AI Agents & Automation
Your inbox, answered before you open it
Quotation requests, supplier chasers, invoice queries, customer complaints. Most of it follows a pattern. We build an agent that reads it, drafts the reply in the right language, and routes anything unusual to a human.
2 to 4
weeks to a working first version
24/7
inbox coverage, no shift handoff
2
languages handled natively, Arabic and English
The direct answer
Email automation is an AI agent that reads your incoming mail, classifies it, and either replies directly or prepares a draft for a human to approve. It is built for teams whose inbox is the bottleneck: sales follow-up, supplier RFQs, customer support, back-office requests. It works in Arabic and English and writes into the CRM or database you already use, so nothing lives only in someone's inbox.
What this removes.
One person is the inbox
Today
A single employee reads every email, decides what matters, and types the same kind of reply forty times a week.
With the system
The agent drafts the routine replies in seconds. Your employee reviews and sends, or lets low-risk ones go automatically.
Arabic and English mixed
Today
Customers write in whichever language is convenient. Replies get delayed while someone finds the right person to answer.
With the system
The agent reads and drafts in both languages natively, matching the tone and register of the incoming message.
Nothing gets logged
Today
A quote goes out by email, gets negotiated by email, and closes by email. None of it reaches the CRM.
With the system
Every thread is summarized and written into Odoo, Zoho or HubSpot as it happens, so the record stays current.
Follow-ups fall through
Today
Suppliers or leads go quiet after the first email and nobody has time to chase them a second or third time.
With the system
The agent tracks response windows and sends timed follow-ups on your schedule, without anyone remembering to do it.
What lands in your hands.
Inbox connector
Reads your existing mailbox or a dedicated address, no migration required.
Classification layer
Sorts incoming mail by intent: quote request, complaint, follow-up, spam.
Bilingual drafting engine
Writes replies in Arabic or English matching the sender's language and tone.
Approval workflow
Drafts wait for a human tap before sending, or auto-send for rules you set.
CRM and database writeback
Logs every thread and outcome into Odoo, Zoho, HubSpot or Supabase.
Attachment and quote parsing
Extracts line items, prices and totals from PDF or Excel attachments.
Systems and platforms we work with
- OpenAI
- Anthropic
- Google Gemini
- Meta
Systems and platforms we work with
- React
- Next.js
- TypeScript
- Node.js
- Python
- Flutter
- PostgreSQL
- Supabase
- Tailwind CSS
- Docker
- GitHub
- Google Cloud
- Figma
Five stages. You sign off every one.
Read each stage as a small contract: what we need from you, what lands in your hands, and the sentence that has to be true before we move on.
- Mapping3 to 5 days
- Connect1 week
- Draft and tune1 to 2 weeks
- Pilot on live mail1 to 2 weeks
- Live operation and supportongoing
Mapping
3 to 5 days
We map the categories of email your team actually gets and agree which ones the agent should touch first.
- Share a sample of real inbound email
- Name the systems you use today
- A written intent map of your inbox
- A first-scope list, ranked by volume
We move on when you sign off on the scope before we write a line of automation.
Connect
1 week
We wire the agent to your inbox and to the CRM or database it needs to read and write.
- Grant read access to a test mailbox
- Confirm CRM field names
- Live inbox connection
- Working CRM writeback in a test environment
We move on when test emails flow through the pipeline and land correctly in your CRM.
Draft and tune
1 to 2 weeks
The agent starts drafting replies against real historical threads. We correct tone, facts and edge cases together.
- Review a batch of drafts each day
- Flag anything factually wrong
- Tuned drafting prompts per category
- An accuracy log against reviewed drafts
We move on when your team approves drafts with minimal edits, three days running.
Pilot on live mail
1 to 2 weeks
The agent runs on real incoming email with a human approving every send before anything goes out.
- Approve or reject each draft
- Tell us what still needs a human
- Daily send log
- A shortlist of rules safe to automate fully
We move on when the approval rate and error rate both hold steady for the pilot window.
Live operation and support
ongoing
The agent runs continuously. We monitor its output, retrain on new patterns, and stay on call for changes.
- Flag drift or new email types as they appear
- Monthly performance review
- Direct support line for changes and fixes
We move on when you have a running system with a named point of contact, not a handover document.
Asked before signing.
How is this priced?
Setup is a fixed project fee based on how many email categories and system connections are in scope, quoted after the mapping stage. Ongoing operation is a monthly fee covering hosting, monitoring and a support line. We do not price by email volume, so your costs stay predictable as your inbox grows.
Can it handle Arabic email properly, including scanned attachments?
Yes. The agent reads and drafts Modern Standard Arabic and Gulf business Arabic natively, not through translation. Scanned or image-based attachments go through OCR before extraction. Mixed Arabic-English threads, common in Saudi business mail, are handled without switching modes.
Does anything send automatically, or does a person always approve?
You decide. Most clients start with full human approval on every send, then move specific low-risk categories, like acknowledging receipt of a quote request, to auto-send once the accuracy log proves out. Nothing moves to auto-send without your explicit sign-off on that category.
What happens to the data, is this PDPL compliant?
Email content is processed to draft the reply and is not used to train any external model. Data residency and retention terms are set out in the contract before we connect anything, in line with PDPL. If your data must stay within specific infrastructure, we scope that during the mapping stage.
Let's look at what's actually sitting in your inbox
Send us a week of real email volume and we will tell you honestly what can be automated and what still needs a person.