Add text to get the translation that 22 AI models agree on
The translation option most small businesses don't know about
We pulled a year of orders and split them two ways: our biggest repeat clients, the law firms and agencies placing ten or more orders a year, against everyone smaller than that.
Small business orders use MTPE 2.7% of the time. Our biggest repeat clients use it on 22.5% of theirs. Eight times less. I assumed that gap was demand. It isn't. It's a conversation that never happened.
How I know it isn't demand
Look at what small clients actually write in the order notes when they do pick MTPE.
Nobody chasing a discount writes that much about quality. One client's entire brief was three words: "Make it perfect."
No domain notes, no process guidance. Just a quality bar, and trust that we'd handle the rest.
These aren't the clients I expected either. Not the price-shoppers. The ones sending nine languages through Crowdin for a technical release. The ones using MemoQ, or routing work through platforms like Showbie, because translation is already one piece of a bigger workflow they've already built. One instruction read "STRICT TAT, no extension," in the same sentence as a quality bar, not instead of one. They want both. That's the actual case for AI-plus-human, and the data's been making it for a while now, we just hadn't looked closely enough to say so out loud.
Who this actually shows up for
Academic material shows up in the MTPE group seven times more than its overall share. Medical content, five times more. Technical documentation, twice as much. These are small teams translating volume they could never afford to put through a human translator alone, line by line, at full price.
Legal and certified documents sit at the other end, essentially zero MTPE. Not because AI-plus-human can't handle the language. Because it isn't the right tool for a certified document, and we've never offered it to those clients in the first place. That part is by design, not a gap.
What happens once someone tries it
57% of small businesses who've ordered MTPE more than once come back for it again. Another 40% have only placed a single order so far, too early to tell either way. That 57% figure isn't a fluke. That's a client who got what they needed the first time.
So the demand was already there, sitting in a customer base we already had. Big repeat clients get this conversation naturally, they talk to a project manager often enough that it comes up. A small business ordering once or twice a year usually doesn't get that same conversation, through no fault of anyone in particular.
That's a gap worth closing.
FAQs
A: MTPE (machine translation post-editing) starts with an AI-generated draft, then a professional linguist reviews and corrects it before delivery. AI-only translation skips that human review step entirely.
A: Usually, yes, but Tomedes' own order data shows that isn't why most small businesses choose it. Clients requesting MTPE mention quality and accuracy in their instructions far more often than clients ordering standard human translation, suggesting the human review layer, not the discount, is the actual draw.
A: No. MTPE is built for volume-heavy content like technical documentation, academic material, and multilingual product content. Certified and legal translation, where a signed certification statement is required, stays a fully human process at Tomedes regardless of company size.
Written by Ofer Tirosh
Connect on LinkedInOfer founded Tomedes in 2007 and now leads the company's push to combine two decades of human translation expertise with AI. He writes about where the language industry is actually heading and the shifts nobody's ready for, the bets that paid off, and the ones that didn't.
Share:
Post your Comment