Access, accounts and roles
Sign-up, login and invitations. Each customer is an account and the role decides what each user sees.
Firebase Auth / NestJS / PostgreSQL
SaaS development: from the first version to paying customers
Digitup Studio is a technical studio in Madrid that designs, builds and runs custom SaaS: users and roles, Stripe subscriptions, usage limits, an admin panel and AI with controlled costs. We start with the smallest version that can already charge customers and take it to production.
In short
- Digitup Studio builds complete SaaS products: access and roles, Stripe or PayPal subscriptions, usage limits, an admin panel and deployment.
- It developed Rescrito (formerly Aithor), an international SaaS with real users, recurring payments, plans and usage limits.
- It picks the AI model (OpenAI, Gemini, Anthropic, DeepSeek or Perplexity) based on quality, cost, speed and privacy.
- Projects start from €1,000 and delivery from 2 weeks; a complete SaaS usually goes beyond that minimum.
- You pay as a fixed-scope project, over 6 or 12 months, or with Build & Run, a monthly fee that includes operations.
- It is run by its two partners, Stas Nikitin (CEO) and Anton Zverev (CTO); the studio has developed more than 50 projects.
What does a SaaS need to charge customers from day one?
Six pieces: access with roles, plans and subscriptions, usage limits, an admin panel, onboarding with emails, and analytics with support. Everything else can wait.
Plans and subscriptions
Sign-up, plan changes and cancellation. Stripe issues invoices, collects payment and retries failed payments (Stripe documentation); your product grants or removes access based on the subscription.
Stripe / PayPal / Webhooks
Usage limits per plan
Each plan sets how much can be used: users, documents or AI queries. That protects your margin.
Redis / Firestore
Admin panel
Customers, plans and usage at a glance; raise limits or block accounts without touching the database.
Angular / Backoffice / Logs
Onboarding and emails
A path to the first useful result and the automatic emails: welcome, receipts and failed payment.
Mailgun / Templates
Analytics and support
Sign-up, activation, payment and cancellation measured, and a clear support channel.
Events / Email / Chatbot
What goes into the first version of a SaaS, and what can wait?
Just enough for a customer to sign up, use the main flow and pay. The rest, once users ask for it.
| Module | In the first version | Once there is traction |
|---|---|---|
| Access and roles | Email or Google and two roles: admin and member | SSO, two-factor authentication and custom roles |
| Plans and payments | One to three plans in Stripe | Free trial, coupons, pay-per-use and PayPal |
| Usage limits | One counter per account on whatever costs you money | Credits, top-ups and usage alerts |
| Admin panel | Customers, plan and usage; block accounts | Business metrics and exports |
| Analytics and support | Key events and email support | Funnels, cohorts and a help centre with a chatbot |
| AI, if there is any | One model per task and one limit per plan | Several providers, caching and RAG over documents |
General guide: the real scope is decided when the solution is defined.
Where do I start if I don't have customers yet?
By checking that someone will pay to have the problem solved, not by building the whole product.
With no customers yet, what helps is an MVP: the main flow and billing connected, ready to show to real users and measure. If the technical foundation is solid, that MVP becomes the first version of the SaaS. More in MVP development for startups.
Before writing code, answer four questions: who pays, what task you save them, how much they would pay per month and what they use today. If only your own company will use the tool, what you need is custom software.
How do you add AI to a SaaS without it eating your margin?
With three decisions made at design time: choose the model for each task, limit usage on every plan and measure AI cost per customer.
Every query to a model costs money depending on the provider, the model and the text processed; with unlimited use on a fixed-price plan, a few heavy users eat the margin. In Rescrito, the academic writing SaaS we developed, the multi-LLM approach lets you choose a provider by task, latency or price, and every plan has usage limits.
A model per task. A cheap one for classifying or summarising; a more capable one where quality justifies it.
Limits or credits per plan. Whoever needs more usage moves up a plan.
Bounded context. With RAG, the model receives only the relevant fragments of your documents.
Per-customer measurement. You see which accounts are profitable and adjust prices with data.
For conversation, an AI chatbot fits; for taking action, AI agents. Cost and usage control for LLMs is covered in AI integration.
What architecture and stack do we use for a SaaS?
Typically Angular on the frontend, Node.js with NestJS or Python with FastAPI on the backend, and PostgreSQL, MongoDB, Redis or Firestore for data, depending on the product and who will maintain it.
If there is a mobile app, Ionic is added. Rescrito uses Angular and Firebase (Auth, Firestore and cloud functions) with its own APIs, Stripe, PayPal and Mailgun. More in technologies.
A SaaS keeps each customer's data separate using one of three multi-tenant models: a shared database with a customer identifier on every record, one schema per customer, or one database per customer. The first is usually enough to start and is the cheapest to operate; the others are justified if a customer demands more isolation.
From the start there is CI/CD, logs, monitoring, backups and access control. If you process personal data on behalf of your customers, Article 28 of the GDPR requires a contract or other legal act governing the processing; review it with your legal adviser, because we do not give legal advice.
How much does it cost to build a SaaS, and how long does it take?
Digitup projects start from €1,000 and delivery from 2 weeks, but a SaaS with billing, an admin panel and deployment usually goes beyond that minimum.
The price depends on the user types, the billing model, the integrations, the AI and whether a mobile app is needed. We do not publish rates: we review the case, propose a technical approach and phases and, if it fits, start with a first version, which can usually be validated within weeks if the scope is clear.
