AI in your marketing. And across the business.
I build AI agents that read the numbers, set up campaigns and keep knowledge and email in order. I can do the same for your business.
Below are examples of what I have built for my clients and in my own business.
Keeps watch. Reads, compares and speaks up.
Competitor monitoring
Every Monday an agent goes through the nearest competitors: products and prices on their sites, their ads on Meta and their search rankings. The report says what has changed since the week before.
The numbers every morning
Yesterday’s spend per channel, new leads and sign-ups land in the team’s channel every weekday. If a source doesn’t respond, the message says so, so an empty field isn’t read as zero.
Ad review
Once a week every ad is held up against rules written down in advance: whether it is wearing out, and whether it has spent enough to be judged. The agent suggests, I decide.
The outside world
Every morning an agent keeps an eye on the conditions outside the business that are critical to it, such as official travel advice for the countries a client travels to. Only when something changes does a message go out, in the source’s own words.
Does the work. Built in full, approved by a person.
New ads
Once a campaign is decided, the agent builds it in the ad account with the images, videos and copy we have approved. It is created paused and only goes live once I have checked it.
SEO articles
An article goes from text to a finished post on the client’s site, with SEO title and description, internal links, category and image. It goes up as a draft, and the client presses publish.
Landing pages
New pages are built in the client’s WordPress to the standards the client’s own team has set for blocks, spacing, colours and images. They go up as drafts for review.
Reports
The monthly report is pulled straight from the ad platforms and becomes a PDF per market. Mid-month updates are written as a draft that I edit and send.
Keeps order. Knowledge, meetings and email where they belong.
Product expert
Staff can ask about the products and get answers from the product pages and datasheets, with a footnote on every fact. If the answer isn’t there, it says so. It is being tested on questions with known answers.
Structured context
Each client has a folder with the briefing, agreements, tone of voice and the decisions made so far. The agent reads it first and works on one client at a time. Any number in a text must trace back to a file.
From meeting to tasks
Meetings are recorded and transcribed, and an agent files them with the right client along with suggested tasks. If it isn’t clear who the meeting is about, it saves nothing.
Draft replies
Every morning an agent finds the emails in a partner inbox that need a reply and drafts one for each, based on what has been agreed with the sender. The agent can’t send on its own.
True for all of them
Written down
Every agent has instructions: what it may read, what it may write, and when to stop and ask.
Reads first
Most agents only read. Anything that can change something waits for a yes.
Checks
Before a run, the agent checks that it is in the right account. If the name doesn’t match, it stops.
Logged
Everything attempted in an ad account is logged, including what was refused.
How I help
I help find the tasks and build the agents.
It starts with the tasks that take up the most time each week. We go through them together, then I build the agents that make sense, with written rules for what they may do. Afterwards I follow up on whether they do what they should.
Your systems
The agents connect to what you already use: CRM, ad accounts, email and chat.
Custom software
When no tool fits, I build one.
A lot can be done with the systems that already exist. The rest is built as small programs that do one thing and connect to what the business already runs on.
AgencyOS
The client folders and the task system in one: every client has a folder and a board, and pipeline, time and shared rules sit in the same place. The agents on this page run on top of it.
feedcantrack
Feed Engine Group’s own tracking platform, which replaced a bought tool. It follows every click from a website to the sale at the affiliate networks and speaks up if a site goes quiet.
Data warehouse and dashboards
Dashboards built for where they will be seen, such as an office screen with leads and sales that reads from across the room. The next step is a data warehouse, so ads, CRM and sign-ups come from one place.
Lead database
Collects the forms from three websites in one place, removes duplicates and passes leads on. It replaced a shared spreadsheet.
Ad connector
My connection to Google Ads, Meta, TikTok and LinkedIn. Every change runs as a dry run first and is logged.
This website
bucca.dk is built and maintained with an AI coding assistant. Changes go live with a push.
Got repetitive tasks? Let’s see if an agent can take them on.
A report, a check or an inbox. From a description of the tasks, I can usually tell what it would take.