Digital Brings
Start a project


Eight products. All shipped.

Digital Brings is young, and we’d rather show you something real than a page of logos we borrowed from a stock library. So here is everything we’ve built: a full school operations platform, and seven open-source products whose source you can read line by line rather than take our word for.

127
Automated tests
MIT
Licensed, the open seven
16
Scam scripts detected
12
Licence types tracked
69%
Support work automatable
24/7
Uptime monitoring built

New · Pilots open

Voxa — the AI calling agent that speaks 15 Indian languages.

Voxa console: calls by hour, outcomes and the callers who need a follow-up

Product · AI voice · Pilots open

Voxa

The problem

Leads go cold while the team is busy, reminder calls eat the front desk’s day, and nobody writes down what was said on the phone.

What we built

An agent that calls leads, confirms appointments and follows up on admissions from your own business number — in 15 Indian languages and English, mixing them mid-sentence the way people actually talk. Every call lands in a console with a recording, transcript, summary and next step.

The hard part

Indian-language speech on phone lines. The whole pipeline runs at 16 kHz end to end, because 8 kHz telephony audio clips exactly the consonants Telugu recognition depends on.

Pricing

Launch pricing: setup from ₹24,999, plans from ₹7,999 a month, calls from ₹3.9 a minute all-in. Reserve a pilot with a ₹5,000 advance.

15 Indian languages + English Your business number Human hand-off
See Voxa Try the console demo Book a pilot

Products

Built for real businesses and real people.

AI School Management System principal dashboard showing attendance, student strength and administrative activity

Platform · Next.js & PostgreSQL · Private build

AI School Management System

The problem

A school runs on registers, WhatsApp groups, printed circulars and whoever remembers what. Attendance lives in one book, marks in another, parent communication in a teacher’s personal phone. Nobody can answer a simple question — was this parent actually told? — without asking three people.

What we built

One platform for the whole school: attendance registers, assignments, study materials, events, announcements, fees, staff and student records, reporting, and parent communication over the WhatsApp Cloud API. Every screen reads and writes real data, and every protected action is authorised on the server — hidden UI is never the control.

The hard part

Permissions answer “may this user do this?”. Scope answers “to whose records?”. Both are enforced on every read and write, so a parent editing the student ID in the URL gets a 404 rather than another family’s child. Roles are data, not code: a school can create its own “Examination Coordinator” and grant permissions without a deploy — and can never lock itself out.

The AI, deliberately fenced in

The assistant never touches the database directly. It reads through the same scoped queries the interface uses, so it can only reason about records that person was already allowed to see, and treats the content it reads as data rather than instructions. Without an API key a deterministic engine answers the same questions straight from school records.

Next.js 16 TypeScript PostgreSQL RBAC + data scoping WhatsApp Cloud API
Private repository — ask us for a walkthrough
WhatsApp communication screen showing approved message templates and the parent delivery log
Teacher attendance register for a class section
The AI assistant answering a question from the school's own records
WhatsLens: your WhatsApp already knows your best customers

Customer intelligence · Browser-only · Open source

WhatsLens

The problem

Dozens of business owners said the same thing: business is slow. Looking closer, the problem wasn’t a shortage of customers — it was that they couldn’t see the ones they already had. Years of WhatsApp chats held every buyer, every budget, every “price kitna hai?” that never got a reply. That chat history is a CRM; nobody treats it like one.

What we built

Export a few chats, drop them in, and WhatsLens surfaces dormant VIPs who went quiet, hot leads asking about price and stock right now, life events mentioned in passing, and conversations where the customer spoke last and nobody answered. It reads English and Hindi.

The decision that mattered

Everything runs in the browser. Nothing is uploaded, ever — because asking a business owner to send us their entire customer conversation history is a non-starter, and it should be.

English + Hindi zero upload free & open source
Source on GitHub
ScamLens: risk you can see at a glance

Consumer safety · Pattern engine · Open source

ScamLens

The problem

AI made scams flawless. Cloned voices, perfect grammar, bank pages pixel-identical to the real thing. The instincts that protected a generation — “it looked fake”, “he sounded off” — quietly stopped working, and nobody sent a memo.