Add up third-party costs (hosting, AI APIs, emails and payment fees) before setting your prices. Payment options: ways of working.
Who runs the SaaS once it is in production?
You decide: you can keep the code and run it with your own team, or let us run it with Build & Run.
We have built and run our own products, such as AlisaBot, and we maintain platforms in production, such as Mallorcab. We know what comes after launch: users who do not understand something, servers, payments, support, bugs and changes in APIs and AI models.
Build & Run is a predictable monthly fee covering hosting, security, support and an evolution allowance, with a minimum commitment normally of 12 or 24 months and a purchase option defined from the start; AI and other intensive services carry agreed limits. In a fixed-scope project you receive the software and access credentials on completion, and you can continue with our software maintenance.
How do we build a SaaS?
We understand
the business
Who pays, why and how much. If it is not worth building, we tell you.
We define the
technical solution
Architecture, data separation, plans, integrations and phases.
We build a
first version
Main flow, billing and a minimal panel, ready for real users.
We integrate
and scale
AI, emails, automations and whatever the first customers ask for.
We launch
and maintain
Deployment, CI/CD, monitoring, support and evolution.
Which SaaS products and platforms has Digitup built?
Rescrito (formerly Aithor) is an international SaaS with recurring payments in Stripe and PayPal, plans and usage limits. AlisaBot is our own product for building chatbots; SacaNota and ClipToTask are AI platforms we have developed.
SaaS / EdTech / Generative AI
AI SaaS for academic writing
View project
Chatbots / AI / Integrations
Platform for building chatbots and integrating AI assistants
View project
EdTech / AI for students
AI platform for studying and producing academic content
View project
Productivity / AI / Screen recordings
From screen recordings to tasks and bug reports with AI
View projectWhen is Digitup a fit for your SaaS, and when is it not?
We fit when there is a product to build and run.
When we fit
- You are going to sell the software to others, by subscription or by usage.
- You need plans, limits, an admin panel and automatic billing.
- You want AI in the product without losing control of the cost per customer.
- You want to deal with the people who decide: it is run by the two partners.
When we are not the right choice
- An off-the-shelf SaaS, already configured, is enough and nobody needs to write software.
- You want to rent individual developers, without a product brief.
- The budget is below €1,000 or delivery has to happen in under 2 weeks.
- Your customers require certifications such as ISO 27001 or SOC 2: Digitup does not have them.
Related services
A SaaS usually touches several of these areas.
Frequently asked questions about SaaS development
How much does it cost to build a SaaS?
Digitup Studio projects start from €1,000, but a SaaS with plans, recurring billing, an admin panel and deployment usually comes in above that minimum. The price depends on the user types, the billing model, the integrations, the AI and whether a mobile app is needed. We do not publish rates: before quoting a price we review the case and propose a technical approach and phases.
How long does it take to launch a SaaS?
At Digitup, delivery starts from 2 weeks, and a working first version can usually be validated within weeks if the scope is clear. Above that minimum, the timeline depends on the project: complex integrations, several user types or a mobile app add phases. Arriving with plans and prices already decided helps, because pending decisions slow down any development.
Do I need to build a SaaS, or is an off-the-shelf one enough?
If an off-the-shelf tool solves your problem with just some configuration, you do not need to build anything, and we will tell you so. Your own SaaS makes sense when you are going to sell the software to other companies or people, by subscription or by usage. If the tool is only for your own company, what you need is custom software: there is no need to charge for plans or to separate the data of different customers.
How do you stop AI from driving costs up?
We choose the model for each task, set limits or credits on every plan and measure AI usage per customer. A cheaper model is enough for classifying or summarising; a more capable one is reserved for where quality justifies it. With RAG, the model receives only the relevant fragments of your documents. That way AI spend stays tied to what each customer pays.
Who owns the SaaS code?
It depends on the engagement model. In a fixed-scope project, the software and access credentials are handed over when the project is completed. With flexible payment, the investment is spread over 6 or 12 months and ownership passes with the last payment. With Build & Run, the software belongs to Digitup during the service, with a purchase option if the project includes one; its terms and amount are defined from the start.
Can you improve a SaaS that already exists?
Yes. Digitup Studio steps into projects that are already built to tidy up the architecture, improve performance, add AI, automate processes or scale features. We also fix architecture or infrastructure problems on existing platforms. As with a new product, we first review the case and propose a technical approach and phases; if it fits, we can then continue with maintenance and evolution of the product.
Can the SaaS also have a mobile app?
Yes. Digitup builds iOS and Android apps with Ionic and Angular, which can share the backend, users and payments with the web version of the SaaS. In Mallorcab, for example, a customer app, a driver app and an operational backoffice live side by side, with driver payments and subscriptions. If at first you do not need to be on the App Store and Google Play, a PWA may be enough to validate the product.
Where is Digitup Studio based?
Digitup Studio is a Madrid studio founded in 2019, based at Calle Alcalá 211. We work mostly remotely and, if the project calls for it, we meet in person. We have clients in Madrid, elsewhere in Spain and on international projects, and we can work in Spanish, English or Russian. The first call lasts 30 minutes, is with Anton Zverev, our CTO, and carries no commitment.
Do you have a SaaS idea, or a product that needs to start charging?
Tell us who would pay and why. We will tell you how we would approach it and, also, whether we think it is not worth doing. The first call lasts 30 minutes, with Anton, and carries no commitment.