Vladyslav Lysechko
+47 462 17 168
NOEN
Book a short call

AI agent · Meta Ads

AI marketing agent that manages Meta ads: Eva

A recruitment agency ran many campaigns in several languages and had nobody watching them all day. Eva is an AI agent that lives in the team chat on Telegram, reads the tasks, calculates cost per qualified lead from the CRM and does the work in the ad account herself, within limits we agreed on.

Want something similar?

By Vladyslav Lysechko, Lysechko Digital · Updated

Client
Recruitment agency in Poland
Channels
Meta Ads, Google Ads, Telegram
Delivered
AI agent, rules, reports, banners
Period
October 2026, running
Eva's log of actions in the ad account and a chat on Telegram
Illustration. The numbers in the top tiles come from Eva's log for 17 to 30 October 2026, the single actions and the chat are examples.
In short

Eva runs every 20 minutes. From 17 to 30 October 2026 she paused 64 ad sets, changed budgets 32 times, swapped 20 creatives and launched 10 campaigns. She reviewed 312 Google Ads search terms, made 107 banners and found wording Meta does not allow in 163 of 500 ads.

64ad sets paused in two weeks
32budget changes within the daily cap
107banners made by Eva
163of 500 ads with wording Meta does not allow

The starting point

The agency ran dozens of campaigns for different jobs, cities and languages. Budgets needed daily changes, weak ads should stop early, and Meta had already removed a post because of its wording. The team had no time to sit in Ads Manager all day.

What I built

  1. An agent that runs itself. A scheduler starts Claude Code with a dedicated skill every 20 minutes. Eva uses 23 Python tools for Meta, Google Ads and the CRM.
  2. Three levels of autonomy. Eva reports, suggests or acts on her own, depending on the size of the change.
  3. Fixed limits. An hourly cost guard for the whole account, an automatic pause after 38 zł with no leads, and scaling only when the CRM shows real leads.
  4. Ad copy checks. Every ad text is checked against wording Meta rejects before it goes live.
  5. Banners. AI backgrounds, text set in HTML and an automatic face check, in 4:5 and 9:16.
  6. Know how. 13 step by step playbooks and 8 written cases Eva uses when something unusual happens.
  7. A watchdog. An alert if Eva or a connection stops.

The result

In two weeks Eva made 64 pauses, 32 budget changes, 20 creative swaps and 10 launches, all within the limits. On the day one campaign was relaunched with new banners and copy, its cost per qualified lead was 26.30 zł against 70.86 zł for the previous version. The team writes to Eva in the chat instead of logging into Ads Manager and gets a report every morning and evening.

Eva does not own the strategy. She does the daily work fast and the same way every time, and asks a human before anything above the limits.

Tools and tech stack

  • Claude Code
  • Python
  • Meta Marketing API
  • Meta Ads MCP
  • Google Ads API
  • CRM API
  • Telegram Bot API
  • launchd

The client is anonymised. The numbers come from the project's own reports and measurements.

Want this for your business? See Meta ads specialist or Google Ads consultant.

Project gallery

A banner made by Eva: AI background, text set in HTML and an automatic check before publishing.
A banner made by Eva: AI background, text set in HTML and an automatic check before publishing.
The same template for another job, without claims Meta does not accept.
The same template for another job, without claims Meta does not accept.
The 9:16 version for Stories and Reels.
The 9:16 version for Stories and Reels.

Questions and answers

What is an AI marketing agent?

A program with a language model that can use tools: read numbers from the ad account and the CRM, then pause, scale or launch ads. Unlike automated rules in Ads Manager, the agent understands tasks written in plain words in a chat.

Can an AI ads manager overspend?

Not here. Eva has an hourly cost guard, a fixed pause threshold and a cap on budget changes. Above the cap she asks a human first.

Which ads can Eva manage?

Facebook and Instagram through Meta Ads, and search terms in Google Ads. Her reports are based on CRM leads, not only on Meta's numbers.

Can I learn to build an AI marketing assistant like this?

Yes, see the guide on building your own AI ads manager. If you would rather have it built for you, you get a fixed price in writing before we start.

Vladyslav Lysechko
Vladyslav Lysechko, Performance marketer, Lysechko Digital, Rånåsfoss, Nes, Norway

I have worked with ads on Google, Meta and TikTok for more than seven years and personally managed ad budgets from $1,000 a day up to $50,000 a month. For the last two years I have built websites, automations and AI assistants with AI every day. I help local businesses in Oslo and on Romerike get more calls and bookings. LinkedIn

More case studies

Systems and dashboardsAds and tracking

Meta ads and CRM dashboard

One marketing dashboard that joins Meta ad spend with CRM leads, and found 41.6% of the budget bringing zero candidates.

84.3%of ad spend matched to campaigns, up from 59.6%

Read the case study

AI and automationSystems and dashboards

WhatsApp recruitment chatbot

A bot that asks candidates the key questions on WhatsApp and creates a finished lead in the CRM with the ad source.

526conversations in four months

Read the case study

Ads and trackingWebsites

Recruitment ads and landing pages

Facebook, Instagram and Google campaigns with a landing page for every job, tracking every lead from ad to CRM.

67leads at about 7.7 zł after approval

Read the case study

All case studies

Let's have a short call

Tell me a little about your business and you get a concrete suggestion with a fixed price.

Send an email Message on WhatsApp

Email vlad@lysechko.com · Phone +47 462 17 168

Book a short call WhatsApp