We connect web shops, payment providers, POS, accounting and booking software. And we apply AI where it measurably saves time: product data, classification, summaries and internal search.
What it is
Most errors happen between systems: an order that doesn’t arrive, stock that’s off, a payment without a booking. We build integrations that make this visible and prevent it.
Who it’s for
For companies with several systems forming one process. And for teams who want to know where AI really helps, from developers who build with it every day.
What you get
Integrations & AI- 01
API integrations
Integrations with Shopify, Mollie, GuestPlan, Tebi, StockX and any system with an API. With retries, idempotency keys and a readable log per message.
- 02
Webhooks and sync
Event-driven sync through queues. A dry-run with a log first, writing only after confirmation, like the Shopify Control Center in Collett Core.
- 03
Payments
Mollie with iDEAL, card and Apple Pay. Status is always confirmed server-side through the webhook before anything becomes final.
- 04
AI in your process
Claude and OpenAI models for product copy, classification, summaries and search across your own documents. With structured outputs, validation and human review where it matters.
- 05
Data enrichment
Completing product data from external sources, like the StockX enrichment in Collett Core.
- 06
Automation
Emails, exports, reports and checks that run on their own, with cron jobs, queues and alerts the moment something deviates.
The stack
Languages, frameworks and tools we use- Languages
- TypeScriptSQLPowerShell
- Runtime
- Node.js 22FastifyNext.js route handlers
- Queues
- BullMQRedisCron
- Integrations
- Shopify Admin GraphQLMollieGuestPlanTebiStockX
- ResendReact Email
- AI
- Claude APIOpenAI APIEmbeddingsStructured outputs
- Contracts
- OpenAPIJSON SchemaZod
- Operations
- LoggingAlertsAudit trail
How we approach it
ProcessInventory
Which systems, which data, who owns it and what goes wrong today. Captured in one overview.
Contract first
Messages and fields defined in OpenAPI or JSON Schema before any code.
Build with dry-runs
Every integration can run dry first with a readable log. Only then does it write.
Monitor
Alerts on failures, an error log per message and a monthly check.
In practice
Built by JABL
The control system behind a sneaker store with three locations.
A French bistro with boules lanes: the website, and separately the system behind it.
An Italian sandwich shop that gets found, by Google and by AI.
Frequently asked
What if an integration fails?
Every integration has retries, an error log and an alert. Nothing fails silently.
Is AI safe with our data?
We use business APIs where your data isn’t used for training, send only what’s needed and keep a human deciding where it matters.
Can you take over an existing integration?
Yes. We start with an audit of what runs and what goes wrong.
Is an integration a lot of maintenance?
Very little when it’s built well: versioned, tested and monitored.
Let’s talk about your project.
One hour, no strings attached. You’ll hear from us within one working day.
Book a callinfo@jablworks.com