Loading…
Use case · AI parsing & clean-up
Email is where structure goes to die: ten senders, ten layouts, one wall of HTML. With HideMy.world you classify mail by asking questions in plain English, and pull out the exact fields you need. No regex, no scraping, no fragile parsers.
An AI rule, verbatim
↓ matched · fields extracted
{
"operator": "LNER",
"booking_code": "ABC123",
"departs_at": "2026-09-14T09:35"
}One question, one mapping. Every future ticket arrives already structured.
The problem
Parsing by hand
Asking instead
Deterministic first
Most email decisions don’t need a model. Sender allowlists, subject checks and header matches are instant, free and perfectly predictable, so they always run first. When a plain rule settles an email’s fate, the pipeline short-circuits and no AI call is made.
Only the genuinely fuzzy cases reach the AI, and even then it reads a stripped-down copy of the email, not the raw HTML. That keeps every evaluation fast, cheap and consistent.
And you never fly blind: every rule can be tested against real sample emails you’ve already received, so you see exactly what would match before anything goes live.
step 1 · free & instant
Deterministic rules run first
sender, subject, headers; a match short-circuits everything below
step 2 · only if undecided
AI rule reads a stripped-down copy
your plain-English question, answered yes/no over lean text
step 3 · on match
Transformer extracts exact fields
booking_code, total, departs_at; named by you, mapped by clicking
out
Delivered as clean data
webhook, Telegram, or forwarded to your inbox
The honest bit
A language rule is only as good as its phrasing, and we won’t pretend otherwise: this is the same disclaimer you’ll see inside the app. Ask something vague like “is this important?” and no model on earth answers consistently. So when a rule is ambiguous, HideMy.world flags it and suggests a sharper reformulation before it costs you a single evaluation. Clear question in, reliable yes/no out.
⚠ ambiguous rule
“Is this email important?”
✓ suggested rewrite
“Is this an invoice or payment reminder from one of my suppliers?”
The toolkit
“Is this a booking confirmation from any UK train company?” That sentence is the rule, exactly as you’d say it. Each incoming email gets a yes or a no, and your pipeline routes on the answer. No query language, no operator precedence, nothing to memorise.
Draft a rule, let the AI tighten the wording, and preview what it would have matched against your recent mail, all before a single live email is touched.
Open a real email, click the booking code, click the date, name the fields. That’s the whole editor. No regex, no XPath, no guessing at markup.
Air France and Emirates format confirmations nothing alike. One transformer, two templates, identical output fields: your code reads one shape forever.
The AI never sees raw HTML soup. Emails are stripped to their essentials first, so each evaluation is quick, consistent, and your monthly quota goes a long way.
How it works
Create an address and write the rule as you’d say it aloud: “Is this a booking confirmation from any UK train company?” No syntax to learn.
Open a sample email in the transformer editor, click the values you care about, name them. Add a second template when another sender formats things differently.
Matching emails arrive wherever you point them (webhook, Telegram, or your inbox) as tidy named fields instead of a wall of HTML.
No testimonials, just configs
Plans
Personal pipelines: Lite
€9.90/mo: unlimited addresses and rules, 250 AI evaluations, 1,000 emails a month and 2 transformers. Enough to structure your bookings, invoices and parcels.
Heavier extraction: Pro
€24.90/mo: 1,000 AI evaluations, 5,000 emails a month, unlimited transformers, 5 team members and the read-only API. For pipelines that do real work.
Deterministic rules are on every plan. Start free with 2 addresses and no card. Compare all plans
FAQ
A stripped-down copy of the email: HTML boiled down to text, headers trimmed to the ones that matter. The AI reads that and answers your plain-English question with a yes or a no. Lean input means faster answers and far fewer tokens per evaluation.
No. The deterministic rules engine always runs first: sender, subject, header checks are free and instant. If a plain rule already decides the email’s fate, the pipeline short-circuits and no AI call happens at all. AI is the fallback, not the default.
We tell you, before it costs you anything. A rule like “is this important?” gets flagged with a suggested reformulation (say, “is this an invoice from one of my suppliers?”), the same warning you see inside the app. Sharp question in, reliable yes/no out.
Yes, always. Run any rule, deterministic or AI, against real sample emails you have already received and see exactly which ones match before you commit. No guessing, no waiting for the next email to find out you got the wording wrong.
Multi-template extraction. One transformer can hold several templates (Air France’s confirmation layout and Emirates’) and both come out as the same clean fields: booking_code, departs_at, passenger. Your downstream code never knows the difference.
Lite includes 250 a month, Pro 1,000. Deterministic rules never count against that, and because the engine short-circuits, plenty of emails are settled without an AI call at all. AI rules and transformers start on Lite; Free covers addresses and plain rules.
Write your first plain-English rule in under a minute, and test it against real emails before it ever goes live.