Find your answer.

The assessment is fixed-scope and quoted before it starts. The findings report is yours to keep either way — it is written to be useful to another supplier if you choose one.

Usually, yes. Most engagements split responsibility explicitly: your team keeps what it runs well, we take what it doesn't want at 3am, and the boundary is written into the SLA.

We deliver from Dubai, London, Doha and Islamabad. The contracted cover window, severity levels and response targets are defined for the service rather than assumed from the office locations.

Ask us directly and we will tell you plainly — including where we have not. Sector references are shared in conversation, not printed on a website without permission.

It is one product with role-based experiences. AI voice agents, human agents, supervisors, contacts, campaigns, follow-up and analytics work on the same operating surface. Prime Alley also designs and operates the telephony and integrations beneath it.

No. It removes repetitive demand and gathers structured information. Your approved escalation rules decide which conversations always go to a person, and the handover includes the transcript, summary and known context.

They describe overlapping jobs. Apex can answer approved routine enquiries, gather structured details, complete connected actions and route exceptions to your team. Hours, languages and the exact use cases are confirmed during call-flow discovery.

Yes—the deployment identifies the AI clearly. Disclosure wording and the route to a person are agreed for the journey and jurisdiction rather than left to an unstated assumption.

Whatever you authorise: opening hours, bookings, status checks, structured intake. The escalation matrix you sign defines it explicitly, and everything else routes to your team with context.

Language and dialect coverage is confirmed against your caller mix and use cases during discovery. We do not treat a generic language label as proof that a specific accent, vocabulary and call condition is production-ready.

Yes. Campaigns can apply pacing, concurrency, retry, business-hours, time-zone, suppression and budget rules. The applicable consent and Do Not Call requirements must be confirmed for each market before launch.

Yes. The white-label and multi-tenant model supports partner branding, customer separation and scoped analytics and billing. The commercial and operating responsibilities are agreed during partner onboarding.

Usually. We assess the current carrier, number type and regulatory conditions first, then document whether the number can be ported, forwarded or connected through an approved trunk before any change is made.

Usually, but it depends on the carrier, number type and local rules. We document whether the safest path is porting, forwarding or an approved trunk before any commitment or change is made.

Often. We first measure the path, capacity, quality-of-service policy and failure points. The report separates configuration fixes from circuit, hardware or architecture changes so replacement is not the default answer.

Yes. The design standardises monitoring, addressing, configuration and escalation while preserving local carrier and regulatory requirements at each site.

The secondary path is tested under an agreed scenario, the result is recorded and the runbook states who makes the failover decision and how service returns to normal.

Apex includes its own contact, company, pipeline, task and project records and can exchange approved customer, conversation and task data with external systems. The exact fields, actions and connection method are confirmed during discovery.

Not automatically. We first establish whether the problem sits in the source data, definitions, pipeline or reporting layer. Existing tools stay where they remain the right fit.

Yes. Conversation, queue and outcome data can form part of the governed model alongside CRM and operational sources, subject to the access and retention rules agreed for the deployment.

You do. Definitions, lineage and operating documentation are included so your analysts can maintain and extend the reporting layer.

Retention, access roles, export and deletion requirements are agreed before go-live and reflected in the deployment design. Regulated or sensitive workflows receive stricter human-handover and evidence controls.

The institution defines the lawful purpose, access roles, retention period and export requirements. Those rules are reflected in the technical design and tested before production use.

Residency is treated as an architecture requirement. We first confirm the authority's rule, the proposed service providers and the actual storage and processing locations; the written design records any gap rather than assuming compliance.

The deployment design documents AI disclosure, the approved interaction record, human-escalation routes and the decision rules available for oversight review. Exact recording and transcript handling follows the authority's policy and applicable requirements.

Yes. You define what the AI may resolve alone and what always reaches a person. The escalation rules are configuration you approve, and transcript or recording handling follows the policy agreed for the deployment.

Not in the operating model we recommend. It can provide approved factual information and gather structured details, but advice, disputes, vulnerable-customer scenarios and material decisions route to authorised staff.

