How we work

Security, guardrails, the stack — and the rest of what we've built.

The stock system is what we lead with. This is everything underneath it: how we keep your data safe, the rules every system follows, the technology we build with, and the other systems already running in daily use.

Built to be trusted

You have to be able to trust it.

Every system we build follows these rules. You never have to ask.

Checks that stop, not just warn

If a check fails, nothing goes out. The log says which check failed.

People confirm what matters

The system suggests. Anything that spends money, changes a price or messages a customer waits for a person.

A record of every action

Who changed what, and when. Approvals are saved with the reason.

Access by role

People see only the parts of the system their job needs.

Mistakes caught before go-live

Every part is typed and tested, so a change breaks in testing, not in front of your team.

Easy to change later

Every database change is versioned and can be undone. The system can grow for years.

Security & compliance

Your data stays yours. Here is how we keep it that way.

We work inside your accounts, with the least access needed, and everything is logged. No certificates we don't hold. Just the rules we follow on every build.

We work inside your accounts

Your data stays where it is. We connect to it; we do not take a copy home.

Least access, removed at the end

Read-only keys where possible, one login per person, and every access removed at handover.

Secrets never in code or chat

API keys and passwords live in a secrets manager and are rotated when a person leaves the project.

Backups you can restore

Managed databases with daily backups. We test a restore before go-live, not after a disaster.

Every action logged

Who did what, and when, on every system we build. Approvals are saved with the reason.

Encrypted in transit and at rest

HTTPS everywhere, encrypted storage on the database, separate test and live setups.

Your data trains no AI

Where AI is part of a build, only what is needed is sent, through a logged gateway, to providers that do not train on it.

NDA first, then the data

We sign a non-disclosure agreement before we see anything, and your own if you prefer it.

What we comply with

  • India's data protection law. Digital Personal Data Protection Act, 2023 and the IT Act, 2000. A named grievance officer, plain-language notices, and deletion on request.
  • GDPR for EU and UK clients. We sign a data-processing agreement and follow it, including data minimisation and deletion.
  • Marketplace and ad-platform rules. Amazon's Selling Partner API data-protection policy and the Meta and Google developer policies. Credentials scoped to what the job needs.
  • GST invoicing. Every invoice carries GST at the applicable rate, with a proper tax invoice for your input credit.
  • Your security questionnaire. We answer it in full, in writing, before we start. If your policy is stricter than ours, we follow yours.

Straight answer

We do not hold an ISO 27001 or SOC 2 certificate. We are a four-person studio. What we do have is a written set of rules, an audit trail on every system, and a habit of answering your security questionnaire in full, in writing, before we start.

Technology

The tools we build with.

All of it in daily use in real systems.

Backend

PythonFastAPINode.js / ExpressSQLAlchemy 2.0Pydantic v2Alembic

Frontend

React 18TypeScriptNext.js / ViteTailwindshadcn/ui + RadixTanStack

Data

PostgreSQLpandasMulti-source ETLReconciliation enginesData-quality audits

AI

Claude APIMCP tool serversLLM gatewaysEmbeddings & image matchingHuman-in-the-loop

Integrations

Amazon SP-API (seller + vendor)Amazon Ads APIMeta Ads APIGoogle Sheets APIQuick-commerce buyer portalsE-mail & WhatsApp

Ops

DockerGitHub ActionsRender / Neon / VercelScheduled workersPlaywrightstructlog + pytest

Also built

The rest of the portfolio.

Not sold on their own right now, but designed and built by the same team and running in daily use today.

What we automate

If someone does it every week, software can probably do it.

This is the kind of work we automate. Each card shows the manual job it replaces and what the system does instead.

Pull your data automatically

01

Replaces · Logging into five websites every Monday to download reports.

A scheduled job signs in to your marketplaces, ad accounts and sheets, pulls the numbers, cleans them and puts them in one place.

  • Many accounts in one run
  • Slow reports are retried until they arrive
  • Old months can be filled in too

Take orders from every buyer

02

Replaces · Typing in purchase orders because every buyer sends a different file.

Each buyer's file is read automatically, checked against your stock and sent to the warehouse.

