A text lands at 6:40 in the morning, before the first coffee is even poured. Three short lines, all in Portuguese, and nobody on the crew reads Portuguese fast enough to answer before the trucks roll out. So it sits.
By lunch somebody finally looks at it. Too late. The customer already texted two other shops, one of them wrote back in ten minutes, and that job is gone. You never even knew it was a real lead.
Massachusetts runs on this. We've got big Brazilian Portuguese and Spanish speaking communities in Framingham, Everett, New Bedford, Lawrence, all over, and a lot of that work goes to whoever answers first and doesn't make the person feel like a burden. Here's the good news: translating an inbound message both ways is one of the things AI actually does well now. The trick is knowing where to trust it and where to slow down.
What these tools can actually do now
The cheap and free options got a lot better the last couple of years. Google Translate, DeepL, and the chat assistants you might already pay for (ChatGPT, Claude, Gemini) can take a message in Portuguese or Spanish and hand you clean English in a second or two, then go the other way just as fast.
For everyday back and forth that's plenty. 'Can you come Thursday?' 'What's the address?' 'The quote is nine hundred, does that work for you?' Short, concrete messages translate cleanly because there's nothing fancy to get wrong.
Where it still stumbles is slang, regional wording, and your trade's jargon. Brazilian Portuguese isn't the Portuguese from Portugal, and Spanish from a Dominican family reads differently than Spanish from Mexico. The tools nail the meaning most of the time. They just miss the flavor now and then, and they mangle technical words, which we'll fix further down.
Where you can trust it to run on its own
For the routine stuff, you can let translation carry the conversation with just a quick glance from you. These are low risk because if something's slightly off, it's easy to catch and easy to fix in the next message.
- Scheduling: setting a day, confirming a time, telling someone you're running twenty minutes behind.
- Directions and logistics: the gate code, where to park, which door, whether the dog's friendly.
- Simple quotes and ballpark pricing: 'roughly twelve hundred for the whole job, I'll confirm after I see it.'
- Reminders and follow ups: 'we're on for tomorrow at 9,' or 'just checking if you still want that estimate.'
- Basic questions about your services and hours.
With all of that, translate, read it once to make sure it sounds like a person and not a robot, and hit send. Figure it saves you a few minutes per message and, more to the point, it saves the job you'd have lost by going quiet.
Where a human still needs to read it first
Some messages carry real weight, and that's where you want a person who speaks the language to check the wording before it goes out. Not because AI is useless here, but because the cost of a small mistake is high.
- Contracts, change orders, and anything someone signs. A mistranslated payment term or scope line can cost you real money or land you in a dispute.
- Medical and dental detail: symptoms, dosages, anything about a copay or what insurance covers. Get this wrong and it's not just awkward, it's unsafe.
- Legal wording, permits, and liability. If a lawyer would care about the sentence, don't let software be the last set of eyes.
- Money disputes and anything emotional. Tone matters most exactly when it's hardest to translate.
The honest version: AI translation is a strong first draft for the important stuff, not the final word. When it matters, have a bilingual employee, a trusted friend, or a paid service read it before you send. Ask us how we know.
Keep a glossary so it stops mangling 'soffit' and 'copay'
Here's the fix for the jargon problem, and it's simpler than you'd think. Every trade has words the general tools get wrong. 'Soffit' comes back as something vague about a ceiling. 'Fascia' turns into a medical term. 'Copay' and 'deductible' get flattened into words that don't mean what your patient needs them to mean.
So you build a little glossary: a short list of your terms and the right translation, written down once. Then you paste that list into your assistant before you translate, or save it as a standing instruction so it remembers. The tool stops guessing on the words that matter to you.
A roofing glossary might pin down soffit, fascia, drip edge, ice and water shield, and square (as in a square of shingles, not a shape). A dental office might lock in copay, deductible, crown, filling, and cleaning. Ten or fifteen terms covers most of what comes up.
Keep it in a note on your phone or a shared doc the front desk can reach. Add a word every time you catch a bad translation. In a month you've got a glossary that makes the tool sound like it actually knows your trade.
Setting this up without a big project
There's no need for new software or a big rollout. Most shops can get this running in an afternoon with tools they already have.
- Pick one tool and stick with it so everyone builds the same habit. DeepL is strong for Portuguese and Spanish; a chat assistant is better when you want it to use your glossary.
- Write your glossary. Fifteen terms, English on the left, the translation on the right.
- Set a simple rule for the crew: routine messages, translate and send; anything with a signature, a diagnosis, or a dispute, flag it for a human.
- Save a couple of canned replies in both languages for your most common messages, so the front desk isn't starting from scratch every time.
Worth doing this week
You don't have to solve all of this at once. Do these few things and you'll stop losing jobs to a language gap.
- Look back through last month's texts and voicemails for any message you were slow to answer because of the language. That's your real cost, right there.
- Pick your translation tool and put it on the phone the front desk actually uses.
- Write your fifteen word glossary and save it where the whole team can grab it.
- Agree on the one rule: routine goes out translated, anything a lawyer or a doctor would care about gets a human read first.
- If it's more than you want to wire up yourself, the team at Massachusetts AI Agency sets this up for local shops and can build your glossary with you. Call 617-398-0033.