Intelligent process automation · Saudi Arabia

Automating a broken process just makes it fail faster

Crux automates work that should be automated, using the simplest mechanism that will hold — and says plainly when a process needs fixing, or a rule, rather than a model.

Why automation fails

Three failures, and none of them are technical

Automation projects rarely fail because the technology could not do it. They fail for reasons that were visible before anyone wrote code, and that a two-week assessment would have surfaced.

  • The process was never documented. What exists is a set of habits held by the four people who do the work, each with their own exceptions. Automating this encodes the confusion and makes it permanent.
  • The exception rate was underestimated. A process that is 70% routine sounds like a strong candidate until the remaining 30% turns out to need more handling than the whole thing did before, because the easy cases no longer subsidise the hard ones.
  • Nobody owned it after launch. Rules drift out of date, models decay as behaviour shifts, and upstream systems change format. Six months later it is quietly wrong and everyone still assumes it works.

The corollary is uncomfortable but useful: the assessment sometimes concludes that the process should be redesigned or retired rather than automated. That answer costs a fraction of the alternative.

The automation ladder

Climb only as far as the problem requires

Each rung costs more to build, more to govern and more to explain than the one below it. The engineering discipline is choosing the lowest rung that solves the problem — not the highest one the budget allows.

  1. 01 Deterministic rules Cheapest · fully auditable

    Explicit logic a person wrote down. Eligibility thresholds, VAT calculation, routing by value. Boring, transparent, and correct every time. If the logic is knowable, stop here — a model adds cost and removes certainty.

  2. 02 Robotic process automation Cheap · brittle

    Software driving a user interface because no API exists. Genuinely useful against legacy systems, and genuinely fragile: a screen layout change breaks it silently. Treat as a bridge to an integration, not a destination.

  3. 03 Machine learning classification Moderate · needs data

    Models that learn a pattern nobody can write down: which category, which priority, which risk band. Requires labelled examples and ongoing monitoring for drift, but handles variation that rules cannot.

  4. 04 Document and language understanding Higher · needs review

    Reading unstructured content — contracts, invoices, correspondence — in Arabic and English, extracting structure. Confidence thresholds route uncertain extractions to a person rather than guessing.

  5. 05 Agentic execution Highest · needs governance

    Multi-step work across systems where the next step depends on the last. Powerful and correspondingly risky, requiring authority limits and approval gates. See our agentic AI work.

Most production automation we deliver combines rungs one and three, with four where documents are involved. Rung five is a deliberate decision, not a default.

Choosing what to automate

Five questions that rank a candidate process

Score every candidate on these before committing. A process scoring badly on reversibility should not be first, however attractive the volume looks.

Criterion Strong candidate Weak candidate
VolumeThousands of cases monthly, so a small per-case saving compoundsDozens of cases, where the build cost never returns
VariabilityCases look broadly alike, with variation in data not in logicEvery case is negotiated and genuinely different
Exception rateExceptions are a clear minority and easy to identifyExceptions are common and only recognisable by an expert
ReversibilityA wrong outcome is caught and corrected cheaplyA wrong outcome reaches a customer or a regulator irreversibly
Data availabilityThe inputs needed already exist in a reachable systemKey inputs live in someone's inbox or in an unscanned file

The most common good first candidate in Saudi enterprises is document-heavy back-office work: invoice and receipt handling, supplier onboarding paperwork, claims intake, or application triage. High volume, moderate variability, and a wrong outcome that a person catches at the next step.

Models or rules

The line most vendors will not draw for you

Selling AI means calling everything AI. But a model where a rule belongs adds cost, removes determinism, and creates an explainability problem you did not previously have. Almost every working system uses both.

Use rules when
  • The logic is knowable and someone can state it
  • The outcome must be identical every time, given the same inputs
  • A regulator may ask exactly why this decision was made
  • The calculation is arithmetic — tax, interest, eligibility, fees
  • Being wrong is unacceptable rather than merely costly
Use models when
  • The input is unstructured — documents, free text, images
  • The pattern is real but nobody can write it down
  • Rules already exist but there are hundreds and they conflict
  • The right answer shifts as behaviour shifts
  • A confidence score is genuinely useful, because uncertain cases can be routed
Explainability and oversight

An automated decision you cannot explain is a liability

Once software decides something about a person — an application, a claim, a limit — the organisation owns that decision. SDAIA's governance expectations and PDPL both bear on this, and so does the customer standing in a branch asking why.

That requirement shapes the architecture, which is the practical reason a rules layer usually survives alongside models: rules explain themselves, and models require deliberate work to do the same.

