrote. I've been turning workflows into systems since my first marketing job. Now AI agents do the first draft and a person approves it.
Open to Director+ roles leading AI-driven growth, marketing or GTM systems, and to walking your team through a live demo of everything on this page. Ten minutes, real systems, no slides.