Dealnote
Voice-recorded deals turned into documents
Dealnote lets two people record their verbal agreement on the spot, in India, and walk away with a properly formatted document instead of writing one up afterward. The Android app transcribes the conversation and turns it directly into structured text.
- 2025
- India
- Android app
- PHP admin + API
- OpenAI speech-to-text
- Document generation
The challenge
Verbal agreements between two parties, a sale, a rental, a settlement, are easy to make but tedious to turn into a usable written record afterward: someone has to remember what was said and format it correctly. Dealnote captures the conversation as it happens and produces the document automatically, so the people involved get a record before they leave the room.
Who uses it
- Users
- Record a conversation, review the extracted party details, and receive the generated document.
- Admins
- Manage user accounts and review generated documents through the web admin panel.
What we built
We built an Android app and a web admin panel on a shared PHP API. Users record a video or audio conversation; OpenAI speech-to-text transcribes it, then OpenAI extracts each party's name and address from the transcript and formats the conversation into a structured document. The app also handles sign-up, OTP verification, document viewing, and purchase history.
Audio/video recording
Users record a conversation directly in the app, up to a 25 MB file size limit for transcription.
Automatic transcription
OpenAI speech-to-text transcribes the recording with punctuation and attention to names and addresses.
Party detail extraction
OpenAI reads the transcription and extracts each party's name and address, flagging anything missing.
Document formatting
OpenAI turns the transcription into a professionally formatted legal document by category.
OTP verification
Sign-up and account actions are confirmed with one-time passwords.
Document & purchase history
Users can view past generated documents and their purchase history in the app.
Admin document review
The web admin panel lists accounts and the documents generated for them.
How it works
- User signs up and verifies their account with an OTP.
- User records a video or audio conversation between the two parties, within the 25 MB limit.
- OpenAI speech-to-text transcribes the recording with attention to names and addresses.
- OpenAI extracts each party's name and address from the transcription, prompting for anything missing.
- OpenAI formats the transcription into a structured legal document by category.
- The user views the finished document and its entry in their document/purchase history.
Where the AI fits
Once a recording is made, the API sends the audio directly to OpenAI's speech-to-text, with a custom prompt asking it to preserve names and addresses accurately. The transcription then goes to OpenAI, first to extract each party's name and address (prompting for whatever is still missing), then a second OpenAI call formats the full transcription into a structured, professionally worded document for the selected category. Recordings over 25 MB are rejected before the transcription step so costs and latency stay predictable.
- OpenAI speech-to-text for transcription
- OpenAI for party-detail extraction and document formatting
- 25 MB file-size limit enforced before transcription
- Missing-party-detail prompts fed back into the extraction step
Every part of the product
Android app
Kotlin
Records conversations, shows transcription and extracted party details, and displays generated documents.
Admin panel
PHP
Web dashboard for managing user accounts and reviewing generated documents.
API
PHP
Shared backend handling auth, OTP, the OpenAI calls, and document storage.
How it’s built
The Android app and admin panel share one PHP API, which handles the two-stage OpenAI pipeline: speech-to-text, then two separate OpenAI calls for party extraction and document formatting. Keeping extraction and formatting as separate calls lets the API re-prompt specifically for missing party details without re-running the whole document generation. Account security runs on OTP verification, and the admin panel gives staff visibility into every document generated without needing access to the raw recordings.
- Kotlin
- PHP
- MySQL
- OpenAI speech-to-text
- OpenAI API
- JWT
- OTP verification
- REST API
Related services: AI app development
What clients say about working with us
Verified Fiverr reviews from our clients, including clients in India, quoted word for word.
The team understands the requirement impressively well and they've built a truly complex and working product for me. They are always up to date with the AI trends and applied immediately in my product too. I'm very much satisfied with their work and highly recommended for any AI projects with Web app and mobile apps.
dnk3201Repeat buyerIndia · AI Websites & Software
"The team knows the craft and are expert in it. I'm impressed with the kind of knowledge they have in this industry and its remarkable to work with them. I highly recommend this team to everyone who is serious about their AI Startups"
sivakumarc45Repeat buyerIndia · AI Technology Consulting
The team understood the requirement and was able to deliver before time. They also took care of the deployment. Great job !
debabratatha136India · AI Websites & Software
Questions about Dealnote
What does Dealnote do?
Dealnote is an Android app that Venzpire built for a client in India. It records a spoken conversation between two parties and turns it into a formatted legal document automatically, with a web admin panel for managing accounts and documents.
Which platforms does Dealnote run on?
Dealnote is a native Android app for recording and reviewing conversations, paired with a PHP web admin panel for account and document management, both built by Venzpire on a shared PHP API.
How does Dealnote use AI?
Dealnote sends recorded conversations to OpenAI's speech-to-text, then uses OpenAI in two steps: first to extract each party's name and address from the transcript, then to format the conversation into a structured legal document. Venzpire built this pipeline so users get a usable document straight from a recorded conversation.
Your product could be next.
Tell us what you are building. Venktesh reads every application himself and replies within one business day.
Apply for a strategy call