Every AI capability in AmalOps, listed in full rather than summarised as three bullet points. All of it included with the module it belongs to, all of it explainable, and none of it acting without a human confirming.
Amal AI drafts, scores, flags, and where permitted executes — but payroll moves money irrevocably and employment decisions affect people’s livelihoods. Where the system runs payroll it acts as maker and a named human remains the checker. Every AI action is written to the audit log with its reasoning. No output is a black box, because a flag a manager cannot understand is a flag they will ignore.
The co-pilot layer that sits across every module, so people ask instead of navigating.
Ask anything in plain English or Arabic and get an answer from live data, with a Take-me-there button that opens the exact screen. Removes the most common self-service failure: the employee who opens the app, cannot find what they need, and emails HR anyway.
Step-by-step guidance for every process in the platform — running payroll, generating a WPS SIF file, configuring overtime rules, calculating gratuity, onboarding a hire, tracking document renewals, checking Emiratisation compliance, revising a salary, exporting an audit-ready report.
Say “Onboard Sara Ali as Sales Executive” and the AI creates the candidate record, attaches the UAE onboarding checklist, raises IT and PRO tasks, drafts the offer letter, and schedules the probation review — then asks you to confirm.
“Show overtime by department”, “Who is blocked from WPS?”, “Rating distribution by department” — grouped reports generated instantly with no report builder.
A one-paragraph management narrative across headcount movement, payroll cost, compliance posture and readiness, generated on demand or as a monthly digest.
The assistant proposes the questions worth asking based on what changed since you last looked, so the blank prompt is never the barrier.
Every run scanned before a dirham moves, with a readiness verdict rather than a wall of warnings.
Every payroll run scored 0–100 with a verdict — Ready to run, Review before running, or Not safe to run — deducting weighted penalties for each detected issue.
WPS submission blockers, expired documents, pay-variance anomalies, negative nets, overtime outliers, duplicate payments, ILOE lapses, Emiratisation shortfall and more, each with one-click resolution.
Each line compared against the employee’s own twelve-month history and their peer group, catching duplicates, unapproved components, and ghost employees that pass every rule-based check.
Every net change beyond a threshold explained by cause — overtime delta, salary revision, deduction — before anyone asks.
Six bank-style checks before submission: record completeness, negative nets, overtime bounds, net-to-basic sanity, IBAN validity, and expired-document exclusions.
One click fixes everything fixable — re-issuing rejected transfers, enrolling lapsed ILOE, opening renewal cases — then reports what still needs a human.
With no criticals outstanding, the AI can execute the run itself as maker, while a named human remains the checker before SIF unlocks.
A standing assessment of whether the last run completed cleanly, with overtime-cap review flags surfaced rather than buried.
A five-step tracker — processed, approved, SIF submitted, bank return reconciled, payslips published — each step auto-ticked as the system detects completion.
Signals that name a person and a next action, months before a resignation letter.
Flight-risk scores per employee, refreshed daily and weighted against your own departure history using tenure, engagement, missed check-ins, overtime load and recognition drought.
Not a departmental heat map but named individuals with their contributing drivers, so a manager can act on a specific conversation this week.
Employees who have gone too long without recognition surfaced explicitly, since time-since-last-recognition is one of the earliest reliable disengagement signals.
Unusual absence clustering flagged as a possible burnout signal — for example a Monday absence pattern rising sharply against the prior quarter in one team.
Prompts the right person at the right moment: check in with an at-risk employee, recognise someone overlooked, chase non-respondents before a pulse closes.
Every risk output carries a proposed intervention, drafted and ready to assign, because a score with no next step produces anxiety rather than retention.
Departure feedback analysed for recurring themes across leavers rather than read one form at a time.
Reading what people actually said, across languages, at pulse scale.
Open-text pulse responses grouped into themes automatically, so a thousand comments become a handful of issues you can act on.
Comments analysed across the languages your workforce actually writes in, with mood shifts and hotspots surfaced by team and site.
Isolates what genuinely moves your score — manager quality, workload, pay, progression clarity — per department rather than as one organisation-wide number.
Answers the question behind the score: why detractors are unhappy, drawn from their own words rather than inferred.
Wellbeing signals read from check-ins, workload and absence together, reported by team so pressure concentration is visible.
Representation and experience gaps surfaced across demographics, including career-clarity differences between employee populations.
A daily or weekly narrative for each manager: who needs a one-to-one, which pulse closes tomorrow, what changed in their team overnight.
Pulse questions and survey structures proposed for the topic you want to measure, rather than starting from a blank question bank.
Surveys timed to reach shift and site populations when they are actually working, which is what makes response rates comparable across a mixed workforce.
From CV to signed offer, the AI screens and drafts while recruiters decide.
Every applicant scored against the role in seconds on skills, experience, salary fit, visa status and Emiratisation impact, with the reasoning shown.
Semantic search across your entire resume database, so a new vacancy starts with people you already know rather than a fresh sourcing campaign.
High-volume applications parsed and scored together, then added as candidates — the case where volume genuinely defeats human review.
Role descriptions drafted from the requisition and rewritten on request for tone, seniority or length.
Role and candidate-specific questions generated per interview, so panels probe the actual gaps rather than working from a generic list.
A structured assessment generated from the interview record, giving panels a consistent basis for comparison.
Interview notes summarised into decisions and next steps, filed against the candidate automatically.
Panel feedback synthesised into a single view, surfacing where interviewers actually disagree rather than averaging it away.
Screens applicants, advances outreach sequences and schedules interviews while you are away, then presents a “while you were away” digest of everything it did.
A single panel governing what the autopilot may do unattended, what needs approval, and what it did last — so automation stays supervised.
Shortlists weighted against live Nafis and Tawteen targets, so quota compliance happens during hiring rather than in a quarterly review.
The paperwork layer: drafted, checked, and chased automatically.
Salary certificates, NOCs, experience letters, warning letters and offer letters drafted in correct UAE format from the employee record, audit-logged and routed for e-sign.
Offers pre-drafted against grade benchmarks and company policy, with UAE national, international hire, fixed-term and executive variants handled distinctly.
Every employee record scanned for missing documents, lapsed attestations and labour-law risks, flagged before an audit finds them.
Opens renewal cases for anything expiring within ninety days, assigns the PRO team, and escalates what has already lapsed.
Renewal workload projected months ahead, so visa and Emirates ID runs are planned rather than firefought.
Live position against target per establishment, with a forward projection reflecting approved requisitions and known notice periods.
Confirmations due surfaced with a recommendation: auto-confirm clean records, flag the rest for manual review, with reasons attached.
Coverage clashes and collisions with the payroll cut-off flagged at the point of approval rather than corrected afterwards.
Evidence assembled, bias surfaced, reports written.
Appraisals drafted from a year of actual check-ins, completed goals and peer feedback, so managers edit a grounded draft rather than starting from a blank form.
Managers rating consistently high or low, or showing recency skew, flagged ahead of calibration.
Objectives with no progress or no check-in surfaced mid-cycle, while the quarter can still be recovered.
Forward performance signals drawn from attendance, task completion and check-in sentiment, identifying who is trending down before the review.
Report outputs across headcount, engagement and helpdesk data enriched with the pattern the numbers show rather than presented raw.
Task and workload suggestions generated from what is actually overdue and who is actually loaded.
Describe the report you want in words and get the grouped output, without building a query.