Compliance requirements are captured in the discovery workshop and carried into the architecture as controls, not bolted on before an audit. Evidence is collected continuously.

Yes. Discovery maps the call flow, data movement and operational responsibilities to the obligations and internal controls you identify. Prime Alley does not claim that software alone makes an institution compliant.

With the posture assessment — one to two weeks, no agents installed without your change process, and a remediation roadmap ordered by what an attacker would actually use first.

Only where a tool is failing you. The roadmap works with what you own where it is effective; rip-and-replace is a recommendation of last resort, and it will say why.

Yes — and the goal is that the next one is boring. Evidence collection runs continuously so audit week is retrieval, not reconstruction.

The public page does not yet publish approved seat, usage or service rates. We still make the commercial model explicit: seats, AI-handled minutes, messaging and integrations for Apex; estate and SLA for managed services; written scope for projects. A scoping call produces a budget range before any paid assessment.

Yes — after one scoping call we will give you an honest bracket, including whether the project is worth doing at all. The written figure comes with the assessment, and it is the figure we stand behind.

A useful first estimate needs your agent count, a recent month of inbound and outbound call volumes, average call duration, the channels you want to operate, the systems Apex must integrate with and any recording, retention or residency requirements. If some figures are missing, we identify the assumptions rather than hiding them inside the quote.

Metered alongside the seat subscription, with usage visible in Apex before it reaches an invoice. The quote defines each applicable unit, included allowance and any overage rate. Budget alerts, automatic pauses and team allocations are agreed during scoping where they apply.

Phone numbers, inbound or outbound carrier usage and third-party messaging charges may be passed through where they apply. The commercial schedule separates these from the Apex platform, AI usage and implementation services so procurement can see who charges for what.

Projects are fixed-scope agreements. The available monthly or annual options for Apex and managed services are confirmed in the quote. Exit assistance, data export and deletion responsibilities are written into the applicable contract before service begins.

The assessment gives you a wave-by-wave schedule. Small estates move in weeks; a multi-country estate is sequenced over months so no single cutover carries the whole risk.

Every wave has a rehearsed rollback with a named decision owner and a hard go/no-go gate before traffic moves. Rolling back is a planned path, not an emergency.

Whichever the workload economics say. The target-architecture report prices both against your actual usage, and hybrid is often the honest answer for regulated data.

Usually not. The service is designed around the work your team wants to retain and the work it wants covered after hours or under a firmer operating discipline. Responsibility is recorded in the onboarding matrix.

The contracted cover window, severity definitions, response targets and escalation contacts are stated in the SLA. A proposal should never use 24/7 as shorthand without defining what actually happens at 3am.

Restore tests run on an agreed schedule. The result, recovery time and any corrective work are included in the service record rather than inferred from a successful backup job alone.

Every handover carries the current hypothesis, evidence, next action and rollback state in a common incident format. The process is measured at the seams between offices, not only inside each one.

That is the default plan. Assessments run any time; anything touching production is scheduled into term breaks, with enrolment-week freezes written into the change calendar.

Admissions enquiries, fee and payment questions, appointment booking and status checks — with anything sensitive or safeguarding-related always routed to a person, by rule.

Under the same residency and access controls as the rest of the estate: calls are logged, transcripts access-controlled, and the escalation matrix keeps clinical matters with clinicians.

It detects urgency and routes those calls to staff immediately — it does not give clinical advice. That boundary is a rule you approve, not model behaviour we hope for.

Yes, when the required order and policy data is available through an approved integration. The design states which information may be returned automatically and which exceptions require a person.

We model the expected call and digital-channel shape, test overflow and failure scenarios, and document the staffing and AI boundaries before the event rather than scaling by guesswork on the day.

That is the target operating model. Customer, order and conversation context can be passed into the human task, subject to the systems and access rules agreed during integration discovery.

No. The design separates what must be common—reporting definitions, security, runbooks and ownership—from what must remain local, such as numbers, languages, carrier arrangements and regulatory requirements.

Yes. The roadmap groups locations by dependency and risk, with a service gate and rollback path for each wave rather than a single global cutover.