AI workflow setup
Not just in a chat window. I build two or three AI agents that run your repetitive marketing work in the background: drafting, research, reporting and monitoring. Your team approves everything before it goes out. Live in three to four weeks, on a fixed fee.
You have the tools.Not the setup.
Most marketing teams now pay for AI. Few have wired it into their daily work, so it saves far less time than it should.
Source: MMA and Decision Lab, The State of AI in Marketing 2026, Southeast Asia (survey of 143 marketers).
From busyworkto working agents.
Five steps, a fixed scope and a person signing off anything that matters.
Map the work
I list your team's recurring tasks and split them in two: low-precision work where a strong first draft is enough, and high-precision work such as pricing, claims and client sends, which stays with people.
Pick the first agents
I score each task on how often it happens, the hours it takes, how structured it is and whether success can be measured. The top two or three get built.
Build into your tools
Agents connect to what you already use, such as Gmail, Google Drive, Slack, Shopify, Klaviyo and your ad accounts. No new platform to learn.
Set the review rule
A one-page rule for what needs approval, what gets spot-checked and what can run alone. Nothing reaches a customer without a person signing it off.
Measure time saved
I take a baseline before we start, then report hours saved, output volume and error rate after 30 days.
The agents Ibuild most often.
Each one drafts, researches or reports on a schedule, and hands the result to a person.
Inbox and follow-up drafter
Drafts replies and chasers in Gmail each morning, ready for you to check and send. Nothing goes out on its own.
Content repurposer
Turns a podcast, video, interview or long article into LinkedIn and social posts built around the strongest quotes.
Conversion test planner
Compares your site with two or three competitors and hands back ranked A/B test ideas with the reasoning behind each one.
Pre-call researcher
Prepares a one-page brief on the person and company before every sales call, separating verified facts from what they claim.
Ad anomaly alert
Flags when spend, cost per acquisition or return on ad spend moves outside a set range on Meta, Google or Amazon, with the likely causes.
Weekly report writer
Turns your GA4 and ad data into a plain-English summary with three clear actions for the week.
Start small.Scale what works.
Fixed fees, agreed before we start. You keep the agents, prompts and review rule.
AI workflow map
- Task map with a low and high precision split
- Top five automations ranked by hours saved
- Draft review rule
Map and build
- Everything in the workflow map
- Two to three agents built and live in your tools
- Final review rule and team walkthrough
- 30-day time-saved report
Build and run
- Agents maintained and improved
- One new agent each month
- Monthly report on time saved and quality
I run my ownbusiness this way.
Every weekday, AI agents draft my replies, follow-ups, client check-ins, media chasers and admin across five inboxes. I approve every send. I build for clients what already works for me.
I am a growth marketer first. I optimised SupplementHub for AI search and ChatGPT is now among its top four revenue sources, and I led the MONPURE launch, which reached £500k in six months.
Goodquestions.
What is an AI agent?
An AI agent takes a goal and delivers a result, rather than just answering a question. It works through several steps on its own, uses your tools and data, and hands back finished drafts for a person to approve.
Will the agents send anything without approval?
No. Agents draft, research and report. Anything that reaches a customer, a journalist or your ad budget is approved by a person first, as set out in your review rule.
Do we need new software?
Usually not. I build into tools you already use, such as Google Workspace, Slack, Shopify, Klaviyo and your ad platforms. Any AI subscriptions are billed to you directly.
Who is this for?
Founder-led brands and marketing teams of two to ten people with a regular load of content, reporting or outreach. It is not for teams that want fully automated sending or AI making high-stakes calls on pricing or health claims.
How quickly will we see results?
The workflow map takes one week. Agents are typically live within three to four weeks, and I report hours saved after the first 30 days.
From theInsights blog.
Plain answers to the questions founders ask before hiring senior marketing help.
Ready to gethours back?
Tell me what your team spends its week on. I will tell you which two or three agents would save the most time.