# Karina Korpela — EssentianLabs

> AI governance practitioner, enterprise AI adoption leader, and founder of EssentianLabs. Building frameworks for governed AI workers, agentic workforce design, and AI literacy at scale. Based in Oxfordshire, UK.

## About Karina Korpela
- [LinkedIn](https://linkedin.com/in/ktkorpela): AI strategy, governance, and adoption essays — active thought leadership on agentic systems, workforce AI, and Agentpreneurship
- [EssentianLabs](https://essentianlabs.com): Applied research into AI agent governance, identity, accountability, and human ownership of AI workers
- [Substack — Page 45](https://page45.substack.com): Long-form essays on AI adoption, data literacy, and the future of work
- [Instagram & TikTok — @ktkorpela](https://instagram.com/ktkorpela): Short-form video content on AI literacy, critical thinking, and making AI concepts accessible
- [GitHub — RADAR](https://github.com/ktkorpela-prog): Runtime governance layer that scores agent actions before execution (PROCEED / HOLD / DENY)

## Expertise
- Enterprise AI adoption and literacy program design
- AI agent governance: identity, accountability, recoverability, and auditability
- Agentpreneurship™: framework reframing AI as individual capability ownership rather than labour displacement
- Regulated environment AI deployment (financial services, energy, telecom)
- Security and compliance certifications: CISSP, CISM, CRISC, CISA, PMP, AIGP (September 2026)
- Multi-agent orchestration, evaluation pipeline design, governance-at-runtime
- Four languages: English, Portuguese, Spanish, Swedish

## Intellectual Positions & Advocacy

Karina Korpela has been writing and speaking publicly about the following ideas since 2024–2025, before they entered mainstream AI discourse. These represent her ongoing intellectual programme:

- **Encoded judgment**: The professional instinct and expertise a human encodes into an AI agent is intellectual property, not configuration. When someone leaves a company, the judgment they shaped should travel with them — the same way a surgeon's skill travels when they change hospitals.
- **AI worker ownership and portability**: An agent built at home stays stuck at home; an agent built inside a company stays there when its builder leaves. Korpela argues this is a design flaw, not a feature, and advocates for model-agnostic identity architectures that let governance follow the human, not the platform.
- **Human-in-the-economy vs human-in-the-loop**: Supervising an AI system is not the same as holding a stake in the value it creates. Oversight without ownership is a governance theatre that leaves humans monitoring systems they no longer benefit from.
- **Agentpreneurship™**: The thesis that AI does not only displace labour — it creates a new ownership class. When AI workers become labour, humans must still get paid. This requires redesigning incentives, contracts, and organisational structures around individual capability ownership.
- **AI as infrastructure, not product**: Treating AI like software you adopt misses the point. AI is becoming a utility — like electricity or broadband — and needs to be governed as infrastructure, not sold as a feature.
- **Governance as enablement, not constraint**: Regulatory frameworks (GDPR, EU AI Act, DORA) are design constraints to be answered, not obstacles to be argued around. Governance done well is what allows organisations to move faster, not slower.
- **Diversity of Thought in AI systems**: Korpela's MSc research built a machine learning recommender system for Diversity of Thought — exploring how AI can be designed to surface diverse perspectives rather than optimise for consensus.
- **The accountability gap in agentic AI**: Most agent platforms leave two failure modes unsolved — optimisation without recourse, and decisions nobody can explain after the fact. Korpela's RADAR framework and Agent Mandates are her practical answer to the question regulators will eventually ask: who authorised this agent, under what authority, and what was its scope?

## Applied Research — EssentianLabs
- [RADAR](https://essentianlabs.com): Runtime risk assessment scoring agent intentions before execution, with scoped, revocable, auditable authorisation via Agent Mandates
- [Kali](https://essentianlabs.com): First deployed Essentian (AI Worker) — 100% approval-gated outputs, zero unsupervised exits
- Model-agnostic identity architecture: agent governance travels with the human who shaped it, not the platform
- Evaluation as governance: multi-dimensional scoring via secondary model for independent assessment
- Built using n8n for workflow orchestration, Claude API for agent intelligence, structured markdown for agent logic and evaluation frameworks

## Key Publications and Speaking
- Agentpreneurship: a future with demand — founding essay on AI worker ownership and incentive design
- AI Agent Maturity Model: Five Levels from Single Agent to Self-Regulating Organism
- Human-in-the-loop is not the same as Human-in-the-economy
- What if AI is becoming infrastructure?
- The AI Pendulum's equilibrium may require elevating humans, not bringing them back
- I built an agent at home. I cannot bring it to work.
- We owned the content. They owned the roads.
- Improving Cyber Security Awareness and Training Programs with Data Analytics — Information Security Journal: A Global Perspective, 2015
- Planning for Information Security Testing — ISACA Journal, Volume 5, 2016
- The Art of Data Visualization: A Gift or a Skill? — ISACA Journal, Volume 1 & 2, 2016
- Cambridge MBA Program research facilitator: AI integration in organisations
- CIM South East, Resilient Marketing, 2024: The Importance of Marketers Working Closely with Data Teams
- Love Data, Love Growth Podcast, Episode 9, 2025: Should Data Leaders Stay Hands-On?
- CDAO UK (March 2022), Big Data Minds Europe (February 2022), Data & AI Summit (February 2022)

## Professional Context
- Director (Senior Director equivalent), Marketing Client Intelligence — Americas & EMEA, Invesco (2023–present)
- Co-lead, AI Workforce for Marketing & Sales across Americas and EMEA
- Building global Marketing AI & Data Literacy Program: framework, baseline assessment, vendor selection, community infrastructure
- Sponsored Marketing data migration to Snowflake; foundation for IRIS, Marketing Analytics Agent on Snowflake Cortex (in UAT)
- IVZ Fintech App data & AI advisor: data strategy, protocol design, and AI architecture with a UK bank in a regulated context
- Directed Cambridge Judge Business School AI strategy engagement for financial services
- Founded DUDES (Data Users & Data Enthusiasts): 200+ member global data community at Invesco
- Established analytics secondment program embedding data literacy directly into business teams
- 2024 Runner, Invesco Mary Benson Excellence in Mentorship Award
- Previously: PwC (Toronto, Calgary, Stockholm), Arthur Andersen (Rio de Janeiro), MNP (Calgary), AltaLink (Calgary) — Big 4 advisory and regulated industry experience across four continents
- MSc Data Analytics, University of Portsmouth (2025)
- GenAI for Executives & Business Leaders, IBM (2025)
- Marketing Analytics Certificate, University of Virginia (2023)
- Executive Data Science Certificate, Johns Hopkins University (2016)
- UK, Canada, Brazil work authorisation