What we mean by a "shortcut"
Custom AI software from Nplay starts with one idea. A shortcut is a small, special program built for one job in your company. It takes a task that needs a person to read, type, check or search, and lets AI do the heavy part in seconds. Your team stays in control: they review, approve and decide. The system removes the waiting and the repetition.
Off-the-shelf software makes you change how you work. A custom system fits the way your company already works, in Arabic and English, and connects to the tools you already use.
Work we can shortcut with AI
- Quotations and proposals: the system drafts a quote or proposal from a short request, using your prices and templates, ready for a person to check and send.
- Customer questions: an AI assistant answers on WhatsApp or your website from your own information, and hands over to your team when needed.
- Documents and invoices: AI reads PDFs, invoices and forms and fills your system, instead of someone typing them in.
- Reports: daily or weekly reports built automatically from your sales, stock or operations data, with a plain-language summary.
- Internal knowledge: an assistant that answers your team's questions from your policies, manuals and past projects.
- Approvals and follow-ups: requests move to the right person automatically, with reminders, so nothing waits in someone's inbox.
What is inside a Nplay system
- A simple web dashboard your team can use from any computer or phone.
- AI connected to your own data: your prices, products, documents and history.
- Connections to WhatsApp, email, Excel or Google Sheets, your website, and existing ERP or CRM systems where they offer access.
- Arabic and English from the start, including right-to-left screens.
- User roles and permissions, so each person sees what they need.
How we build it
- Find the slow work. In a short consultation we list the tasks that take the most time and pick the one where a shortcut pays back fastest.
- Design the shortcut. We map the steps and show you how the system will look and work before we build it.
- Build with AI inside. We develop the system and connect it to your data and tools.
- Test with your team. Real users try it on real work, and we adjust it.
- Launch and measure. We go live and measure the time it saves, then decide on the next shortcut together.
Part of a bigger plan
Custom systems work best alongside the rest of your digital tools. Nplay also builds AI agents, WhatsApp AI agents, AI tools and system integration and websites and web platforms. We start from the problem in your business, then choose the technology.
Custom AI systems FAQ
What kind of company is this for?
Any company where people spend hours on repeated tasks: trading and distribution, services, real estate, clinics, schools, factories and government suppliers. If a task follows the same steps every time, it can usually be shortened.
How much does a custom AI system cost?
It depends on what the system needs to do and what it connects to. We usually start with one focused shortcut, then grow. After a consultation we send a clear quote.
How long does it take to build?
A focused first version can be ready in weeks rather than months. We agree the timeline after we understand the task and your data.
Will AI replace my employees?
No. The goal is to remove the boring, repeated part of the work so your team can spend time on customers and decisions. People stay in control and approve what matters.
Is our company data safe?
We agree with you where the system and data are hosted, who can access them, and which AI services are used, before we start. Access is controlled by user roles.
Can it work with the systems we already use?
Usually yes. We connect to Excel, Google Sheets, email, WhatsApp and many ERP and CRM systems. Where a system has no connection option, we find the safest practical way.
Custom AI software: before and after examples
The easiest way to understand custom AI software is to look at a task before and after. These are typical workflows we see in companies in Saudi Arabia and Jordan.

Quotations
Before: a salesperson copies details from WhatsApp and email into a spreadsheet, looks up prices, and builds a quote by hand. After: the system reads the request, pulls prices from your list, drafts the quote in Arabic or English, and the salesperson checks and sends it.
Daily reports
Before: a manager collects numbers from three systems every morning. After: the system gathers the data, writes a short summary, highlights anything unusual, and sends it at 8:00.
Documents and invoices
Before: staff type data from PDFs and scanned invoices into the ERP. After: the system extracts the fields, checks them against purchase orders, and only sends unclear cases to a person.
Customer requests
Before: requests arrive on WhatsApp, email and phone, and some get lost. After: every request lands in one queue, tagged by type and urgency, with a suggested reply.
HR and onboarding
Before: HR answers the same policy questions every week. After: an internal assistant answers from your own policy documents and passes sensitive questions to HR.
How a custom AI software system is built
| Layer | What it does |
|---|---|
| Inputs | WhatsApp, email, forms, files, your ERP or CRM |
| AI step | Reads, classifies, extracts, summarises or drafts, in Arabic and English |
| Business rules | Your prices, approvals, limits and exceptions |
| Actions | Creates records, sends messages, updates dashboards |
| Controls | Permissions, logs, human approval for sensitive steps |
| Interface | A simple web app, a dashboard, or a WhatsApp chat your team already uses |
Most of the value comes from the rules and connections around the AI, not the AI model alone. That is why we spend time understanding your process before writing code.
Build vs buy: when custom AI software makes sense
Off-the-shelf tools are good when your process matches how the tool works. Custom AI software makes sense when:
- Your process is specific to your industry, region or company.
- You need Arabic that understands how your customers really write.
- The task touches several systems that do not talk to each other.
- You want to own the logic and data, not rent it per user forever.
- Off-the-shelf tools would need so many workarounds that staff stop using them.
Often the best answer is a mix: keep the tools that work, and build a small custom layer that connects them and removes the manual steps.
Data, privacy and control
Companies in Saudi Arabia and Jordan rightly ask where their data goes. We design every system with clear answers:
- Read-only access first; write access only after testing.
- Hosting options that respect your policies and local data protection rules.
- Logs of what the system read, decided and changed.
- Human approval for payments, contracts, prices and anything sensitive.
- No training of public AI models on your private data.
A realistic project plan
- Discovery (1 to 2 weeks): we map one process, collect real examples and agree the success measure.
- Prototype (2 to 4 weeks): a working version with your data, used by a small group.
- Pilot: daily use, feedback, fixes and the first numbers.
- Roll-out: more users, more connections, and training.
- Support: monitoring, updates and new shortcuts as your team finds them.
Timelines depend on your systems and data, so we confirm them in writing after discovery.
What affects the cost of custom AI software
- How many steps and systems the process touches.
- The quality of your data and documents.
- Languages, user numbers and expected volume.
- Hosting requirements and security reviews.
- Ongoing support and new features after launch.
We usually recommend starting with one high-value process. It proves the value quickly and keeps the first budget small.
7 mistakes to avoid
- Starting with "we need AI" instead of a specific slow task.
- Automating a process nobody has written down.
- Skipping the people who do the work today.
- Giving the system full access on day one.
- Testing only on clean examples, not messy real ones.
- No owner inside the company after launch.
- Measuring nothing, so the value is never clear.
Why Nplay for custom AI software
Nplay has built interactive and software projects since 2017, from our headquarters in Amman, with an office in Riyadh and a development team in Cairo. We build websites, AI agents, WhatsApp assistants, dashboards and immersive systems, so we can connect AI to the channels your customers and staff already use. Nplay is a registered trademark in Sweden since 2020.
Our promise is simple: we build the shortcut. We look for the slow, repeated steps in your company and replace them with custom AI software your team actually wants to use.
Custom AI software: more questions
Do we need a big IT team?
No. We handle the build and support. You need one person who knows the process and can make decisions.
Can it work with our ERP or CRM?
Usually yes, through APIs, database connections or exports. We check your systems during discovery.
Can our staff use it on WhatsApp?
Yes. Some systems work best as a WhatsApp assistant for staff, others as a web dashboard.
Tell us one task that wastes hours every week. We reply with how a shortcut could handle it.
WhatsApp usBook a discovery call