Our Services
One product, deployed to fit the organisation. What varies is the scope, the material behind it, and how far it is allowed to reach into your systems.
One product, deployed to fit the organisation. What varies is the scope, the material behind it, and how far it is allowed to reach into your systems.
What we deliver
The assistant itself
Chat, voice, document understanding, multiple agents and human handover — in Arabic and English, on infrastructure you control.
Knowledge onboarding
Turning your policies, procedures and product material into something an assistant can answer from accurately. This is where most of the real work is.
System connections
Read-only, scoped connections to the systems you already run, so answers reference a real order, ticket or invoice rather than a general policy.
Agent design
Scope, tone, limits and escalation rules written per agent and per language — including the list of what it must never say.
The live-agent portal
Where your team picks up escalations with the full conversation already in front of them, so nobody is asked to repeat themselves.
Measurement
A baseline before launch and the metrics that show whether it is working — including the ones that do not flatter the system.
How a deployment runs
- Scope it narrowlyOne audience, one channel, the twenty questions that actually dominate your contact volume. Taken from last month's tickets, not from memory.
- Settle governance firstWhere data sits, how long it is kept, who can see what, and how a person exercises their rights. Before the first conversation, not after.
- Bring the materialPolicies and procedures rather than marketing pages. Answer quality follows source quality more closely than anything else you control.
- Write the limitsThe list of what it must never discuss. A thirty-minute exercise that prevents the category of incident that ends projects.
- Launch narrow, read everythingOne channel first, and read the actual conversations weekly. Every deployment discovers a question nobody predicted.
- Widen on evidenceThe next agent or connection is added because the measurement supports it, not because the plan said month three.
What we do not do
We do not replace your contact centre, and we will not sell a project on that promise. The assistant removes the repetitive layer and leaves judgement, complaints and difficult cases with the people who were always best placed to handle them.
We do not build you a platform. If you have an engineering team that wants to own every component, a global cloud platform serves you better than we do.
We do not run projects without material or an owner. A grounded assistant cannot find what does not exist, and material with no named owner goes stale quietly.
Common questions
With the twenty questions that dominate your contact volume, on one channel. It gives most of the value at the smallest risk surface.
Not to begin with. A great deal works from your documents alone. Connections are added when a question genuinely needs live data.
Yes. That is the deployment this was designed around.
Current material, a named owner for it, and a decision on retention. The technical setup is the short part.