SIRAJسراج
Nabil Mobile/3 branches, Amman /Engagement 001
0 systems connected Demo — simulated data
Step one — connect

Siraj reads what you already run.

Read-only. No migration, no new system for your staff to learn.

POSPoint of sale18,412 sales · 14 months
TKTRepair tickets3,120 tickets
STKBranch stock sheets3 branches · live
WAWhatsApp Business2,847 threads
PDFSupplier invoices412 documents
Step two — the problem, in your words
No specNo ticketNo dashboard to configure
Step three — investigate

Working the data like an engineer on site.

Reasoning — live
Records read
0
Sales
0
Repairs
0
Messages
Repair turnaround — days to close
1d 14d average 6.2 days
Step three — testing what it thinks

It checks its own answer before giving it to you.

Hypothesis 1 — not enough technicians.
Tested against 3,120 tickets: technicians were idle 41% of working hours. Capacity is not the constraint.
Rejected
Hypothesis 2 — repairs are waiting on parts.
63% of total repair time is a ticket sitting open, waiting for one component to arrive. Screens account for most of it.
Holds — kept digging
Hypothesis 3 — the part was already in the building.
Cross-referenced branch stock sheets against order dates: 38% of parts ordered from the supplier were in stock at another branch that same day.
Confirmed
The finding you didn't ask for
3.1×

Customers whose repair took more than four days were 3.1× less likely to buy their next phone from you.

The slow repair isn't costing you repair margin. It's costing you the next sale — which is roughly nine times larger.

● Estimated JOD 47,000 / year
How it got there — cross-system join
repair tickets sales history whatsapp threads
Repeat purchase within 12 months
No dashboard was ever going to ask this question.
Step four — build

Working software, not a slide with a recommendation.

parts_router.py — new
Shipping in this fix
+Cross-branch parts lookup, before any supplier order
+Per-part reorder points from real failure rates
+Day-3 WhatsApp update with a loaner offer
+Owner alert when a ticket passes 4 days
Built against your constraints

Runs on the sheets and WhatsApp number your staff already use. No new app for the technicians. Arabic messaging, sent from your own business account.

Step five — backtest

Replayed against 14 months of your own history, before touching anything live.

3,120 historical tickets, re-run through the fix
Amber = would have closed in under 3 days
Projected turnaround
6.2d
Projected repeat purchase rate
31%
Flagged for a human

14 edge cases where the loaner offer shouldn't fire — warranty disputes and water damage. Excluded pending your call.

Step six — your call

Nothing ships without you.

Step seven — deploy

Live in your business in under four minutes.

Parts router deployed to all 3 branches
Reorder points written into the stock sheet
WhatsApp templates approved and scheduled
Owner alerts routed to Nabil's phone
Measurement started — baseline locked at 6.2 days
Step eight — measured, 30 days later
Repair turnaround — the KPI it promised to move
6.2d
deployed ● Verified against your own data
Repeat purchase rate
31%
Revenue recovered — month one
Annualised: JOD 74,000
What Nabil did

Approved one change. Then went back to running his shop.

And then it gets smarter

Every engagement teaches the next one. Your data stays yours.

Shared layer — patterns only, never data
Cross-location stock blindnessv4 · learned here
Service delay → churn couplingv2 · learned here
Reorder point driftv7
Queue congestion at batch releasev3
Unprompted — engagement 002

The same pattern is running in your accessories stock.

Siraj found it on its own, using what it learned here. Estimated JOD 19,000 a year. Want it to take this one?

Start engagement Not now
Idle t + 00:00 Read-only · nothing written without approval
SIRAJ
Tell it the problem. It ships the fix.
sirajai.com
Connect 0:00
space play/pause · ← → chapters · 1–4 speed · R restart · H clean view · F fullscreen — “Record video” captures the full run and downloads it