a · Multimodal vision
Reads what others can't.
Handwritten receipts, faded thermal prints, glossy QR-only slips, mixed-language line items, non-English merchant names — all parsed into the same clean schema.
Snap a receipt. Talk to your phone. Paste a screenshot. AnytimeBudget extracts the transaction — merchant, date, line items, amount — and hands it back structured. No forms. No typing. No spreadsheet.
We don't ship "AI receipt scanning" as a checkbox feature. AI is the engine. Multimodal vision and language models do the heavy lifting so you can move on with your day.
Speak or type an expense and it's structured before you put the phone down. Receipt scans take a little longer — multimodal vision earning its keep.
Receipt scan, voice, image, file. Submit any of them; get the same shape of result back.
Every entry auto-converts into your home currency. Daily rates, cached locally for offline use.
Handwritten receipts, faded thermal prints, glossy QR-only slips, mixed-language line items, non-English merchant names — all parsed into the same clean schema.
"Split with Lana." "Round up to fifty." "Minus the tip." "Same as yesterday but at lunch." Speak the way you'd tell a friend — the AI infers the rest.
¥, ฿, €, ₩, ₫ — recognized as units, converted on the spot. A bowl of ramen entered in JPY shows up settled in HK$ next to the ferry fare.
The engine is one thing. The interface around it is another. Four entry surfaces, each tuned to the moment you're capturing.
Point your phone at the receipt. Multimodal AI itemizes merchant, date, line items, tax, and tip — the wait is short enough to put the phone down and pick it back up. Edit anything that's off — categories learn from your corrections.
"Forty bucks at the noodle shop, split with Lana." Long-press the mic, talk like a human, and it parses amount, merchant, category, and split. No forms.
160+ currencies with daily rates. Record a Tokyo ramen in ¥, see it settled in HK$, and let trip totals reconcile themselves. Cached locally — no roaming required.
Heatmap calendars, category trends, and on-pace budget projections that update as you spend. The math runs on-device. No cloud crunching required to know last Tuesday cost you $312.
The path from photo to filed transaction has no detours and no parking spots.
Photograph the receipt, hold the mic and talk, or paste a chunk of text. The app encodes the payload and opens a TLS connection to our backend.
The backend forwards your payload to a multimodal AI model. The model returns structured JSON — merchant, date, line items, amounts, currency, tax. Voice and text typically finish in under two seconds; receipt scans take a little longer while vision does the heavy lifting. The payload exists only in the memory of the request; nothing is written to disk.
The transaction lands in the app for you to glance over, edit, and file. From here on it lives on your device. The backend has already forgotten the request.
AI does the heavy lifting — but on our side, no copy of what you sent or received is ever written down.
Full mechanics in the Privacy Policy →
AI runs on your quota, not your data. Free covers casual use. Plus and Pro raise the ceiling for travellers, freelancers, and anyone who'd rather not type.
Unlimited manual entries, charts, accounts.
For the regularly-curious. Most people pick this.
For freelancers, travellers, and the receipt-heavy.
Prices in Hong Kong dollars. Billed monthly through Apple or Google. Cancel anytime in your store account.
The full app — receipt scanning, voice entries, 160+ currencies, private by architecture. Free to start. Plus and Pro raise the monthly AI ceiling when you need it.