Skip to content
One minuteWhat would we build for you?Pick your sector and what to fix first. See the three agents we would start with, and try one.Get your first planStart · 01Custom AI solutionsAny problem, diagnosed first.TimeframeDiagnosis in 2 weeks, build in 6 to 12You keepA working solution, its code and its documentation.View solutionStart · 02Model selectionThe right LLM for each agent.Timeframe2 to 4 weeksYou keepAn evaluation set and a model plan you can re-run.View solutionStart · 03Security and privacy reviewBefore any agent touches data.Timeframe2 to 3 weeksYou keepA signed-off security design for every agent.View solutionStart · 04AI training and enablementTeams that run agents.TimeframeSessions over 4 to 8 weeksYou keepTeams that run and extend your agents.View solutionStart · 05Araby.ai platformArabic models, ready to build on.TimeframeLive in daysYou keepArabic-first tools today, and a path to your own agents.View solutionBuild · 06Custom agents and systemsBuilt from scratch, for your process.TimeframeMVP in 8 to 12 weeksYou keepAgents, source code and runbooks.View solutionBuild · 07AI Customer ServiceEvery channel, one answer.TimeframeLive in 4 to 8 weeksYou keepOne agent across WhatsApp, web, app and voice.View solutionBuild · 08Voice agentsCalls answered, in Arabic.Timeframe6 to 10 weeksYou keepA voice agent on your numbers, with transcripts.View solutionBuild · 09Vision agentsImages, scans, and video, understood.Timeframe6 to 10 weeksYou keepA vision agent tuned to your images.View solutionBuild · 10Analytics agentsForecasts, churn, and anomalies.Timeframe6 to 10 weeksYou keepModels, alerts and the data pipeline behind them.View solutionRun · 11Agentic workflowsMulti-step work, end to end.Timeframe6 to 12 weeks per workflowYou keepWorkflows that run themselves, with a full log.View solutionRun · 12ERP and CRM integrationAgents that write back.Timeframe4 to 8 weeks per systemYou keepTested connectors you own.View solutionRun · 13API and database toolsSafe read and write access.Timeframe2 to 6 weeksYou keepA tool catalogue your agents share.View solutionRun · 14Legacy systemsConnected, even without APIs.Timeframe4 to 10 weeksYou keepAdapters that outlive the agents.View solutionRun · 15Agent monitoringTracked, measured, alerted.TimeframeLive from launchYou keepDashboards, alerts and traces.View solutionRun · 16Private deploymentOur cloud, yours, or on-site.Timeframe2 to 8 weeks, depending on the optionYou keepAgents, models and keys inside your walls.View solution

Start · (02) · Model selection

The right modelfor every agent.

We test open-weight and commercial models on your own tasks, in Arabic and English, and pick per agent for accuracy, cost, speed and where the model is allowed to run.

Timeframe

2 to 4 weeks

You keep

An evaluation set and a model plan you can re-run.

(02)

Model selection

The problem

Public leaderboards measure English trivia, not your Arabic forms. One model for everything is either too expensive for simple tasks or too weak for hard ones.

What we do.Step by step.

(01)

Build an eval set

Real tasks from your teams, in both languages.

(02)

Test the candidates

Open-weight and commercial, side by side.

(03)

Weigh the trade-offs

Accuracy, cost per task, latency, hosting.

(04)

Route per agent

The right model for each job, swappable later.

An evaluation set and a model plan you can re-run.

(02)

What you get

  • Arabic and English evaluation set
  • Benchmark report
  • Model routing plan
  • Cost forecast per agent

Timeframe · 2 to 4 weeks

Where it fits

Start, build, run. Each stage ends in something you own.

  1. (01)Start · Find where agents pay back
  2. (02)Build · Agents that finish the job
  3. (03)Run · Agents inside your systems
How we work →

Questionspeople ask.

Yes. We include open-weight models you can host yourself.

Bring us one process. We will show you the AI working.

A one-hour meeting with our team, on a real example from your organization.