Starter
$19/mo
A single store that needs visibility: 30 days of events, twenty diagnoses a month.
opsdeck
Incident Intelligence
Find why orders, webhooks and fulfillment break — fix them faster.
Every webhook, order and fulfillment becomes a readable timeline. When something breaks, the AI incident engineer explains what happened, why, and what to do — using only the events that actually happened.
Live on the Shopify App Store
The problem
Stuck orders, failed webhooks, fulfillment that never moves, inventory that won't sync, a third-party app that quietly drops a tag: anyone running a Shopify store next to an OMS, ERP or WMS burns hours reconstructing what happened, jumping between the admin, logs and three other dashboards. opsdeck is built for the people who get paged at 2am.
How it works
- 01
Incident
Something breaks: an order stalls, a webhook fails, a fulfillment never starts, a tag disappears without anyone noticing.
- 02
Evidence
Every webhook and operational event is normalised into one readable, timestamped technical timeline, per order.
- 03
Diagnosis
The AI incident engineer explains the most likely cause using only the events in the timeline, and lowers its confidence when the evidence is thin.
- 04
Resolution
An operational playbook: how to verify, the suggested fix, risk level, affected systems and the next best action.
What it does
- Resolution playbook: probable cause, evidence, how to verify, suggested fix, risk and next action
- Timeline per order: a readable technical history, in the order it actually happened
- What changed: compare before and after an incident to surface the change that caused it
- Blast radius and history: how many orders share the same failure, and whether it happened before
- Alerts on stuck orders, failed webhooks, fulfillment delays and payment or sync anomalies
In numbers
- 11
- webhook topics handled, business and GDPR
- 3
- distinct AI surfaces, each with its own strict JSON schema
- 5
- alert rule types, from stuck orders to sync anomalies
- 30/90/180
- days of event retention, depending on the plan
The product
The engineering call
The resolution playbook isn't prose — it's strictly typed JSON. Incident title, probable cause, a confidence from 0 to 1, evidence entries each tied to a real event on the timeline, hypotheses with their missing evidence spelled out, how to verify, suggested resolution, risk, affected systems and next best action.
The system prompt is explicit: the model may only use the evidence it was given, may not invent missing events, and must lower its confidence when the evidence is insufficient. The marketing promise — every diagnosis is backed by events that actually happened — is verifiable in the code.
The store's operational profile (tools in use, carriers, tag conventions, volume and timing baselines) is built deterministically from the event stream at zero token cost, then condensed into a brief injected into the prompts: diagnoses become specific to that store instead of generic. That's a deliberate cost-control decision, not a detail.
- Remix
- NestJS
- Polaris + App Bridge v4
- OpenAI
- BullMQ + Redis
- Prisma + PostgreSQL
- Event store
- Tailwind v4
Pricing
Pro
$99/mo
Growing stores that can't go dark: 90 days, two hundred diagnoses, advanced alerts.
Agency
$199/mo
Teams running many stores: 180 days, unlimited diagnoses, priority support.
Find why orders, webhooks and fulfillment break — fix them faster.
A Barikreativa product: design, architecture and full-stack development in house. Published on the Shopify App Store.