---
title: "Bringing AI into a whole company's work, starting from what the experts don't ask for."
url: "https://simonebova.com/en/runs/intarget"
lang: "en"
author: Simone Bova
alternates:
  it: "https://simonebova.com/md/it/runs/intarget"
  en: "https://simonebova.com/md/en/runs/intarget"
  pl: "https://simonebova.com/md/pl/runs/intarget"
---

# Bringing AI into a whole company's work, starting from what the experts don't ask for.

How I run the AI-native agentic platform a marketing company works in: what experts don't ask for, who signs off on quality, and knowledge built out of questions.

Name: Intarget

Terrain: Enterprise · marketing

When: 2025 → now

Role: Leading AI and AI product development

| Fact | Value |
| --- | --- |
| Terrain | Enterprise |
| Role | AI leader |
| Status | In production |
| Rule | Only what gets used stays |

## The situation

Intarget wanted AI in the everyday work of the people who write briefs, media plans and strategies, for the company and for its clients. I've worked on it since 2025, and today I lead the company's AI and AI product development. The goal was easy to state and hard to reach: a tool experienced people choose to open every morning because it makes their work better. This is the method. How the platform works inside stays at Intarget.

## What I did

I built the platform the company works in: a full-stack, AI-native agentic platform, a system of action, not a system of record. I run it from research to release. I kept the loop short: colleagues flag what's wrong while they work, I fix it and ship it. Models review the code first, then I do.

The decisions that mattered most were about trust. I removed from the interface everything the model had to keep up to date on its own, and kept only the indicators that come from work actually done. I moved permissions from the approved plan to each single action. I cut the agent catalogue down to the agents that have something of their own. For client knowledge, I stopped asking for documents and started asking questions.

## What changed

The platform is in production, and people who don't write code use it for real work. Feedback comes out of everyday work and turns into releases quickly. When a feature goes unused, I remove it. That rule is what keeps the platform readable for the person opening it in the morning.

## What I learned

In the first meetings with the people who write briefs and media plans, the requests were almost always to automate a step they could already see. Nobody named the steps that had hardened into habit, because after years in the job those steps are the standard, and they were often exactly where AI changed the work most. Collecting requests carefully wasn't enough to find them. Since then I ask the same question at every first meeting: what's the one thing that would change your life that we haven't mentioned, because it seems impossible? The building starts from that answer.

A strong model writes a decent attempt at communication for a luxury brand, because it knows the trade from what it has read. How that brand should be told now, with this season's trends, is known by the person who works with the client. Those same people, though, started out asking AI for far less than it can do. So the ambition and the sign-off on quality go to different people. I set the ambition, above the expert's first request. The sign-off stays with the expert at every step, and tests and evaluation cases keep it the same from one run to the next.

What the company knows about its most important clients lived partly in the CRM and partly in the heads of the people who own the relationship, and when someone left, part of it left with them. There had been three attempts with hand-filled presentations, all stalled for the same reason: nobody owned the updates and no routine made them sustainable. A better format wouldn't have changed anything, because any document updated by hand asked for work nobody had time for. So today whoever owns the client answers questions and confirms what goes in. When I design a knowledge base, I start from who answers, which questions, and after which meeting.

*// the numbers exist. they stay at Intarget.*

[Runs](https://simonebova.com/md/en/runs)

[simone@yempik.ai](mailto:simone@yempik.ai) · [Book a call](https://calendar.app.google/LKDrQPKGV2EAFCMN9)