Retail and restaurant operations managers
Branch and area managers who need fast answers across stores without building reports by hand.
Meet Orchestra · AI analytics for POS
AnalyticMena brings AI analytics to your POS data, led by Orchestra, its conversational AI command center. Ask a business question in Arabic or English, watch it pull from sales, cashiers, inventory, and procurement, and get a finished PDF, Excel, or DOCX back. Schedule it to send every morning and Orchestra waits for your confirmation before anything runs.
Orchestra listening
Query
Sales performance from 1 May to 10 May by store
4
Data sources
PDF XLS DOCX
Artifacts
Logged
Audit state
The command center
Type or speak a business question and Orchestra figures out what you mean, pulls the right data from your sales, cashier, inventory, and procurement systems, writes an executive summary with KPIs, and hands you a finished PDF, Excel, or DOCX. Ask it to send that report every morning and it shows you a confirmation card first.
Orchestra never creates a scheduled task silently. Every recurring or sensitive action stops at a confirmation gate, and nothing runs until you approve it. The model proposes; it cannot act on its own. Add a visible run lifecycle, full audit trails, permission checks before every tool call, and three execution paths that trade speed for flexibility, and you get a command center rather than a chatbot bolted onto a dashboard.
Sales performance
1–10 May · by store
482,000
Net sales
9,120
Orders
52.8
Avg ticket
Type the question. Orchestra extracts the dates, stores, and grouping and routes it to the right data. You skip the filter menus and the SQL.
Every run ends in a downloadable PDF, Excel, or DOCX with an executive summary, KPI cards, tables, insights, and recommendations. Nothing to copy, paste, or reformat for the board meeting.
Scheduled deliveries, emails, and alerts all stop at a confirmation card. The AI proposes the task; you approve it. Every confirmation and cancellation is logged.
The live progress stream shows understanding, planning, permission checks, data fetching, analysis, and artifact generation. Long jobs feel shorter and failures point you to a recovery action.
Shipped capabilities, built on advanced AI and large language models, with the backend executing every action and persisting every run.
Ask business questions in plain Arabic or English, like "sales performance from 1 May to 10 May by store." Orchestra reads intent, extracts filters for date, location, and grouping, and routes the query to the right modules. An inline voice button lets you speak the question instead of typing it.
One question can pull from sales, cashier and payment, inventory, procurement, and profitability at once. Ask for "hourly store sales, cashier performance, and available stock" and Orchestra fetches and combines the sources instead of you assembling them by hand.
An Intent Classifier picks the route: a Fast Path for known reports that returns in under a second, an Agent Path where a planner and executor chain registered tools for multi-step work, and a Generation Path where the model writes Python in a sandboxed container with a self-fix loop for one-off analysis.
As a request runs, the interface streams each step over SSE: queued, understanding, planning, checking permissions, fetching data, analyzing, generating response, generating artifact, completed. Redis pubsub drives it, with DB polling as a fallback when Redis is unavailable.
Orchestra produces PDF, Excel, and DOCX reports with a standard structure: title, date range, filters used, executive summary, KPI cards, detailed tables, key insights, recommendations, and a generated timestamp. Empty-data cases return a clear message instead of a broken file.
Ask Orchestra to send a report on a schedule and it builds a confirmation card showing the task name, action type, report type, recipients, delivery channel, schedule, timezone, start date, and filters. The task is created only after you click Confirm. Nothing is scheduled silently.
All scheduled tasks live in one panel filtered by status: active, paused, failed, completed, cancelled, draft, with pause, resume, run now, cancel, and archive. Open execution, delivery, and audit logs per task. Every run event, confirmation, and permission check is logged with timestamp and actor.
Governed by you
Orchestra never schedules or sends anything silently. When you ask for a recurring report, it proposes a confirmation card showing the task name, report type, recipients, channel, schedule, timezone, start date, and filters, then waits. The task is created only after you click Confirm, and every confirmation and cancellation is logged. The model proposes; you decide.
Operators get ad-hoc analysis at conversational speed without waiting on IT or learning the report catalog.
One question spans sales, cashiers, inventory, and procurement, so you stop stitching spreadsheets together to see the whole picture.
The confirmation gate, audit logs, and per-user permissions mean scheduled reports and emails only happen on purpose, with a record of who approved what.
Once confirmed, daily, weekly, and monthly reports reach the right inboxes on schedule, with delivery logs you can check and retry.
Arabic and English input and responses, with the broader suite supporting RTL layout for MENA teams.
Branch and area managers who need fast answers across stores without building reports by hand.
CFOs and accountants who want recurring KPI reports, reconciliation views, and an audit trail.
Analysts who run ad-hoc, multi-dimensional questions and need exportable, board-ready deliverables.
Stock controllers and buyers tracking variance, waste, transfers, and supplier performance alongside sales.
Leaders who want cross-functional insight, and IT buyers who prioritize permissions, audit compliance, and multi-tenant governance.
Part of the POSMena platform
Coming after Phase 1. These are planned, not yet available.
Orchestra will assemble live dashboards from your questions, so a recurring view builds itself instead of being wired up by hand.
A multi-agent layer where specialized agents take on longer-running, multi-step jobs across modules under the same confirmation and audit rules.
Scheduled reports delivered to WhatsApp alongside email, each one still passing through the confirmation gate before it sends.
Orchestra is the conversational AI command center at the core of AnalyticMena. You ask a business question in plain Arabic or English, and Orchestra routes it to the right POS data, analyzes it, generates a downloadable report, and can schedule recurring delivery after you confirm.
No. Orchestra never schedules a task silently. When you ask for recurring delivery it shows a confirmation card with the schedule, recipients, channel, timezone, and filters, and creates the task only after you click Confirm. Every confirmation and cancellation is logged.
Orchestra accepts and responds to queries in Arabic and English. The broader AnalyticMena suite also supports right-to-left layout for Arabic and additional interface locales.
PDF, Excel, and DOCX. Each report follows a standard structure with a title, date range, filters used, executive summary, KPI cards, detailed tables, key insights, recommendations, and a generated timestamp.
An Intent Classifier picks one of three paths. Known reports take a Fast Path for sub-second results. Multi-step requests use an agent planner and executor that chain registered tools. Unique analysis takes a Generation Path where the model writes Python that runs in a sandboxed container with a self-fix loop. The model plans and reasons; the backend executes every action.
Yes. A live progress stream over SSE shows each step of the run: understanding, planning, checking permissions, fetching data, analyzing, generating response, and generating the artifact, with DB polling as a fallback if Redis is unavailable.
Orchestra checks permissions before every tool execution. A per-user permission model scopes which reports, stores, and data levels each user can reach, with roles such as superuser and organization admin, and a full audit trail across runs, confirmations, deliveries, and permission checks.
Yes. You can upload CSV, Excel, PDF, DOCX, and TXT files into a conversation. Orchestra extracts metadata such as column, row, and page counts and makes the content available for analysis in the same query context.
Bring AnalyticMena to your POS data and let Orchestra do the routing, analysis, and reporting. Ask in Arabic or English, get a finished artifact back, and keep every scheduled action behind the confirmation gate.