DEV Community: acceptedsoul_11

Talk2Invoice: Turn a Job Note Into a Client-Ready Draft With Open-Weight AI

acceptedsoul_11 — Sat, 17 Oct 2026 06:34:29 +0000

This is my submission for the Hacktoberfest Open-Source AI Challenge: Week 2 — Side Hustle.

What I built

Talk2Invoice helps an independent service worker turn a short note about a finished job into a client-ready invoice draft.

The person can type a note or use optional browser dictation, choose one of 16 currencies and one of seven draft languages (English, Hindi, Bengali, Spanish, French, Portuguese, or Arabic), then ask an open-weight model to suggest the client and line items. They can review and edit every field, add a tax rate if it applies to them, and print or save a PDF when they are satisfied. The interface is in English for now; draft-language support does not localize every control or provide country-specific tax guidance.

These hand-drawn SVG interface previews use fictional sample records. They are not captures of a live app session, a real customer transaction, or an inference result from a live model.

Who it is for

The first likely users are independent repair workers, tutors, cleaners, and freelancers who finish a job away from their desk and want to send a clear bill without learning an accounting system.

Someone might pay for a tool that saves them a few minutes at the end of every job and keeps customer details on a device they control. That is a product hypothesis, not a validated sale: I have not yet asked a service worker to test it or confirmed willingness to pay.

How I built it

The UI is plain JavaScript served by a small Node.js server. When a user requests a draft, the server calls an Ollama-compatible local model and requests structured JSON. The suggested values go back into an editable form.

The recommended model is Qwen2.5 1.5B, an open-weight, multilingual model available through Ollama. The model is replaceable with another Ollama model by changing one setting.

I deliberately left the model out of decisions the app can make reliably. Code calculates quantities, rates, tax, and totals. If the model does not provide a rate, that field stays blank and the person must fill it in. The invoice remains a draft until the person checks it.

Why open innovation matters here

The job note can include a client’s name, work performed, and agreed prices. Running the model through an Ollama service on a machine the user controls keeps that note out of a hosted AI API by default. Invoice data and business settings are stored in the browser, and the model is contacted only after the person requests a draft.

Open models also make it possible to change the model or adapt the language and workflow without tying the tool to one provider’s account or pricing. Local inference still uses memory, power, and time; I have not measured whether it is cheaper than a hosted model, so I am not making a cost-savings claim.

One privacy detail deserves a clear explanation: browser voice dictation may use the browser provider’s speech service. Typing is available for people who want to avoid that service. Talk2Invoice does not record or upload audio itself.

Demo and code

Source: Talk2Invoice on GitHub
Run it: install Ollama, run ollama pull qwen2.5:1.5b, then use docker compose up --build from the repository.
Open the app: http://localhost:3000
Tests: run npm test. The API contract is tested against a mock Ollama server; this does not evaluate real model accuracy or speed.

The first model download needs internet. After that, the local Docker setup reuses the model volume.

What is not finished yet

Talk2Invoice does not send email, collect payments, sync between devices, or file taxes. Its interface is currently English-only, and its draft-language choices do not replace full localization or market-specific tax rules. It has not been tested with an external service worker, and I have not validated whether anyone would pay for it. The next useful step is to watch one independent worker create an invoice after a real job, with permission, then fix the parts that slow them down.

Tax is an arithmetic field, not advice. AI-generated details can be wrong; check every amount, date, and name before sharing an invoice.

Why I chose this workflow

There are already several invoice and receipt sorting projects in this challenge. Talk2Invoice takes a different part of the work: the moment after a service is complete, when a worker needs to turn what they just did into a clear invoice. The small scope keeps the app approachable while leaving room to add a market-specific template or language after talking to real users.

Source notes

StrukLokal: Local-First Receipt OCR for Freelancers describes a local OCR workflow that treats extracted values as suggestions, then asks a person to correct and approve them.
I built an offline invoice sorter for freelancers, then measured whether local inference is actually cheaper reports that privacy and control were clearer benefits than assumed cost savings, which is why this project makes no cost claim.
Schema-Valid Isn't Correct: 4 Checks Before You Trust LLM Extraction explains why structured output still needs evidence checks, arithmetic validation, and human review.

Mind Music Walks: a little room to breathe

acceptedsoul_11 — Sat, 10 Oct 2026 16:30:19 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.

What I Built

Mind Music Walks is a gentle audio companion for taking a short walk when your mind feels busy, low, restless, or simply ready for a pause.

Before heading out, choose how you feel, how much time you have, and whether you’d like soft tones, birdsong, rain, or outdoor ambience. The guide offers a few calm prompts, then gives you space to notice the world around you. It’s made for everyday walks in India, with English and Hindi support.

Demo

The app is running locally for now. To try it, clone the repository, run npm start, and open http://127.0.0.1:4173. The README has the setup details.

