FriendlyAI

They sell access to AI.
We sell control over AI.

A Finnish high-technology company building the parvi family — agent-native systems with cryptographic governance. Designed for a world where AI must prove what it does.

Finnish High-Tech NATO Country Built In-House EU AI Act Ready
The Company

Most AI companies ship models that talk.
We build systems that prove.

Founded by Renata Baldissara-Kunnela — a senior software tester with over 15 years of experience before moving into AI development. That background is not a footnote. It is the foundation of the company.

15+
Years of QA and security testing
6
Products in the parvi family
100%
Built in-house, original engineering
0
Wrappers, reskins, thin API layers

The core conviction: The only question that matters is, "how do you know it is right?" A tester asks that. Every product in the parvi family answers it — with cryptographic proof.

The Name

parvi means a flock of birds.

Not one bird. Not one agent. Not one model. A flock moves as one organism. Each bird sees, reacts, and adapts, yet the whole formation travels with a shared intelligence no single member holds alone.

Swarm intelligence, engineered. The products are not isolated tools. They are a coordinated system where agents observe, decide, act, and prove — together.
Each member matters. Every product is powerful alone. The combination is what nobody else can replicate.
Adaptive. The flock learns from every engagement. Memory compounds. The system improves.
SWARM INTELLIGENCE
The Product Family

Six products. One governed stack. Each layer reinforcing the next.

parviGov
The governance hub. Cryptographic proof of every agent action. LAG (Merkle chain), AWP (signed receipts), PayBotFin (authorization layer). The spine of the family.
parviSight
The agentic browser. 26 MCP tools. Agents that see the web and prove what they did. Dual-projection architecture. Every action flows through parviGov.
parviClaw-core
The execution engine. Real PTY terminals. Multiple LLMs through one governed pipeline. Claude Code, Codex, GPT, Gemini — same governance.
AgenticTestari
Security testing. Powered by Sentinel-Hub. AI defenders trained against AI attackers in the RAID Lab. Swarm versus swarm warfare.
Graph Files
Document knowledge. Like M-Files, but agent-native and governed. Content-addressed, hash-based identity. Every node and edge witnessed.
parviALMA
Agent memory. Governed, persistent memory that learns from every engagement. Agents accumulate knowledge across sessions. What is remembered is accountable.
The Category

We do not compete in an existing category. We create a new one.

Agent-native systems with cryptographic governance. No existing company combines these two properties.

Agent Platforms

  • No governance
  • No cryptographic proof
  • Has agents

Governance & Audit Tools

  • Has governance
  • No agents
  • Not agent-native

FriendlyAI — Both

  • Agent-native, designed for AI as primary user
  • Cryptographic governance on every action
  • Built in-house, as an integrated stack

This is not a feature comparison. It is category creation. The parvi family occupies a space where there are no direct competitors because the combination does not exist anywhere else.

The Moat

Any single product is strong. The combination is what nobody else can replicate.

ACT GOVERN PROVE TRUST DEPLOY LEARN parvi Family Closed Loop

Each Product Alone

  • parviGov without parviSight: ledger with nothing to record
  • parviSight without parviGov: browser with no proof
  • parviClaw-core without parviGov: execution with no accountability
  • Sentinel-Hub without the stack: scripts without evidence
  • Graph Files without governance: just another document store

Together: The Closed Loop

  • Agents act
  • Actions are governed
  • Governance produces proof
  • Proof builds trust
  • Trust enables deployment
  • Deployment generates data
  • Data trains better agents
Compliance

EU AI Act compliance, from day one.

The EU AI Act demands that companies can explain what their AI did, prove it was safe, and show evidence of compliance. Most companies will scramble when enforcement begins.

Most Companies

Scramble when enforcement begins
Bolt on logging retroactively
Hire consultants for retroactive proof
Try to prove what black-box systems did

FriendlyAI

parviGov designed to produce regulator evidence
Every action has a signed receipt
Every action has a numbered ledger place
Every decision traceable, verifiable, auditable

We did not build governance for compliance. We built it because it is the correct way to engineer agent systems. Compliance is a consequence.

What We Build — and What We Do Not

Every piece of technology is built in-house. Not wrappers. Not reskins.

The Market

  • Rent intelligence from a provider
  • Wrap it in a chat box
  • Call it a product
  • Thin layer over someone else's API
  • Sell access to AI

FriendlyAI

  • FriendlyAI writes the Rust
  • FriendlyAI writes the Python
  • FriendlyAI designs the cryptographic protocols
  • FriendlyAI builds the agent swarms
  • Sells control over AI

From the ledger to the browser to the execution engine to the combat arena — original engineering. When something in the stack works, it works because it was understood, specified, and shipped by the team itself.

Finnish High-Tech

Operating from Finland — a NATO country.

Finnish engineering is known worldwide for discipline, transparency, and reliability. This is not a marketing label. It is a commitment.

Strong institutions. Operating under strong rule of law and a culture of engineering rigor.
Understands defense. Finland understands defense, security, and the responsibilities of building autonomous systems.
EU data residency. All cloud deployments run from Finland, satisfying EU AI Act data sovereignty requirements.
Stands behind the work. Builds things that work, proves they work, and stands behind them.

Finland

NATO Member
EU Member
World-class engineering tradition

The Founder

The tester's mindset is the DNA of the company.

RB

Renata Baldissara-Kunnela

Founder — Senior Testing Expert Turned AI Developer

Over 15 years of QA and security testing experience. That testing background shapes every product in the parvi family.

How do you know it is right?

A tester asks this. The parvi family answers with cryptographic proof.

How do you prove it?

Every agent action is witnessed, signed, and verifiable. Not because it sounds good in a pitch.

What happens when it fails?

The quality gate catches it. Score below 7 means halt. No override.

Not because it sounds good in a pitch. Because that is how a senior tester thinks. Fifteen years of breaking systems, finding what fails, and proving what works.

The Bottom Line

They sell access to AI. We sell control over AI.

They

  • Rent intelligence and wrap it
  • Build tools
  • Log actions
  • Test with static rules
  • Scramble for compliance

We

  • Engineer systems that hold intelligence accountable
  • Build a family
  • Prove actions cryptographically
  • Train defenders in combat
  • Were born compliant

FriendlyAI. Finnish high-tech. The parvi family. Built in-house. Governed cryptographically. Designed for a world where AI must prove what it does.

FriendlyAI

The parvi family. Built for a world where AI must prove what it does.

Let's build governed AI together.

Whether you need governance, security, document knowledge, or the full stack — FriendlyAI builds systems that prove what they do.

Visit friendlyai.fi Contact Us
Company friendlyai.fi
Governance parvigov.com
Security agentictestari.com
Contact dev@friendlyai.fi
YouTube @FriendlyAI_fi
EN PT FI
01 / 12