Artificial intelligence
AI that works with the systems you already run.
Artificial intelligence is Datapoint's newest line of work, built on our experience with information systems. We support public and private organisations across the whole adoption cycle — focused on administrative processes, the systems already in place and the people who use them.
What we do
Services designed for results, not for demos.
-
Assessment and roadmap
We assess processes, systems, data and maturity, and deliver a roadmap prioritised by impact and risk.
- Opportunity map
- Personal-data assessment
- 90-day plan
-
MCP integration
We connect AI models to your systems with secure, auditable and reusable MCP servers.
- Databases, ERP, DMS, CRM
- Per-user permissions
- Human approval for writes
-
Assistants for teams
Internal assistants that search, summarise and prepare work, integrated with Microsoft 365 or your intranet.
- Search across systems
- Drafts and reports
- Teams integration
-
Assistants for citizens and customers
24/7 support with grounded answers, real request status and hand-over to a person when needed.
- Answers with sources
- Bookings and requests
- Accessible and multilingual
-
Document automation
Extraction, classification and routing of documents — invoices, applications, referrals, submissions.
- PDF and scan reading
- Validation against rules
- Workflow integration
-
AI training and literacy
Training for leaders, officers and IT teams, aligned with the AI literacy obligation in the AI Act.
- Responsible use
- Prompt engineering
- Governance and risk
-
Hosting and operations in the EU
We operate models, MCP servers and monitoring in the European Union, through our partners, or on your premises.
- 24/7 monitoring
- Security updates
- Usage reports
-
Support and evolution
Ongoing tracking of indicators, better answers and new connectors as needs grow.
- Direct contact with the technical team
- Regular reviews
- Shared roadmap
Methodology
Start small, measure, and only then expand.
-
Assessment
Processes, systems and data mapped; use case chosen with agreed indicators.
-
Design
MCP tools, permissions, personal data and human approval points defined.
-
Proof of concept
A working assistant connected to test systems and evaluated with real users.
-
Production in the EU
Implementation with European hosting, monitoring, auditing and security.
-
Evolution
Training, indicator tracking and new processes.
Responsible AI
The rules we follow on every project.
Trust is built with concrete choices, from day one.
- Human oversight: no decision affecting people is taken by AI alone.
- Transparency: people know when they are interacting with AI and where each answer comes from.
- Minimisation: AI accesses only the data it needs for the task.
- Processing in the EU: data and models within Europe, through our partners.
- Measurement: quality, errors and impact tracked with agreed indicators.
Pricing
A predictable pricing model, always with a cap.
No unpredictable invoices. The cost of AI usage is measured in the proof of concept, before you commit to operation.
-
Implementation per platform
A fixed price for each connected system: MCP connector, permissions, testing and documentation.
- Fixed scope and price
- The same price for new platforms
-
Monthly operation
A fixed monthly fee per platform: EU hosting, through our partners, monitoring, updates and support.
- Security updates
- Usage reports
-
Usage with a cap
Bundles of requests or processed documents, sized on the actual consumption measured in the proof of concept.
- Alerts before reaching the limit
- You never pay more than the agreed cap
For public bodies
A clear way to get started, within public procurement rules.
We have worked with regional public administration since 2018, under contracts awarded through open tenders and direct awards.
- A free AI literacy session for senior staff and technicians — generic, without analysing the body's processes, and aligned with the AI Act's literacy obligation.
- Assessment and proof of concept with defined scope, deliverables and price, procured through the Public Contracts Code procedure that applies to each value.
- Start with one platform and extend to the rest at the same unit price, when that option is provided for in the initial procedure.
- Payment after acceptance, against criteria set out in the specifications.
- Transparency: our public contracts can be consulted on the BASE portal (in Portuguese).
Free literacy session
A first session to understand what AI can and cannot do.
Price per platform
Fixed amounts that help estimate the cost and choose the procedure.
Planned extension
Every new platform reuses what has already been built.
People in control
No decision affecting people is taken by AI alone.
Frequently asked questions
About our services.
How long until we see results?
A proof of concept on a well-chosen process shows results in 3 to 6 weeks, against indicators agreed at the start.
How do public bodies procure your services?
Through the Public Contracts Code procedure that applies to the value of each phase. We define scope, deliverables and a fixed price per platform, and the option to extend to the remaining platforms can be provided for in the initial procedure. Our public contracts are on the BASE portal (in Portuguese).
How is the cost of AI usage calculated?
AI models charge for the volume of text they process, which varies with usage and data sources. Instead of reselling that consumption to you, we measure it in the proof of concept and propose a usage bundle with a cap: you receive alerts before the limit and never pay more than agreed.
Can we use our own AI model account?
Yes. If your organisation already has a contract with a model provider, we connect the solution to that account and charge only for implementation and operation.
Do we need an in-house technical team?
It is not required. We handle implementation and operations; where there is an in-house team, we work together and transfer knowledge.
Where would you like to start?
An assessment, a proof of concept or just a conversation — choose the first step.