The insight

Scammers work from scripts. The digital-arrest call, the KYC-expiry SMS, the ₹3,000-a-day task job, the UPI collect trick. The packaging is AI-fresh; the script underneath never changes. And scripts can be pattern-matched.

What we built

Paste a suspicious message and get an instant verdict: the exact script it matches, the evidence quoted back from the message, a risk score, and what to do next — including the 1930 helpline if money has already moved. 16 patterns, scored entirely on your device.

16 scam patterns 100% on-device no signup
Source on GitHub
LicenseGuard: one licence was already gone and nobody knew

Compliance · Works offline · Open source

LicenseGuard

The problem

Ask a business owner when their trade licence expires. Most don’t know. Every business in India runs on several — Trade Licence, FSSAI, Shops & Establishment, Fire NOC, Weights & Measures, Pesticide, Fertiliser. Each has an expiry date and almost none of them warn you.

Why it’s dangerous

Nothing happens the day a licence expires. The business carries on exactly as before — until an inspection, a penalty, or a sealed shop.

What we built

A single dashboard with live countdowns for every licence, four clear statuses (Expired, Expiring Soon, Renewal Window, Active), renewal timelines for 12 common Indian business licences, calendar export and WhatsApp reminders. No account, works offline, data never leaves the browser.

12 licence types calendar export no account
Source on GitHub
AmIAutomatable: AI exposure by role, nine jobs and nine numbers

Career tool · Bring-your-own-key AI · Open source

AmIAutomatable

The problem

The biggest career risk isn’t AI — it’s not realising your role is changing until it already has. Displacement rarely arrives with a warning; it looks like a hiring freeze, a restructure, responsibilities quietly shifting.

What we built

A tool that answers “which parts of my job can AI already do?” in about 30 seconds. Choose from 9 pre-analysed roles or paste your own weekly tasks, and get a task-by-task split: what AI can do today, what works as AI-plus-you, and what remains your competitive advantage — plus three skills worth learning next.

The decision that mattered

It runs in your browser on your own free API key. No server, no account, no collection of what you typed about your own job.

9 roles analysed bring your own key no data collected
Source on GitHub

Infrastructure

Three tools we got tired of rebuilding.

The hardest part of software often isn’t the feature — it’s the plumbing you rewrite in every project. We solved these once and released them, and now client work starts further along.

Relay — automation you can read

TypeScript · MIT

Relay

Every automation starts as fetch, ask a model, notify someone — then drowns in retries, timeouts, rate limits and logging until the workflow is the smallest part of the file. Relay moves that into the engine and leaves you declarative JSON.

6 step types29 tests
GitHub
Sentinel — is the site up?

TypeScript · MIT

Sentinel

Most monitoring tools want a database, migrations and a container to answer one question: is my service online? Sentinel answers it with monitoring and a status page built entirely on the Node standard library — zero runtime dependencies.

0 dependencies39 tests
GitHub
Tessera — same place, six spellings

Python · MIT

Tessera

Real datasets never agree: NORTHVALE, Northvale, North Vale. Most mapping software silently drops the mismatches and still draws a convincing map — until someone notices a whole district missing. Tessera reports every unmatched record instead.

3 match tiers59 tests
GitHub

The unglamorous part

The tests found things we never hit while building.

127 automated tests across those three tools. Writing them was the least interesting week of the project and easily the most valuable.

Race condition

A monitor occasionally reported the wrong state — the kind of bug that shows up once a fortnight in production and never when you’re looking.

Unicode crash

A crash that only happened on Windows. Plenty of Indian businesses run Windows; “works on my machine” is not a shipping standard.

Silent data loss

A CSV parser turned 1,240.5 into 1 — producing a believable, completely incorrect map. No error, no warning. The worst kind of bug there is.

All three would have reached production. This is what we mean when we say tested — and it’s the same habit we bring to work we build for you.

Client case studies

This section is deliberately empty.

We could fill it with invented testimonials, borrowed logos and revenue figures nobody can check. Almost everyone does. We’d rather this space stay honest until there’s real client work to put in it — with the client’s name, their permission, and numbers they’ll stand behind.

If you’d like to be the first name here, we’ll make it worth the space.

Start a project

Want something like this, built for your business?

Same standard, same testing, same handover — pointed at your problem instead of ours.

Start a project See what we do
WhatsApp