Every service business has the same inbox problem: a customer writes three paragraphs about their deck, their dog, and their vacation plans before finally mentioning that the job is a fence repair, it's urgent, and they have a budget in mind. Someone still has to read the whole thing and pull out the actual job details. This weekend's build fixes that without hiring a developer: a small app that takes a messy inquiry and turns it into a clean, structured summary plus a draft reply, built with Google's no-code tool Opal.
What Opal Actually Is
Opal is a free, experimental tool from Google Labs that lets you describe an app in plain language and get back a working mini-app, built as a visual chain of steps rather than code. You type what you want, Opal lays out the workflow as connected steps on a canvas, and you can edit any step, swap the AI model it uses, or add a new one. Google expanded it from an initial handful of countries to more than 160 countries, including Canada, so it's available to try without a VPN or workaround.
It's worth being clear about what it is not: a hosting platform for mission-critical software. It's a fast way to prototype a small, specific tool and share it as a link — which is exactly the scope of a weekend project.
The Build: A Quote-Request Cleanup Tool
Here's the app we're building: paste in a rambling customer message, and get back (1) a short bullet list of the actual job details and (2) a draft reply you can edit and send. Nothing here auto-sends anything — you stay in control of what goes out.
- Sign in. Go to Opal with a Google account — no download or install required, per DataCamp's walkthrough of the sign-in and setup flow.
- Start a new app. Click to create a new app and describe it in a single prompt, something like: "Take a customer's message about a service request. Pull out the job type, location, timeline, and budget if mentioned, as a short bullet list. Then draft a short, friendly reply confirming we received the request and asking for anything missing."
- Let it build the workflow, then look at the steps. Opal turns that description into a chain of steps on a canvas — typically one step that extracts the structured details and another that drafts the reply. You can click into either step and edit its instructions directly if the output isn't quite right.
- Test it with a real example. Paste in an actual rambling inquiry (anonymize any real customer's info first) and run it. Check whether the bullet list is accurate and whether the draft reply sounds like your business, not a generic bot. Adjust the step instructions — “keep the reply under four sentences,” “use a casual tone,” “never promise a price” — until it does.
- Share it. Once it's working, publish the app and you get a shareable link. Anyone on your team can open that link, paste in a message, and get the cleaned-up version back — no account setup on their end required for the basic use case.
Before You Put It In Front of Customers
Treat the output as a draft, not a final answer, every time. A review of Opal for founders found it's a strong tool for prototyping and demos but flagged real gaps for production use — no custom domain, no offline mode, and no straightforward export path if you later want to move the logic somewhere else (HackerNoon's review). That's fine for this build: the job here is to save someone ten minutes of re-reading a rambling message, not to run your customer intake pipeline.
Two practical rules for this specific tool: never paste in anything with a customer's financial or health information, and always have a person read the draft reply before it goes out. The app should make your team faster at triage, not replace the judgment call on what to actually say.
Takeaways
- Opal lets you build a small, specific app by describing it in plain language — no coding, and it's free to try.
- Start narrow: one input, one or two clear outputs. A quote-request cleanup tool is a good first build because the win is obvious and low-risk.
- Edit the generated steps until the tone and format match your business — the first draft from the prompt is a starting point, not the finished app.
- Keep a human reviewing every output before it reaches a customer. This is a triage aid, not an autoresponder.
If this works well, the same pattern — paste in messy text, get back a structured summary and a draft response — applies to voicemail transcripts, Facebook Messenger inquiries, or after-hours web form submissions. Build the narrow version first, see if your team actually uses it, then decide if it's worth expanding.