  • Add a new buyer with a settings file, not new code
  • Stock is never promised twice
  • See what you can fill before you say yes

Plan what to reorder

03

Replaces · The reorder spreadsheet nobody fully trusts.

Sales speed, stock on hand and stock on the way are turned into a forecast and a reorder plan. The plan goes to the right person to approve.

  • Propose, review and approve in one place
  • Per warehouse and per fulfilment centre
  • Warns you if the data is old

Match up your numbers

04

Replaces · Month-end hunts for why two reports don't agree.

Every run, sales are matched against ad spend, payouts against orders, and stock against the books.

  • Down to the rupee
  • Tax handled the same way everywhere
  • One meaning of 'this week' for everyone

Reports that build themselves

Start here

Replaces · An analyst's whole Monday.

Dashboards and weekly packs rebuild on a schedule. They go out only if every check passes.

  • One dashboard for all your brands and channels
  • Bad data stops the report, not just a warning
  • Runs on its own, week after week
See the full build, price and the 14-day plan

Check your product listings

06

Replaces · Someone clicking through product pages by hand.

Live product pages and photos are compared with your master list. Anything that doesn't match is reported.

  • Checks photos, not just text
  • Different rules per category
  • Catches problems before they cost sales

Keep an eye on your ads

07

Replaces · Exporting campaign reports into pivot tables.

Spend, sales and results are tracked against last period, with alerts and simple rules acting on your ad accounts.

  • Alerts when an account looks off
  • Search-term and bid signals
  • A person confirms anything that spends money

Handle customer cases

08

Replaces · Complaints tracked across inboxes and spreadsheets.

Every case gets an owner, a status and a deadline. You see what's open and what's stuck.

  • Each team sees only its own cases
  • Search by order, product or case number
  • Customers get updates by e-mail and WhatsApp

Ask your data questions

09

Replaces · Waiting a day for someone to run a query.

Ask in plain words, like "which products lost margin last month?", and get the answer from your own numbers.

  • Answers come from your database, not a guess
  • Every question is logged
  • It suggests. People decide.

Alerts and reminders

10

Replaces · Remembering to check.

Stock-outs, lost buy-box, hidden listings, price breaches and missed deadlines go to the right person by e-mail or WhatsApp.

  • Every alert says who owns it and what to do
  • Quiet when nothing has changed
  • Every message is logged

One true margin per product

11

Replaces · Five versions of the margin sheet.

Cost, marketplace fees, tax and ad spend are worked into one margin number that every other tool uses.

  • One number, worked out in one place
  • Per product and per channel
  • Feeds your pricing and deal checks

Don't see your job here? If it has steps and data, we can automate it. We go deepest on Amazon: Seller Central, Vendor Central, the SP-API and Ads API, and the quick-commerce portals. Tell us the job and we'll tell you how.

Before and after

What it looks like, start to finish.

Real patterns from systems running today, with names and client data removed. Pick one.

A weekly report that builds itself

Today, by hand

  • Reports downloaded by hand from several accounts
  • Pasted into a master workbook, formulas fixed by hand
  • Teams disagree because each counts 'the week' differently

With the system

7 accountsPull reportsClean & combine24+ checksPublish

What keeps it safe

  • Sales matched to ad spend, brand by brand, to the rupee
  • Row counts compared with last week
  • One failed check and nothing goes out

Result: Runs every Monday before anyone is in. The team starts the week reading numbers, not building them.

Who it's for

Businesses with data in too many places.

We know online retail best. The same ideas work for any team buried in exports, spreadsheets and hand-offs.

Online and D2C brands

Selling on Amazon as a seller or a vendor, on Flipkart, Myntra, Meesho, quick-commerce and your own store, with data in every portal.

Distributors and wholesalers

Orders from many buyers, stock in several places, and fill rates that matter.

Finance and operations teams

Reconciliations, margins and weekly reports held together with spreadsheets.

Growing service businesses

Case handling, approvals and internal tools that have outgrown shared inboxes.

Working with us

Terms that protect you.

01

Fixed scope, price and date

Agreed in writing before we start.

02

A working demo every week

On a test site you can open yourself.

03

Pay per milestone

You pay when an agreed demo is accepted.

04

You own the code

Code and documents are handed over as we go.

05

90 days of support

Fixes after go-live, then a yearly support plan if you want it.