01 · AI produces
Standard code, tests and documentation are produced with AI, under the direction of an engineer who specifies each piece of work.
Business software, built to order
Custom AI application development in Switzerland, reviewed by senior engineers
When no software on the market fits the way your business works, Djtal, the operational AI specialist for Swiss companies, builds the application. Our engineers develop with AI, a senior engineer reviews every line, and the code we deliver belongs to you, in your own Git repository. Lead times come down to weeks and the project costs less, with engineering quality unchanged. Zoho Creator and Zoho Catalyst are our preferred platforms. We also build applications independently of any platform.
You own the code
Delivered to your Git repository, documented and tested · end the contract and the application stays with you
EU data sovereignty
Amsterdam + Dublin · ISO/IEC 27001 · SOC 2
Your data stays out of AI training
No client data is used to train AI models, under a contractual clause checked on every project
Proven in production
Applications running in production at FRC, OLF and Terre des hommes
In brief
What is a custom business application with AI built in?
A custom AI application is business software built for one company when no product on the market fits its process. Djtal, the operational AI specialist for Swiss companies, builds these applications in Switzerland: AI writes the code, a senior engineer reviews every line, and you own the result in your Git repository, on Zoho Creator and Catalyst or standalone.
How we build
Built with AI, reviewed by senior engineers.
Our engineers have been developing with AI since 2025. This website, our 25 AI agents in production and our internal tools were all built that way. The gains have been measured, and they come with conditions. Productivity rises by 16 to 30% in teams that have reorganised how they deliver, and routine tasks take up to 46% less time (McKinsey, 2025). Work on a well-scoped task was completed 55.8% faster (controlled study by GitHub, 2023). You get shorter lead times and a lower project price, at the same standard of engineering.
Less flattering studies exist too, and we cite them. Experienced developers working with AI outside any method were measured 19% slower (METR, 2025). Code duplication rose eightfold (GitClear, 2025), and 45% of generated code proved vulnerable when it went unreviewed (Veracode, 2025). Our method counters each of these risks with senior review, automated tests and continuous integration.
02 · A senior engineer reviews
Every line goes through senior review, automated tests and continuous integration. ‘Almost right’ code is stopped before it reaches production.
03 · You own it
The code we deliver is standard, documented and tested, and it lives in your Git repository. If you end the contract, the application stays with you, in working order.
Four ways to build an application, compared
Each approach has its place. We use this comparison in every scoping exercise, including when it leads us to recommend a platform in place of a custom build.
| Approach | What it gives you | Its limits |
|---|---|---|
| Full-code | It gives full performance and complete freedom, with software shaped exactly to your business. | It is costly to build and complex to maintain, and it depends on the developers who wrote it. |
| Low-code | It costs far less and moves quickly on standard needs (forms, workflows, reports). | It performs less well, hits a ceiling on complexity and comes with a platform subscription that runs for life. |
| No-code | It is enough for a form or a prototype put together in a few hours. | For a real business application it is an illusion, and it soon hits its limits on access rights, volumes and integrations. |
| AI-assisted development | It brings custom-built performance at lower cost, in weeks. AI produces the code, and a senior engineer specifies and reviews it. | It demands a team that has mastered the method: systematic review, testing and continuous integration. |
As an Authorized Zoho Partner since 2017, we sell low-code too. When Zoho Creator is enough, we recommend Creator. When it reaches its limits, we build a bespoke application. The scoping work shows which, with figures to back it up.
Two platforms, one team
Creator and Catalyst complement each other, from low-code to pro-code.
We combine these two Zoho platforms depending on the problem at hand. Creator covers business applications that need to go live quickly. Catalyst comes in when you need custom development, AI connected to your processes, or logic too unusual for low-code.
Zoho Creator
Low-code, with a business app ready in a few weeks
Creator models a business process that off-the-shelf software leaves out. Forms, screens, automations, reports and portals go up quickly on a platform already in production at FRC, OLF and Terre des hommes, natively integrated with your Zoho CRM, accounting and inventory.
