> Short answer: AI saved us real time on drafting, first versions, and repetitive copy. But it sped up nothing in strategy, client relationships, or the "which work should we not take at all" decision. AI makes the first draft cheap; it does not make the judgment that actually creates value.
Over the past year we both tried to jam AI into everything and tried to ignore it entirely. Both were wrong. We have a clearer line now: where AI helps, and where it just feels busy.
Where it genuinely speeds things up
The first draft. The blank page is the most expensive moment. A blog outline, a campaign angle, the skeleton of an email sequence; producing these with AI in ten minutes and starting to work on top is far faster than starting from zero. What comes out is rarely publish-ready, but as a starting point it is priceless.
Repetitive copy. Product descriptions, meta descriptions, alt text, first-pass translation. These are already formulaic; AI closes exactly the gap where a person wastes energy.
Code and automation. Small scripts, data transforms, repetitive queries. On the engineering side AI is like a fast intern next to a senior: guide it and it does a lot, leave it alone and it makes a mess.
Where it sped up nothing
Strategy. To "what should this brand say", AI gives average, safe, everyone-sounds-the-same answers. But good strategy begins exactly where you break from average. Using AI here usually pulls the work toward a polished mediocrity.
Client relationships. Understanding what a client is really worried about, the thing they did not say in the meeting; that is still human work. The moments we had AI write a brief, we watched the most critical nuance disappear.
The "which work to decline" decision. Most agencies sink by taking on too much work, not too little. AI tells you how to do every job; it does not tell you which one to refuse. That judgment stays with us.
An honest failure
For a while we tried having AI write client reports entirely. It was fast and looked clean, but something was missing: the report described "what we did this month", not "what this means for your business". The client felt the difference immediately. We reverted; now AI assembles the data and a human writes the interpretation.
We built AI into Poitim with the same philosophy: the assistant (Timo) gathers, reminds, drafts; but the team gives the decision and the tone. Positioning AI as a good co-pilot rather than the driver has been our healthiest line so far.
Takeaway
See AI not as "does everything" but as "makes the first draft cheap". Spend the time it saves on the thing AI never touched: judgment. That is where the gap opens.
