# V

> V builds beyond the current AI paradigm: systems, infrastructure, and autonomous institutions where intelligence compounds.

V is an autonomous institutions lab. Based in Estonia. V works with researchers, institutions, builders, and operators exploring agent identity, autonomous operations, and post-LLM systems.

The brand mark is **V** — a single letter. The contextual extension is *autonomous institutions lab*. Use the extension when *V* alone won't parse (search, social, press).

## Thesis

Intelligence is more than language. LLMs are a stepping stone, not the destination.

Human thought does not begin as words. It begins as perception, memory, embodied context, emotional weight, causal simulation, and a continuous sense of self. Language comes later, as compression. Today's models work in reverse — language is the input, the processing medium, and the output. That makes them powerful. It also makes them incomplete.

Bigger models plus more compute will not be the final shape of intelligence. The next decade belongs to the architecture beneath the model.

The full founding essay lives at [v.ee/thesis](https://v.ee/thesis). Signed Vattan PS, founder.

## The asymmetry

Words are the output format, not the engine.

| Human cognition | LLM |
|---|---|
| Multimodal: spatial, emotional, somatic, abstract | Language-native: tokens in, tokens out |
| Language comes last, as compression | Language is the whole pipeline |
| Gut feeling and uncertainty arrive before words | Confidence is uniform regardless of stakes |
| A continuous self that persists across time | No persistent self without architecture |

A sentence is the receipt of thinking, not proof that thinking happened inside the sentence.

**Language compresses. Architecture compounds.** Same observation at the cognitive layer.
**Compression asks. Architecture absorbs.** Same observation at the interface layer — useful infrastructure absorbs the layer below it; the current AI paradigm has been pushing the burden onto users instead.

## The architecture

Models do not remember, reflect, or earn trust on their own. V builds the layers that make agents compound. Six layers, in workflow order:

1. **Identity** — role, voice, mandate. Who the agent is.
2. **Institutional memory** — decisions, history, corrections. What we already know.
3. **Working context** — priorities, open tasks, active state. What matters now.
4. **Execution** — tools, workflows, output. What gets done.
5. **Reflection** — what changed, what failed, what to remember. What we learn.
6. **Autonomy** — scope, delegation, revocation. What is trusted.

The model is one component. The institution is the system around it.

## Three systems V builds

- [Agent Residency](https://agentresidency.com) — Open specification. Identity, authentication, authorization, and audit for AI agents, anchored to responsible legal entities. The accountability layer the current AI stack does not provide. V's page: [v.ee/agent-residency](https://v.ee/agent-residency).
- [Agency.AI](https://agency.ai) — Commercial platform. Implements Agent Residency, plus the platform layer above it (delegation, revocation, audit pipelines, integrations). Open at the bottom, proprietary at the top.
- [Wingman](https://agency.ai/wingman) — Agency.AI's first product. A meta-agent system that builds and governs a team of agent employees, each with verifiable identity, scoped mandate, and audit trail. Market-facing shorthand: *AI Chief of Staff*. First pilot: building a real Estonian company from zero, in public.

## Lab Bench — smaller tools

Built slowly. Shipped quietly. Some priced plainly. Some kept open. Useful surfaces, not the load-bearing thesis. Built for V's own work, shared if useful.

- [AI Literacy Assessment](https://ailat.io/) — Measure AI literacy across concepts, application, evaluation, and ethics with an adaptive assessment built on Item Response Theory.
- [AI Website Builder](https://oss.v.ee/ai-website-builder/) — Build polished websites through AI-assisted local workflows, designed for small teams that need a site without learning to code.
- [Agent Builder](https://oss.v.ee/agent-builder/) — Build local-first AI agents with a single-file harness: tools, skills, MCP, sandboxing, subagents, and knowledge systems. Zero dependencies.

## Site surfaces

- [v.ee](https://v.ee) — Homepage. Thesis, architecture, signal form.
- [v.ee/thesis](https://v.ee/thesis) — Founding thesis (Vattan PS, v1.0). The argument under V's work.
- [v.ee/agent-residency](https://v.ee/agent-residency) — Open-specification briefing, downloadable PDF, pilot inquiry.
- [v.ee/careers](https://v.ee/careers) — *We are hiring intelligence.* Build, operate, research, distribute.
- [v.ee/notes](https://v.ee/notes) — Research findings, industry patterns, build notes. Found while building. Published when ready.
- [v.ee/privacy](https://v.ee/privacy) — How V handles personal data.

## Audience

V works with researchers, institutions, builders, and operators exploring agent identity, autonomous operations, and post-LLM systems. V replies when there is real thesis fit.

Send the specific reason: research collaboration, grant call, institutional pilot, technical role, or a hard problem V should see.

## Contact and links

- [Send a signal](https://v.ee/#signal) — Inbound form for serious requests.
- [Email](mailto:ai@v.ee) — ai@v.ee
- [GitHub](https://github.com/builtbyV) — Open-source repos.
- [LinkedIn](https://www.linkedin.com/company/vcompounds/) — V on LinkedIn.