- A quick start, with the first screens in a few weeks
- Native links with CRM, Books, Inventory and Desk
- Secure portals for customers or employees
- Room to grow without a full rebuild
When your business fits the ‘form → workflow → report’ mould.
Zoho Catalyst
AI-native custom development, for work that breaks the mould
Catalyst is Zoho's serverless pro-code platform, with Zia AI services built in (OCR, NLP, vision, prediction, generation). It supports fully custom development (APIs, microservices, functions, orchestration), integrated with the Zoho ecosystem or used on its own when you need a platform without assembling every component separately.
- Native Zia AI services (OCR, NLP, vision, prediction, generation)
- European hosting (Amsterdam and Dublin)
- Native integration with Zoho One, yet usable on its own
- No infrastructure to manage (serverless, DevOps-free)
When you need complex business logic, operational AI or a solid API to expose to other systems.
Why Catalyst, why now
Serverless pro-code, AI-native and hosted in Europe.
Catalyst started life as a serverless platform. Zoho then built its Zia AI services into it, covering OCR, NLP, computer vision, prediction and text generation. Today it is a professional development environment in which AI is a platform primitive, available from the first line of code. That suits a company that wants a custom application without assembling its own AI infrastructure or depending on the large US cloud providers.
01 · Native Zia
Catalyst code calls the Zia AI services directly, among them OCR, text classification, vision, AutoML and ConvoKraft. With a single platform, the API key, the billing and the architecture all stay with Zoho.
02 · EU data sovereignty
Catalyst runs in European data centres (Amsterdam, Dublin) and holds ISO/IEC 27001 and SOC 2 certification. Your data stays within the European area, a point Swiss companies can make to their own customers.
03 · Integrated or independent
Catalyst works natively with CRM, Creator, Books and Desk, and it can also be used on its own, as a PaaS. One platform replaces a stack of AWS, Vercel, OpenAI and a DevOps team.
Djtal has been working on Catalyst since 2021, from the platform's earliest versions, and now has seven applications in production on it, among the first Catalyst deployments in Switzerland. We know the platform's strengths and pitfalls in detail. Find out more about Zoho Catalyst.
Apps in production
Three client applications in three different sectors.
FRC
Fédération romande des consommateurs (consumer association for French-speaking Switzerland)
FRC works with four major applications built on Creator and Catalyst, among them a fast search tool for member services and an application on its website for professionals that manages comparative tests.
Creator + Catalyst
OLF
Book distribution and publishing
A tailored, secure B2B portal and a Creator application that extends CRM and Desk serve the customer service team, with AI-assisted automated pre-analysis and categorisation of tickets.
Creator + AI services
Terre des hommes
Humanitarian NGO
A Creator application brings together children's files, medical and social follow-up, volunteer management and the organisation of convoys to hospitals, with secure access for external parties.
Creator + external user portal
The most common trigger
Replacing the Excel file that runs a critical process.
Replacing a shared Excel file with a business tool connected to your ERP or CRM is a few weeks' work at Djtal, with role-based access rights, a change history, validated data entry and integration with the systems already in place. Scoping starts with a 4-hour initial audit (CHF 800).
The symptom is always the same. A shared workbook runs a vital process (scheduling, case tracking, pricing calculations), its versions drift apart, and nobody dares touch the formulas any more. The application that replaces it keeps the flexibility of the original tool and adds what Excel lacks: several users at the same time, access rights, a change history and a connection to your CRM or accounting system.
Which one, when
Creator, Catalyst or both?
The rule stays flexible. It depends on how specific your business is, how much logic needs coding, how much AI you need and what budget you have. These pointers guide our first assessment.
Creator alone
When the business fits the ‘data entry → workflow → report’ pattern, as with registration management, standardised case tracking or a conventional customer portal. It gets going quickly, and the cost stays under control.
