Blog · Published · Updated
AI agent for content marketing: from research to publish with editorial control
Short answer
In content marketing, an AI agent gathers sources for a topic, writes a draft to the brief and the brand's style, checks every number against its source and prepares a version for editing. An editor approves the text, and publication waits for a person. The agent costs from €3,000 net at Syntalith and, with 40 texts a month at 240 minutes each, pays back in 0.9 months.
How we workDetails ↓
- 01Call30 minutes with an engineer, free
- 02PrototypeA working system on your process and data, free
- 03ImplementationTarget and fixed price in the contract; 50% at signing
- 04Acceptance50% once the agreed target is met
- net per agent for one process, build paid once
- from €3,000
- from signing the contract to acceptance
- 6–16 weeks
- free prototype on your cases
- €0
We belong to the OpenAI and Anthropic partner programmes and learn through their official courses. What this means ↓
Author: Syntalith · 3 min read
Statistics
Marketing and AI: statistics 2026
0.9 mo.
payback of a €3,000 agent with 40 texts a month: €5,584 of work today, €2,251 with checking and the model. Scenario.
40%
According to a study published in Science (2023), ChatGPT cut the time 453 professionals spent on writing tasks by 40%.
Source: Noy and Zhang, Science (2023) ↗
34.7%
According to Eurostat (2025), 34.7% of EU enterprises that use AI use it for marketing or sales.
Source: Eurostat, Use of artificial intelligence in enterprises (2025) ↗
25.1%
According to a Harvard Business School study (2023) of 758 consultants, tasks within the model's capabilities went 25.1% faster with AI.
48%
According to the European Commission (2026), 48% of EU SMEs and mid-caps name a lack of expertise as a barrier to AI.
Source: European Commission, Flash Eurobarometer on AI and algorithmic management (2026) ↗
Figures checked: 22 October 2026
Why content slows down as it scales
A marketing team plans 40 texts a month and publishes 15. Every text means research, a draft, checking numbers, style corrections, an expert's approval and loading it into the CMS. Gathering sources and making sure every number has an origin take the most time. As the pace grows, so does the risk that a statistic from an aggregator, which nobody checked, ends up in a text.
The agent works to a brief: the topic, the audience, the keywords, and the allowed and excluded sources. It gathers materials, writes a draft in the brand's style and gives the primary source with its date beside every number. It marks a number without a source for removal. The editor gets the draft with a list of claims to check, and an expert assesses the substance.
- A brief with the audience, the goal and the allowed sources
- Every number with a primary source, its year and its sample
- An expert checks texts on law, health and finance
- Publication only after the editor's approval
Copyright and the AI Act matter too: the DSM Directive governs text and data mining, and the European Commission is preparing rules on the transparency of AI-generated content.
What it costs and when it pays back
- Today
- 40 texts × 240 min = 160 h (€5,584)
- With the system, a person checks the output in 40% of the time
- €2,234
- Model, 4 calls per case, Claude Sonnet 5.5
- €17
- Cost of one case
- €139.60 today, €56.27 with the system
- Saving
- €3,333 a month
Payback of €3,000
0.9 months
The volume, the time and the checking share are the scenario's assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat. Model: 40,000 input and 4,000 output tokens per call at Anthropic's rates, converted at the ECB rate of 1.1205 USD per euro on 22 October 2026. Checking takes 40% of the time because the editor and the expert read the whole text and check the sources.
The count covers the work around a case: searching, reading, writing and passing things on. What remains is the time to check the output, 40% of today's work in the scenario. Before the quote we measure both times on 20 to 100 cases the team has already handled, in a €0 test that takes 5 working days.
What the agent should not do
The agent does not publish without an editor, invents no quotes, clients or results and does not rewrite other people's texts. It does not decide the strategy either: the team sets topics and priorities from traffic and sales data. According to a study in Science (2023), ChatGPT cut the time of professional writing tasks by 40%. That result concerned single texts. A process with editing and source checks has to be measured separately.
A few texts a month
ChatGPT or Claude with a good brief
training from €1,150
A dozen or so texts, one editor
brief templates and a source checklist
no build
Dozens of texts, several channels, checking numbers
an agent that prepares drafts for editing
from €3,000
Where to start
The first scope covers one type of text and one channel. We compare the time from brief to approval and the number of editorial corrections with the current process. We build this as part of AI agents. For the free process review, a description of one case from recent weeks and the tools it ran through is enough. You do not need to send confidential documents, and you get a written plan within 2 working days.
Worked example · Scenario
How we count the payback: team collaboration tools
We run the same count for your team in the free process review.
€4,534
a month for finding agreements and handing over cases in a team of 10
Payback: 1 mo.
3 hours a week per person for finding agreements and handing over cases is our assumption. A 30% cut sits within the range of controlled studies (MIT: 40%, Harvard and BCG: 25%). €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat, for companies with 10 or more employees. Licence: ChatGPT Business or Claude Team at 20 USD per person a month on annual billing, about €17.85 at the ECB rate of 1.1205 USD per euro on 22 October 2026. Rollout: from €1,150 for up to 10 people, from €2,300 for 11 to 40 people.
See the full calculation →
