KarminaAI StudioAgents · Services · Training

EXPERTISE · PUBLIC ADMINISTRATIONS

AI in public services needs public judgement.

Karmina AI Studio accompanies administrations and institutions to turn artificial intelligence into a useful, governed and understandable capacity: from strategy and training to pilots, agents and their supervision.

Deciding ·
·

StrategyKnowledgeGovernanceTrainingPilots with controlMeasure

WE DO NOT START BY THE Tool

Technology changes every week. Public responsibility remains.

An administration does not need a spectacular demonstration. You need to know what makes sense to transform, with what knowledge, under what rules, and with what supervision.

AN OWN CONTEXT

Innovate while protecting service, rights and trust.

The same technology can be useful in an internal task and inadequate in a decision that affects people. Context is not a legal note at the end: it defines the entire project.

Very potential

Intensive processes in information, teams that know frictions and services where a small improvement can reach many people.

More responsibility

Rights, data, transparency, contracting, explainability and continuity of service are part of the solution from day one.

Scattered knowledge

Regulations, procedures, resolutions, data and criteria coexist in different systems, websites, folders and people.

Unequal rhythms

Maturity is not homogeneous. Management, services, technology and teams need different routes but the same compass.

Citizen Confidence

The institution must be able to explain what the system does, where a response comes from, who has reviewed it, and how it can be corrected.

INSTITUTIONAL KNOWLEDGE

Documentary heritage needs to become infrastructure.

We don't dump folders into a tool. We decide what is official, who answers from each source, what version is in force, who can access it and what the system must do when it does not have a reliable answer.

Understanding the Knowledge Foundation
RegulationsProceduresDataResolutionsFAQsCriteria

Property

Validity

KNOWLEDGE
FOUNDATION

Permissions

Traceability

PeopleServicesDecisions

THE CONTEXT IS FIRST

The same answer does not apply to all organizations.

We can dedicate a free video call to understand the sector, limits, people and urgency before talking about technology.

Video call · free · no obligationBook a free video call A short conversation to understand your starting point and bring clarity.

A COMPREHENSIVE CONSULTATION

Six capabilities. One shared system of responsibility.

Strategy

Assessment and roadmap

Current uses, needs, risks, capacities and a portfolio prioritized according to public value, viability and effort.

Governance

Applicable policies and rules

Responsible, classification of risk, authorized uses, transparency, supervision, incidents and purchase criteria.

Knowledge

Knowledge ready to work

Official sources, validity, ownership, permits, citation, maintenance and limits converted into usable infrastructure.

Agents

Pilots connected to real processes

Shortened assistants and agents, with test cases, registrations, error control and Human Gates before scaling.

Learning

Training according to the role

Management, technical profiles, service managers and general staff with their own practice, criteria and responsibilities.

Measure

Continuous monitoring and improvement

Quality, time, adoption, incidences, impact and maintenance to decide what continues, what changes and what stops.

Karmina team working on a hybrid meeting

Barcelona · Madrid
Strategy · Creativity · AI · Governance

KARMINA · AI STUDIO · UTOPIQ

An institution needs more than a technology provider.

Karmina brings a decade of experience in strategy, communication, processes, content and projects for organizations, administrations and institutions. This trajectory has taught us to understand complex contexts and turn them into plans that teams can apply.

Karmina AI Studio transforms this knowledge into AI systems, agents, training and services connected to the actual job. Utopiq reinforces the look on artificial intelligence, ethics, governance and sustainability.

Portrait of Ignasi Llorente
Project Statement
In public communication, AI only makes sense if it helps make services clearer, more accessible and useful. It is not about automating to impress, but about ordering knowledge, making visible where each response comes from and always keeping a person responsible for the decision.

START FOR THE FASHION

Where can it add value while keeping responsibility clear?

Not all processes need to be turned into agents. Sometimes, the right answer is to order the procedure, improve the data, create a guide or apply a simpler automation.

01

Internal knowledge

Locate procedures, regulations and current criteria with source, version and responsible.

02

Attention to Citizens

Prepare basic information, classify consultations and refer cases that need a person.

03

Document management

Compare, summarize, extract and order documents so that the professional review is more agile.

04

Public Policies

Synthesize background and evidence, detect gaps and prepare first reviewable drafts.

05

Accessible communication

Adapt content by channel, language, format and reading level preserving the official source.

06

Processes and Coordination

Order agreements, notices, control lists and dependencies between services and teams.

FROM IDEA TO IMPACT

A path with oversight throughout the lifecycle.

The project can begin and end at any stage. The dimension is marked by institutional need, not available technology.

  1. 01

    Listen

    Frictions, services, affected people and uses that already exist.

  2. 02

    Prioritize

    Public value, risk, available knowledge, effort and maintenance capacity.

  3. 03

    Govern

    Responsible, sources, permits, limits, tests and stop criteria.

  4. 04

    Provar

    A pilot bounded with real cases, exceptions and human supervision.

  5. 05

    Measure

    Quality, utility, adoption, incidences and review load.

  6. 06

    Scaling

    Only that which demonstrates value enters into the operational and institutional register.

TO TERROR THE PROJECT

Share the context and constraints.

If you have already completed, brief, internal document or a specific question, send us. We will do a first reading with industry criteria.

CASES AND APPLIED EXPERIENCE

When we can name the organisation, we explain the whole system.

Karmina has worked with numerous administrations, institutions and public organizations. Some of this experience is subject to confidentiality; what we can publish is presented with verifiable context, sources and results.

Terrassa City Council Artificial Intelligence Master Plan 2026–2028

CAS DOCUMENTED · TERRACE ADJUSTMENT

From dispersed use to an institutional policy of AI.

Assessment, participation, leadership alignment, governance, literacy and a 2026–2028 roadmap approved by the municipal plenary.

341 answers160participants25 identified proposals
Read the full case
REPRESENTATIVE APPLICATION

Knowledge of public service

An internal assistant who responds with current procedures, cites the source and derives the exceptions.

REPRESENTATIVE APPLICATION

Multi-Actor Coordination

Agreements, dependencies and verifiable tasks between technical, legal teams, suppliers and public officials.

REPRESENTATIVE APPLICATION

Communication and accessibility

Adaptations by channel, language and reading level with official source and validation circuit.

RED LINES

Innovation also means knowing what must not be delegated.

The AI can prepare, order and propose. Public accountability remains human.

CONTINUE EXPLORING

Strategy, knowledge, training and the regulatory framework.

01

LLM Strategy

Order the adoption, the priorities and the roadmap.

Explore ↗
02

Knowledge Foundation

Prepare sources, criteria, permissions and tests.

Explore ↗
03

Learning

In-company training adapted to roles and risks.

Explore ↗
04

Guide AI Act

Understand what to prepare within the organization.

Explore ↗

QUESTIONS BEFORE YOU START

You do not need to arrive with a defined solution.

Do you know which agent you want to build?

No. The diagnosis can conclude that an agent, a simpler automation, a documentary improvement, a training, a director’s plan or a combination of actions is appropriate.

Can you start without opening a huge project?

Yeah. It can begin with a limited diagnosis or with a single priority process, as long as there is a responsible person and clear criteria to evaluate it.

Can internal documents be used?

Yes, when the sources, permissions, purposes and technical environment have been classified. Sensitive documentation is not entered into unauthorized tools.

How is data protection and regulation done?

The legal, security, transparency and oversight criteria are incorporated into the design and are coordinated with the competent heads of the institution.

Can you work with existing technology, legal and suppliers?

Yeah. Karmina AI Studio can act as a strategic and functional layer, defining requirements and coordinating with areas and suppliers that are already part of the institutional ecosystem.

Let’s talk

What does the institution need to govern better?

Tell us about your situation, even if the challenge is not fully defined. A member of our team will respond to understand the process and discuss a first step.

hello@karmina.ai