Catalyst alone
When you need a solid API to expose, a business microservice or AI processing in the background, without a form-style interface. It also suits needs that are entirely independent of Zoho One.
Both
This is the most common case in our projects. Creator handles the screens, forms and business workflows, and Catalyst the advanced logic, AI, APIs and custom integrations. Our largest deployments (FRC, OLF) are built on this architecture.
The next stage of the path: when a process is worth running continuously, the application hands over to a supervised AI agent, designed to order and run over the long term.
Price and terms
Published rates, set by level of expertise.
A custom app calls for several kinds of specialist. The hourly rate depends on the level of expertise each task requires.
Initial audit
CHF 800
4 hours · CHF 200/hour
We scope the project, check feasibility, recommend Creator, Catalyst or both, and give you an initial budget estimate. For an overall view before you build, there is the AI strategy audit.
Architect
CHF 180/hour
CHF 1,440/day
The architect designs the architecture, makes the technical choices, models the data and carries out senior code review.
Senior development
CHF 160/hour
CHF 1,280/day
Senior engineers handle Creator and Catalyst development, integrations, testing, go-live and documentation.
All prices are in CHF and exclude VAT; 8.1% Swiss VAT is added. Well-scoped projects can be billed as a fixed price per deliverable. After go-live, ongoing maintenance and improvements are billed at the senior development rate, with a flexible monthly commitment.
Frequently asked questions
What is a custom AI application?
A custom AI application is business software built for your company when no product on the market fits your process, with artificial intelligence designed in from the start. Djtal develops it on two complementary Zoho platforms, Creator (low-code, delivered in weeks) and Catalyst (AI-native serverless pro-code). Djtal can also deliver it independently of Zoho.
Do we need to be on Zoho for custom development?
No. Creator and Catalyst integrate natively with the Zoho ecosystem, and Catalyst also works on its own as a serverless platform. If your starting point is AI alone or another environment, Djtal builds the application from scratch, with or without an existing system.
What is the difference between a custom application and an AI agent?
A custom application is software your teams use actively (screens, forms, workflows). A supervised AI agent runs a process continuously, in the background, under human control. Djtal treats the two as complementary, and its AI agents page covers that side. The application is often the foundation, and the agent the next step.
How much does a custom AI application cost in Switzerland?
At Djtal, everything starts with an initial audit: CHF 800 for a 4-hour audit covering scoping, feasibility, a recommendation between Creator and Catalyst and an initial budget estimate, with no obligation to continue. Development is then billed by level of expertise (architect at CHF 180 an hour, senior development at CHF 160 an hour), or as a fixed price per deliverable for well-scoped projects. Prices exclude 8.1% Swiss VAT.
Can we replace an Excel spreadsheet with a business application?
Yes. It is the most common project among Djtal's Swiss clients. The application takes over the workbook's logic (data entry, calculations, lists) and adds what Excel lacks: several users working at the same time, role-based permissions, a change history and a connection to your ERP, CRM or accounting system. Allow a few weeks for the project, scoped by a 4-hour initial audit at CHF 800.
Do you develop with AI, and what difference does that make for us?
Yes. Djtal's engineers develop with AI, under systematic senior review, with testing and continuous integration. Djtal built this website and its in-house agents the same way. For you, that means shorter lead times and a lower project price, with standard, documented code delivered to your Git repository, which you own.
Do your applications use our data to train AI?
No. Djtal keeps all client data out of AI model training, and this clause is checked on every project. On Catalyst, hosting is European (Amsterdam, Dublin), certified to ISO/IEC 27001 and SOC 2.
Getting started
Every custom application we build starts with an initial audit.
The audit takes 4 hours. We scope your project, work out the right combination of Creator and Catalyst and hand you an initial budget estimate. It costs CHF 800, with no obligation to continue.
Custom applications are door 5 of Djtal's AI path: six services, one path.
Further reading: Bottle returns: a 100% automated process · The dream of a clean inbox, now a reality
Last updated: