Custom Shopify app development, and the eight products we run.
Custom Shopify app development is what you commission when the workflow you need does not exist as something you can install — B2B pricing tiers, order routing on rules only you use, a catalogue import shaped like your supplier’s file, a margin calculation your finance lead would recognise. We build those for merchants. We have also designed, shipped and now run eight products of our own. Two of them are built on Shopify’s APIs — LineDrop and ProfitOS — and the other six were built for jobs that have nothing to do with a storefront. All eight are named and linked below, including the one that has no page on this site yet.
Every figure in the build record was read off the product page it links to — the first three on 17 August 2026, Autopilot on 19 August 2026 and F2WP on 21 August 2026 — so you can check each one in two clicks. The rest of the shelf sits under the record, with a link each and its status stated rather than implied.
- LineDropShopify Admin API
- ProfitOSShopify Admin API plus six others
- AutoBDB2B outreach. No Shopify in it.
- AutopilotMeta Marketing API and your own site
- F2WPWordPress, from a design file
- HQConnectLead capture, follow-up and booking
- VarnoSize and fit on the product page
- Fig2ShopShopify Online Store 2.0 sections
No install counts, no ratings, no revenue figures — see what this page does not claim.
The five with a page here, and what each one had to touch.
Most agency pages about app development describe a process. This one describes codebases that exist, because the useful question is not how we work — it is what a real build turned out to require. These are the five written up in full; the other three are in the list under the table.
| App | What it does | What it had to touch | Why an installable app could not |
|---|---|---|---|
| LineDropShopify Admin API, embedded in the Shopify admin | Imports a season’s catalogue from an Excel linesheet into Shopify and attaches images to the right product by SKU code, from a Google Drive folder or a ZIP. | Reads 4 columns straight out of the file the supplier sent — style name, SKU code, size and MRP — with no reformatting asked of the merchant. | The linesheet belongs to the vendor, not to Shopify. A generic CSV importer makes the merchant reshape a document somebody else controls, then attach every image by hand. |
| ProfitOSShopify Admin API plus six non-Shopify APIs | Sends yesterday’s RTO-adjusted profit to the store owner on WhatsApp at 9 AM IST, after ad spend, COGS, shipping and gateway fees. | Joins 7 external systems — Shopify, Meta Ads, Google Ads, Shiprocket, Delhivery, Razorpay and Cashfree — and reconciles returns over a rolling 30-day window. | No single vendor holds all seven. The join is the product, and the return window is specific to Indian cash-on-delivery economics rather than to global ecommerce. |
| AutoBDNot Shopify. Not ecommerce. | Finds B2B prospects, studies each one, emails the named decision-maker rather than a generic inbox, and reports who opened and replied. | 6 stages, from finding a company to reporting a reply. There is no store, no catalogue and no Shopify API anywhere in it. | Listed here because this is the path it was published on. It is business-development software for B2B service firms — see the note directly below this table. |
| F2WPNot Shopify. WordPress, from a design file. | Turns a Figma link, a screenshot or a live URL into editable WordPress sections, handed over as a plugin you install on your own site. | 3 ways in and 3 ways out — Figma, image or URL; Gutenberg block, shortcode and Elementor widget, from one generation. | What it produces is a file on the customer’s host rather than a page on ours. Every hosted builder in that category keeps the site. |
| AutopilotNot Shopify. Meta’s Marketing API and your own site. | Researches a market, writes the ad angles and creative, and builds the Meta campaign — paused. It can pause spending by itself and can never start or raise it without an approval. | 16 jobs running today and 9 that are not, each of the 9 named with the credential or approval it is waiting on rather than folded into the feature list. | The guardrail is the product. An installable tool that optimises spend has to be trusted with the budget; this one is built so that it cannot start spending at all. |
Each row is a summary. The long version of how LineDrop imports a linesheet, of how ProfitOS reconciles RTO-adjusted profit of what AutoBD does end to end of what Autopilot will and will not do with a budget and of where an F2WP rebuild stops sits on each product page, and every claim in the table above can be checked against one of them.