  • Reconstructable decisions — inputs, rule set version, model version, score and threshold recorded for every automated outcome
  • Reason codes people understand — in Arabic where the customer is Arabic-speaking, not an internal model artefact
  • Human review on consequential cases — with the reviewer seeing what the system saw, not just its conclusion
  • Confidence thresholds that route rather than guess — uncertainty is information, and it should trigger a person
  • Drift monitoring — automation rates, override rates and score distributions watched as operational metrics
  • Override tracking — when reviewers consistently overturn the system, that is the earliest signal something has shifted
What we build

The framework components

Decision engines

Rules and model scoring combined behind one interface, with versioning, simulation against historical cases before release, and a full record of every decision made.

Rules + ML · Versioned

Document automation

Classification and extraction from Arabic and English documents, with confidence thresholds routing uncertain cases to review rather than accepting them silently.

Arabic OCR · Extraction

Workflow orchestration

Multi-step processes with conditional routing, retry and compensation logic, and durable state — built on Airflow, Temporal or equivalent rather than a bespoke scheduler.

Airflow · Temporal

Human review workflows

The queue, the interface and the escalation path for exceptions. Usually the least glamorous component and the one that determines whether people trust the system.

Review queue · Escalation

Compliance checking

Automated validation against ZATCA e-invoicing requirements, internal policy and sector rules, flagging issues before they reach a person or a regulator.

ZATCA · Policy rules

System integration

Connection to SAP, Oracle, Dynamics and Salesforce, and to Saudi platforms including Nafath for identity, ZATCA for invoicing, and sector systems where access permits.

SAP · Nafath · ZATCA
Delivery

One process, shadow run before it acts

The shadow-running period is not caution for its own sake. It is the only way to find out whether the automation agrees with your people before it replaces them on a decision.

  1. Process assessment

    Document how the work is actually done, measure volume and exception rate, and score candidates against the five criteria. Ends with a ranked shortlist and any recommendation to fix rather than automate.

    2 weeks
  2. Design and rung selection

    Choosing the lowest rung that solves it, defining thresholds, review paths and what "correct" means, agreed with the process owner and whoever carries the risk.

    1–2 weeks
  3. Build

    Decision logic, extraction or models, orchestration, review interface and audit logging, integrated with the systems that hold the data.

    4–6 weeks
  4. Shadow run

    Running alongside the existing process without acting, comparing its decisions to your team's. Disagreements are the most valuable output of the entire engagement.

    2–3 weeks
  5. Supervised live

    Acting, with review on everything consequential, until override rates stabilise and the exception patterns are understood.

    2–3 weeks
  6. Widen

    Additional processes on the same framework, which is considerably faster once orchestration, review and monitoring already exist.

    Ongoing

A first process typically reaches supervised production in 8 to 12 weeks. Subsequent processes on the same framework move considerably faster.

Questions

Answered plainly

When should a process use AI rather than rules?

Rules when the logic is knowable and stable — eligibility thresholds, tax calculation, routing by value. Models when the input is unstructured or the pattern is real but nobody can write it down. Most working automation combines both, with rules governing anything that must be deterministic and auditable.

Why do automation programmes fail?

Usually because the process was never documented, so the automation encodes the existing confusion; because the exception rate was underestimated and exceptions now cost more than the original process; or because nobody owned it after launch and it decayed quietly while everyone assumed it worked.

How do you decide which processes to automate first?

Score candidates on volume, variability, exception rate, reversibility of a wrong outcome, and whether the data needed is actually available. High volume with low variability and reversible outcomes is the right first target. High-value irreversible decisions come later, if at all, and always with human approval.

What does explainability mean in practice?

That any automated decision can be reconstructed afterwards: which inputs, which rules, which model version, which score, which threshold. For regulated sectors and decisions affecting individuals, a reason code the customer can understand — in Arabic where the customer is Arabic-speaking — is part of the requirement, not a nicety.

Does automation work on Arabic documents?

Yes, with more attention than English. Arabic OCR quality varies with scan quality and font, handwriting remains difficult, and mixed Arabic-Latin content is common in Saudi commercial paperwork. Confidence thresholds and routing to human review matter more here than in an English-only pipeline.

How long until the first automation is live?

A single well-chosen process typically reaches supervised production in 8 to 12 weeks, including shadow running. Extending across a portfolio is faster once the framework, monitoring and review workflows exist.

Start here

Find out which of your processes is worth automating

A two-week assessment documents how the work is actually done, measures volume and exception rates, and returns a ranked shortlist — including any process we would advise fixing or retiring instead of automating.