What AI solutions can .brutalis implement?

Assistants over your documents, semantic search, data extraction from invoices and forms, request classification, summaries and transcripts, agents and automations with human approval, plus AI features in apps. Each project starts from a use case tested on your data.

Do you train your own AI models?

No. We integrate existing models, through an API or hosted on your infrastructure, and fit them to your process through documents, instructions and checks. If a case needs fine-tuning, we discuss it separately.

Is my company’s data safe?

We send the model only what is necessary, sign data processing agreements with providers and, where the option exists, ask that data is not used for training. For sensitive data we can use models hosted on your own infrastructure.

What does the AI Act mean for my company?

Regulation (EU) 2024/1689 requires, among other things, that you tell people when they are talking to an AI system (art. 50) and that you have staff with adequate AI literacy (art. 4). High-risk cases have much stricter requirements. We help you classify the project correctly.

Can AI make mistakes?

Yes. We reduce the risk with answers grounded in documents with cited sources, tests on real examples, confidence thresholds and human approval for important decisions. We do not recommend unsupervised AI where errors have serious consequences.

How much does an AI implementation cost?

It depends on the case, the data volume and the integrations. The cost has two parts: the implementation and the model’s usage cost, which we estimate up front. You get a written quote, per stage, starting with a measured prototype.

AI in your business processes, measured and under control.

Document assistants, data extraction, automations and agents with human approval, implemented with tests, visible costs and respect for the GDPR and the AI Act.

Get a quotePrices

Artificial intelligence is worth using where it saves real time: answers to repeated questions, search across documents, reading invoices and orders, sorting email, summaries, preparing quotes. Together we pick a use case, test it on your data and decide on the numbers whether it is worth it.

We use existing models, through an API or hosted on your infrastructure; we do not train foundation models. Our work is integrating them into processes, with checks, limits and a person who decides when it matters.

What we implement

Conversational assistants on your site or in apps that answer from your documents and cite their sources
Search and questions over internal documents (RAG): procedures, contracts, catalogues, knowledge bases
Data extraction from invoices, orders, forms and emails, with validation and delivery into your software
Automatic classification and routing of requests, emails and tickets
Summaries, transcripts and meeting notes, with human review
Assisted content generation (product descriptions, template replies), with approval before publishing
Agents and automations that carry out steps in your software, with human approval for important actions
AI features inside your app: semantic search, recommendations, smart completion

How we work, step by step

We choose a use case with clear value and a success criterion (time saved, fewer errors)
We gather the data and build a test set from real examples
We build a prototype and measure it: accuracy, response time, cost per request
We launch a narrow pilot, with monitoring and a way back to the manual process
We go to production only if the results justify it and you have accepted the limits

Models and tools

We pick the model by task, price, language and privacy needs, not by fashion.

API models from major providers (for example OpenAI, Anthropic, Google, Mistral)
Open models hosted on your infrastructure or in an EU cloud, when data cannot leave the company
Vector search and databases (for example PostgreSQL with pgvector) for documents and memory
Connection to your software through APIs and protocols such as MCP, where they fit
Automation tools (custom code or workflow platforms) and monitoring of cost and quality

What we do not promise

AI models can be wrong or make up answers. That is why we add cited sources, confidence thresholds, human review for important decisions and a way back to the manual process. We do not guarantee results without a test on your data, and we do not replace people where the decision affects other individuals.

Visibility in AI answers

If you want your company to be found and cited in answers from AI engines (AEO, GEO), we work on clear content, correct structured data and files such as llms.txt. Details are on the SEO optimisation service page.

See SEO optimisation

Questions about AI solutions

What AI solutions can .brutalis implement?: Assistants over your documents, semantic search, data extraction from invoices and forms, request classification, summaries and transcripts, agents and automations with human approval, plus AI features in apps. Each project starts from a use case tested on your data.
Do you train your own AI models?: No. We integrate existing models, through an API or hosted on your infrastructure, and fit them to your process through documents, instructions and checks. If a case needs fine-tuning, we discuss it separately.
Is my company’s data safe?: We send the model only what is necessary, sign data processing agreements with providers and, where the option exists, ask that data is not used for training. For sensitive data we can use models hosted on your own infrastructure.
What does the AI Act mean for my company?: Regulation (EU) 2024/1689 requires, among other things, that you tell people when they are talking to an AI system (art. 50) and that you have staff with adequate AI literacy (art. 4). High-risk cases have much stricter requirements. We help you classify the project correctly.
Can AI make mistakes?: Yes. We reduce the risk with answers grounded in documents with cited sources, tests on real examples, confidence thresholds and human approval for important decisions. We do not recommend unsupervised AI where errors have serious consequences.
How much does an AI implementation cost?: It depends on the case, the data volume and the integrations. The cost has two parts: the implementation and the model’s usage cost, which we estimate up front. You get a written quote, per stage, starting with a measured prototype.

Other services

Web design
Business websites with custom design, built for phones and for Google. Every page has one job: move the visitor toward a call, a booking or an order.
Web development
We turn the approved design into a fast, secure, easy-to-maintain site written in Next.js and React.
Custom software development
Business apps, APIs, integrations and automations built for how your company works, with tests and code handed over.
Online stores
Online stores with products, cart, card payment and courier delivery, built to sell and easy to manage.
UI/UX design
We study how people reach you and how they decide, then design the interface that gets them to what they want faster.
Brand identity
Logo, colours, typography and usage rules, so your brand looks the same everywhere: on the site, on social media and on paper.
Digital content management and coordination
We look after what shows up about your business online: your Google Business Profile (formerly Google My Business), social accounts, posting calendar and replies to messages, in one monthly plan.
Mobile apps
iOS and Android apps designed for speed, ease of use and simple maintenance.
SEO
Technical audit, keyword research, content and a local profile, measured monthly in Google Search Console.
Maintenance and support
Security updates, backups, monitoring and small changes, in a monthly plan with everything it includes written in the quote.

Tell us what you are building.

Send a short brief and we will come back with a clear quote.

Get a quote