Eight products, counted one way: a link each.
One row per product that has a page on this site or a live address of its own. That is the whole rule, so the number is something you can settle yourself by counting the rows and opening them, rather than something you have to take from us.
Until 21 August 2026 this page said four, and named four. It was wrong, and it was wrong on the page whose entire argument is that you should go and check. The three it left out were live the whole time.
| Product | What it runs against | Where you can check it |
|---|---|---|
| LineDrop | Shopify Admin API | A page here |
| ProfitOS | Shopify Admin API plus six others | A page here |
| AutoBD | B2B outreach. No Shopify in it. | A page here |
| Autopilot | Meta Marketing API and your own site | A page here |
| F2WP | WordPress, from a design file | A page here · f2wp.headlinehq.in |
| HQConnect | Lead capture, follow-up and booking | A page here · leadgen.headlinehq.in |
| Varno | Size and fit on the product page | A page here |
| Fig2Shop | Shopify Online Store 2.0 sections | No page here yet — a written deep dive, and the live app · fig2shop.headlinehq.in |
OutreachOS is not a ninth row. The deep dive we published under that name points at leadgen.headlinehq.in, and that address is already in the table above, listed under the name the site itself uses: HQConnect appears in its page title, its comparison table and its pricing, and the word OutreachOS appears nowhere on it. We have not published anything about how the two names relate, so this table counts addresses rather than names. The deep dive says the same thing above its first link. Read 23 August 2026.
Two of the eight are Shopify apps. This page is named after those two.
AutoBD is business-development software for B2B service firms: it finds ideal prospects, studies each one, emails the real decision-maker with a pitch written for that prospect, and reports who opened and replied. Autopilot is an AI performance-marketing system a business runs on its own Meta ad account. F2WP builds WordPress sections out of a design. HQConnect and Varno are outreach and fit software. None of those five touches a Shopify API.
They sit on this path because it is the record of software this studio designed, built and runs, and that is a more useful thing to publish than a page which quietly drops the six that do not fit the heading. If you came here for Shopify work, LineDrop and ProfitOS are the two builds that answer your question, and the build record marks which is which.
Autopilot is also the one place on this site where advertising comes up, so it is worth saying here: it is a product a client operates themselves. We do not run anybody’s ad account.
Three times you should not commission a custom app.
We sell this work, so read the next three paragraphs with that in mind. They are still the three reasons we most often tell a merchant to keep their money.
Somebody already sells it
If a few hundred other stores need the same thing, a subscription written by a company that supports it full time will beat anything commissioned once and maintained never. Pay for the subscription.
The rule changes every month
Logic that moves wants a surface you can edit without a deploy. Shopify Flow, n8n and a webhook will serve you better than a codebase you have to book time against.
Nobody has agreed to own it after launch
A custom app is not a deliverable, it is a dependency. Somebody has to answer when Shopify deprecates an API version. If that person does not exist yet, the build is premature.
An app, or an automation?
The two overlap so heavily that a custom app often is the automation, so we draw the line on what gets delivered rather than on what it achieves.
An app is a thing we hand over
It has a codebase, a version number and usually a screen someone logs into. You own it, and somebody has to maintain it when Shopify moves an API version. That is this page.
An automation connects what already exists
Shopify Flow, n8n, webhooks and the Admin API, wired so a job runs and reports. No new product is left behind. That work lives on the Shopify store automation page.
The short test: if it needs a screen someone signs into, it is an app. If it needs a job that runs on a schedule and tells you what happened, it is an automation.
Three commitments we make before any code is written.
Read-only against your real data first
The first build reads and writes nothing. It is the same standard we hold our AI agent work to: prove the numbers against production data, get them checked, and only then let anything write back.
