There's an old saying in computing that's having a moment again thanks to AI: garbage in, garbage out. Feed a system bad information, and no amount of intelligence on the other end will save you from a bad result. For design firms starting to bring AI into their operations, this is the single most important thing to understand before you automate anything.
It's tempting to think AI is smart enough to fill in the gaps — that if your client notes are a little messy or your vendor list is half out of date, the AI will just figure it out. It won't, and it shouldn't have to. An AI sourcing agent pulling from an outdated product catalog will confidently hand you discontinued items. A lead-scoring agent reading vague intake notes will misjudge a serious client as a low priority, or the reverse. A scope-tracking system checking against the wrong budget number will either cry wolf constantly or miss the real overage entirely.
The agent isn't wrong. The input was.
Vendor and product data. If your sourcing system is pulling from a product catalog that hasn't been updated in six months, it's sourcing from a world that no longer exists. Discontinued fabrics, outdated pricing, vendors who've changed their ordering minimums — all of it quietly degrades the quality of every recommendation downstream.
Client intake forms. A rushed or vague intake form gives your qualification agent almost nothing to work with. "Budget: flexible" tells an AI agent nothing useful. "Budget: $15–20K, has gone over budget before for the right piece" tells it everything it needs to score the lead accurately and draft a follow-up that actually fits.
Project notes and style profiles. If the style profile in your client record says "modern, neutral" and hasn't been touched since the first consultation, but the project has since shifted toward warmer tones and more pattern, every AI-assisted sourcing pull from that point forward is working from a stale snapshot of a client who's moved on. And isn't it nine times out of ten that the client ends up wanting the exact opposite of the style they swore they wanted?
Scope and budget fields. An automated scope-tracking system is only as honest as the number it's checking against. If the "original budget" field never got updated after a legitimate client-approved change, the system will flag perfectly normal spending as an overage — and you'll start ignoring the alerts altogether, which defeats the entire point of having them.
This is good news, not bad news. It means you don't need a more sophisticated AI system to get better results. You need cleaner, more current data going in. A few low-effort habits make an outsized difference:
AI agents don't compensate for messy systems — they amplify whatever's already there, good or bad. A firm with clean, current data and a modest automation setup will outperform a firm with cutting-edge AI running on stale inputs, every time.
If you're building out automation for your firm — sourcing, scope tracking, client communication, whatever it is — the highest-leverage thing you can do before turning it on isn't picking the fanciest AI model. It's making sure what you're feeding it is actually true.
Book a 30-minute Honeycomb call and we'll walk through where AI can quietly take work off your plate — no jargon, no pressure.