Industries · HR & Talent
HR & Talent
Payroll at twenty-thousand-employee scale and compliance-native recruiting are different machines than the HR tools sold for them. We've built one, mapped the other and we're building what comes after.
The misfit pattern
- Payroll that has to reconcile biometric attendance from dozens of sites against statutory machinery (EOBI, provident fund, gratuity, income-tax slabs) for tens of thousands of employees, while a global HCM models none of it natively.
- Recruitment treated as a hiring pipeline when in the Gulf it is a government-relations workflow. A hire is not done at offer-accept but after GOSI registration, visa, medical and COC/MOFA attestation clear.
- An applicant tracking system built for dozens of CVs straining under thousands, with candidate, salary and approval data split across an ATS, a CRM and an ERP that do not reconcile.
- Offers made outside any guardrail (no salary-band enforcement, no internal-equity check, no authority matrix), so every exception is an email instead of a logged, approved rule.
- AI bolted onto an HR data model that was never built to support it. Demos that impress and pipelines that cannot actually train, explain or audit a decision.
What we build here
- Real-time payroll and attendance at national-statutory complexity: biometric capture across multi-location, multi-company structures, with EOBI, provident fund, gratuity and tax-slab logic built at the data layer rather than configured around it
- Compliance-native recruiting platforms: GOSI, visa and COC/MOFA attestation tracked through onboarding, authority-matrix exceptions modeled as system rules, the Government Relations role a first-class part of the hire
- Applicant tracking built for volume and multi-tenancy: object- and action-level access control, strict tenant isolation, bilingual English/Arabic (RTL), agency portals, white-label per client
- Offer engines with guardrails: salary-scale validation, internal-equity comparison, market benchmarking and band-breach exceptions that route through an approval chain instead of an inbox
- AI-ready by architecture rather than afterthought: RAG-ready data models, candidate embeddings and event logging granular enough that match-scoring, explainability and talent rediscovery become a phase rather than a rebuild
- Agentic HR operations: the platform we are building now, where AI runs the repetitive HR-ops load on top of a data layer designed to carry it
We have run real-time payroll and attendance for twenty thousand employees. Not as a line on a slide. It is the system one of Pakistan’s largest textile manufacturers has depended on since 2019, reconciling biometric punches from dozens of locations against the country’s full statutory machinery: EOBI, provident fund, gratuity, income-tax slabs, formula-built variable earnings and leave policies down to Hajj and Umrah. Payroll at that scale is not software you configure. It is software that has to be born knowing the rules.
Recruiting taught us the same lesson from the other direction. When we mapped what it takes to run hiring properly in the Gulf, the pipeline turned out to be the easy part. The hard part is everything a global ATS treats as someone else’s job: a hire that is not done until GOSI registration, the visa, the medical and the COC/MOFA attestation clear, all coordinated with a Government Relations Officer the system has to treat as a first-class user. Offers that cannot breach a salary band without an authority-matrix exception that is logged and approved rather than whispered. Two languages, right to left. Strict tenant isolation, because the same platform runs one company’s hiring and white-labels to the next.
That is why the HR platform we are building now is AI-first by architecture rather than by press release. Everyone is bolting AI onto HR this year. Most are bolting it onto data models that recorded nothing a model can learn from. We have seen what it costs to retrofit, so we build the RAG-ready data layer, candidate embeddings and granular event logging first, then let AI take over the repetitive HR-ops load on a foundation designed to carry it: match scoring that explains itself, talent rediscovery from the applicants you already have, screening that scales without losing the audit trail.
The arc is the point: from running payroll for twenty thousand people, to recruiting that is born compliant, to HR operations an agent can actually run. The unglamorous layer underneath (the statutory rules, the government workflows, the audit-complete events) is exactly what makes the glamorous part work. Software templated to a global average makes your HR team the integration layer. We build the layer that already knows your rules.
Common questions
Why do global HR platforms fail at Pakistani and Gulf payroll?
Labour law is the real obstacle. Each market needs its own implementation of complex and constantly changing rules. The Gulf adds real-time compliance that has to be satisfied as events happen rather than at month-end.
What makes GCC recruiting different from a standard ATS pipeline?
A hire is not done at offer-accept. It is done after GOSI registration, visa, medical and COC/MOFA attestation. An ATS that stops at "accepted" ignores the most delay-prone part of hiring.
Can payroll really run in real time at 20,000 employees?
Yes. Biometric capture across dozens of sites feeds a payroll engine that holds the statutory rules natively. We have run exactly that since 2019.
What does "AI-first HR" actually require?
A data model built for AI before the AI arrives: RAG-ready records, candidate embeddings and event logging granular enough to train on. Bolt it on later and you re-architect.
Do you support Arabic and bilingual HR workflows?
Yes. Full English and Arabic with right-to-left interfaces, because Gulf recruiting and onboarding rarely run in English alone.
Recognize the pattern?
A two-week Fit Assessment maps your specific misfit, prices it and returns the call: stay, extend or build.
Book a Fit Assessment