Your repository from the first commit
The code sits in your GitHub organisation from commit one, not in ours, and not delivered as a zip at the end. If the engagement stops, you keep a working codebase and its history.
One workflow, named in a sentence
If the scope cannot be said in one sentence a merchant would repeat to their operations lead, it is not one app. It is two, or it is an automation, or it is not ready.
If the app has to run across a time-zone gap, the terms we publish for that are on the working with an Indian Shopify agency page rather than buried in a proposal.
What this page does not claim.
There are no install counts on this page, no merchant counts, no ratings and no revenue attributed to any of the eight. Nobody independent has audited those numbers, and a figure we cannot show you the instrument for is worth less than the blank space it would fill.
What is here instead: what each one does, which systems it had to join, and why an installable app could not do the job — taken from the product pages, the first three read on 17 August 2026, Autopilot on 19 August 2026 and F2WP on 21 August 2026. Follow any link on this page and check the claim against the page it came from.
The count is a claim too, and it is the one this page got wrong for longer than any of the others. It is now a rule you can apply yourself — one row per product with a page or a live address — rather than a number we assert.
Questions worth asking any app developer.
What counts as a custom Shopify app?
Software we design, build and hand over that runs against your Shopify store through Shopify’s APIs — usually with a screen someone logs into, always with a codebase and a version number. It is what you commission when the workflow you need does not exist as something you can install: B2B pricing tiers, order routing on rules only you use, a catalogue import shaped like your supplier’s file, a margin calculation your finance lead would recognise.
What actually decides the cost?
We do not publish a figure for custom app work, because scope drives it almost entirely and a number without a scope is a number designed to be wrong. Three things move it most: how many external systems the app has to join, whether it writes back into Shopify or only reads, and whether it needs an interface a person uses or just a job that runs and reports. ProfitOS is expensive on the first of those three; LineDrop is expensive on the third.
Do you build on the Admin API or the Storefront API?
Both, for different jobs. LineDrop writes products and variants, so it lives on the Admin API inside the Shopify admin. Shopper-facing surfaces on a headless build read through the Storefront API. Which one your app needs follows from whether it changes data in the back office or presents it to a customer.
How many products are there, and how is that counted?
Eight, and the rule is one row per product that has a page on this site or a live address of its own. Both are things you can open, so the number is checkable rather than asserted. Until 21 August 2026 this page said four and named four, which was wrong while three of the others were already live. OutreachOS is not a ninth: the article published under that name points at leadgen.headlinehq.in, and that address is already in the table, listed under the name the site itself uses, HQConnect. We have not published anything about how the two names relate, so this table counts addresses rather than names.
Which of the eight are actually Shopify apps?
Two. LineDrop runs on the Shopify Admin API inside the Shopify admin, and ProfitOS runs on the Admin API plus six non-Shopify services. The other six have no Shopify surface at all: AutoBD is B2B outreach software, Autopilot is performance marketing on a Meta ad account, F2WP builds WordPress sections from a design, HQConnect handles lead capture and follow-up, and Varno does size and fit on a product page. They appear here because this page is the record of what the studio has built and runs, and leaving six products off their own parent page is how a site ends up publishing a number that does not survive being checked.
Does Autopilot mean you run ads for clients now?
No, and that distinction is the whole point of it. Autopilot is software the client runs on their own ad account with their own budget. Every ad it builds is created paused, the only action it takes by itself is pausing something that is losing money, and anything that starts or increases spend waits for an explicit approval. We build and automate stores; we do not buy media, and there is no version of this where we hold the card.
Who owns the code when the project ends?
You do, and you own it throughout rather than at handover. The repository is yours from the first commit, and the written handover covers what the app does, what breaks it and who to call.
Describe the workflow. We will tell you whether it needs an app.
Send us the thing your team does by hand every week. If an app you can install already solves it, we will name the app and say so — that answer costs you one email and saves you a build.