AI & AUTOMATION PRODUCT MANAGER RIYADH / KSAOPERATOR-BUILT PRODUCT THINKING
ABDELRHMAN REFAAT / AI & AUTOMATION PRODUCT MANAGER

AI products that
move the business.

Owning AI and internal-platform products at PetroApp after nearly four years building at Taager—across autonomous agents, operational systems, and executive automation.

CAREER BRIEF / VERIFIED
30-SECOND READ
PROVEN ANNUAL VALUE CREATED$155K+
AI + AUTOPRODUCT
01Find
02Frame
03Build
04Measure
05Scale
01CONVERSATIONAL AI02MULTI-AGENT SYSTEMS03INTERNAL PLATFORMS04BI + AUTOMATION

Operator context.
Product discipline.

I started inside warehouse, logistics, and CX workflows before moving into product. Today I apply that operating context to AI agents, fueltech platforms, executive automation, and measurable business outcomes.

QC ANALYSTOPSPRODUCT MANAGERLEAD AI PMAI + AUTOMATION PM

Every product has
a number attached.

These are post-launch outcomes from products owned end to end—from diagnosis and strategy to rollout and iteration.

ANNUALIZED VALUE / USDPORTFOLIO TOTAL
$155K+
01AI recovery agent$72K
02CX platform migration$48K
03CTWA acquisition agent$25K
04AI support agent$10K
$0$36K$72K
CONVERSION+15%

CVR lift from an AI WhatsApp storefront.

UNIT ECONOMICS50×

Cheaper address acquisition per order.

$0.50 BEFORE$0.01 AFTER
OPERATIONS+30%

Same-day warehouse shipout improvement.

THROUGHPUT
SHIP VELOCITY13+

Products moved from ambiguity to production.

Four bets.
Four changed systems.

Each case study shows the uncomfortable middle: constraints, product decisions, experiment design, launch mechanics, and measured results.

FULL PORTFOLIO13 shipped products
AI SYSTEMS6 in production
DEEP PROOF4 case studies
Open the complete impact dashboard

From messy workflow
to trusted autonomy.

A practical system for building AI products: grounded context, bounded action, observable decisions, and economics that survive production.

STAGE 01 / DIAGNOSE

Find the expensive truth

Map the workflow, pull the data, and isolate the failure pattern before proposing AI.

Failed-delivery calls exposed a $72K annual automation opportunity.
RAG + AGENT ARCHITECTURE
TRACEABLE BY DESIGN
KNOWLEDGE
01Policies
02Product data
03Conversation history
GROUND + REASON
RETRIEVAL
ORCHESTRATOR
TOOLS + ACTIONS
01CRM update
02WhatsApp response
03Human escalation
EVALS + GUARDRAILSBUSINESS TELEMETRYFAILURE LEARNING

90 days from signal
to scalable value.

A roadmap is a sequence of de-risking decisions—not a feature calendar. Click each phase to inspect the product logic.

PHASE 01 / DAYS 00—30

A ranked opportunity map and one measurable AI bet.

  • Shadow the operation
  • Model unit economics
  • Define the eval set
Problem confidence90%
OPERATOR → PRODUCT MANAGER → AI PRODUCT LEAD
“AI is infrastructure, not a feature. Start with the workflow that should exist if intelligence were native from day one.”
01

Start with business pain

The first artifact is an economic model, not a prompt.

02

Design the whole system

Model, data, workflow, humans, and incentives are one product.

03

Instrument the outcome

Quality metrics and business metrics must move together.

04

Ship to learn

Shadow, stage, observe, then expand the autonomy boundary.

AVAILABLE FOR THE RIGHT PROBLEM

Let’s build a system
that compounds.

Open to AI product leadership, consulting, and ambitious conversations about automation in MENA.

Connect on LinkedIn
Riyadh, Saudi ArabiaRESPONSE WINDOW / 24—48H