Code

Mind Music Walks on GitHub

How I Built It

I built it as a small, local-first web app. For English prompts, it can ask Gemma 3 running locally through Ollama to shape a short guide around the selected mood. Hindi prompts are written into the app so they stay natural and dependable. The browser handles the voice and ambient sounds, and the app shell can be cached for offline use after the first visit.

There are no accounts or analytics. Mood choices aren’t saved, and reflection notes stay in the browser on the device. If Gemma isn’t available, the walk still works with the built-in guide.

Why Does Open Innovation Matter?

Open models made it possible to keep the guide on the same device as the person taking the walk. That means the mood they choose doesn’t need to be sent to a hosted AI service. It also means the model, prompts, and fallback behavior can be inspected and changed by the community.

For a project about stepping away from screens, I wanted the AI to help prepare the walk, not become the walk.

Prize Categories

Overall Touch Grass challenge
Best Use of Gemma

Interview Mitraha: a little local AI confidence boost for my friend

acceptedsoul_11 — Fri, 02 Oct 2026 13:03:48 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend.

What I Built

Interview Mitraha is a little interview-practice companion I built with a friend in India in mind. The days before an interview can make even a simple “Tell me about yourself” feel heavy. I wanted to give my friend a calmer place to rehearse—without an audience, a score, or another expensive subscription.

You choose the role you’re preparing for, and Mitraha opens with one gentle question. From there, you can type or speak in English, Hindi, or Hinglish. If you use voice, you get to review the transcript before sending it. Mitraha follows up like a patient practice pal, helping you find your own words instead of handing you a script to memorize.

I haven’t handed it to my friend for feedback yet, so I don’t have a reaction to quote. That real-world tryout is still ahead of me.

Demo

There isn’t a hosted demo: I wanted the model and interview practice to run on the learner’s own computer. After installing Ollama and Node.js, the first run downloads Gemma (about 3.3 GB). Then start the app in PowerShell:

.\Start-Interview-Mitra.ps1

Open http://127.0.0.1:4173/ and choose a role. The repository’s README has the complete setup guide.

Code

Repository: github.com/acceptedsoul-11/Interview-Mitraha

acceptedsoul-11 / Interview-Mitraha

First Learning milestone in hacktoberfest-2026

Interview Mitraha

A friendly, local-first interview practice partner powered by Gemma.

Interview Mitraha gives a friend preparing for their next opportunity a low-pressure place to rehearse answers. It opens with one warm-up question, lets them type or speak a reply, and keeps the practice conversation on their own computer with Gemma 3 running through Ollama.

The name combines Mitra (friend) with the Sanskrit form Mitraḥ (मित्रः). The interface is designed to feel like a supportive practice pal: no scores, no accent ratings, and no claim that an AI can decide whether someone is ready for a job.

What it does

Starts a role-specific practice session with a gentle opening question.
Sends answers to Gemma 3 4B through a local Ollama server.
Supports typed answers and browser voice recognition in English, Hindi, and Hinglish.
Shows a voice transcript for review before sending it to the model.
Reads a Gemma reply aloud…

View on GitHub

The project includes Windows start/stop scripts, setup notes, the test suite, and a GitHub Actions workflow.

How I Built It

The UI is plain HTML, CSS, and JavaScript. A small Node.js server runs on loopback and sends each checked, bounded conversation to Ollama on the same computer. Ollama runs the open-weight gemma3:4b model; it writes the opening question and the follow-up replies. There is no hosted model API in the conversation path.

flowchart LR
  A[Browser: type or review a transcript] -->|127.0.0.1| B[Node.js: validate and bound messages]
  B -->|local Ollama API| C[Gemma 3 4B]
  C --> B --> A

I kept the app small on purpose: no account, no conversation database, and no runtime npm dependencies. The nine automated HTTP tests run with Node’s built-in test runner and a local mock Ollama endpoint. I also tried a typed answer against the real local Gemma model.

Voice has one important caveat. Speech recognition comes from the browser: some browsers process it on-device, while others may send audio to their configured speech service. Mitraha makes that boundary visible and lets you review the transcript; pressing Send sends the text to local Gemma. Browser speech synthesis can read Gemma’s answer aloud when available.

Why Does Open Innovation Matter?

Interview answers often include personal stories. With local Gemma, typed answers and model inference stay on the learner’s computer. There’s no cloud model account, and no per-answer API bill. Once the model has been downloaded, typed practice can work offline.

Open weights also give the learner room to change the experience. They can inspect or edit the prompt that gives Mitraha its patient tone, or swap to a smaller compatible Gemma model if their computer has less memory. A closed API could provide good answers, but it wouldn’t give the same control over the model and the path the conversation takes.

I don’t want to overstate the privacy story: voice transcription depends on the browser’s implementation, and local inference still needs capable hardware and electricity. I’d rather say exactly where the boundary is than make a blanket “everything is private” claim.

Prize Categories

Best Use of Gemma