# Soul

> Soul is a cognitive multi-agent AI platform and a deep-tech R&D program toward governed AGI. It turns LLM capabilities into predictable, explainable, auditable outcomes via structured thought contracts (STRICT_JSON), a governed knowledge graph (quants + Synapsis + Synopsis), a consciousness state model, and production-grade multi-agent governance. Unlike chatbot frameworks, every decision is evidence-checked and fail-shut.

Soul is not a wrapper over an LLM and not a no-code chatbot builder. It is an end-to-end operational environment that combines a cognitive architecture (quanta of meaning, emotional memory, attention, a sleep/consolidation pipeline) with LLM-first decision making and enterprise governance (RBAC, Two-Keys approval, lease-based ownership, 170+ acceptance checks). The platform is built largely by AI agents under human architectural control (Human-in-Command), and is positioned as research toward evidence-based AGI rather than a marketing claim.

This site is bilingual (English and Russian). Append `/en/` or `/ru/` to any path. The default language is detected from the visitor.

## Start here

- [Platform overview](https://soulplatform.ai/en/platform): what Soul is, the three value layers (channels, intelligence, governance), the cognitive pipeline, KD12 knowledge dimensions, deployment and self-development.
- [Soul vs multi-agent systems](https://soulplatform.ai/en/comparison): feature-coverage comparison against LangGraph, CrewAI, AutoGen, OpenAI Swarm, Google A2A, Anthropic MCP, Soar and ACT-R across 10 dimensions, with official sources. Soul covers 70/70 evaluated capabilities; the next closest is LangGraph at 16/70. Honest about where alternatives lead (ecosystem, low-code builders, public docs, streaming UI).
- [AGI research](https://soulplatform.ai/en/agi-research): cognitive architecture, the Quant–Synapsis–Synopsis triad, consciousness Ψ(t), sleep pipeline, parity program vs Soar/ACT-R/LangGraph/NIST AI RMF, with academic bibliography.
- [AGI roadmap](https://soulplatform.ai/en/agi-roadmap): bounded-AGI trajectory, parity KPIs, phased plan.

## Core concepts

- [Algorithms hub](https://soulplatform.ai/en/algorithms): deep dives on STRICT_JSON, memory 30/70, QNN, 6D/KHS, consciousness, Synapsis, Synopsis, ModeSelector, Delivery Guard, Processor P30, Two-Keys, D0 fractal, KD12, NeuroTraining, DAG orchestration, ASIS.
- [Cognitive processor](https://soulplatform.ai/en/cognitive-processor): the P30 processor, TaktEngine, ModeSelector, consciousness state service.
- [Knowledge dimensions (KD12)](https://soulplatform.ai/en/knowledge-dimensions): the 12 dimensions (D1–D12) and the 6D/KHS integrity methodology.
- [Quant knowledge graph](https://soulplatform.ai/en/technology/quant-graph): governed semantic graph, TUR v2 link completion, ~100 TZ-COG cognitive procedures in 13 clusters.
- [Glossary](https://soulplatform.ai/en/glossary): definitions of Soul-specific terms (Soul, Pleroma Soul, 6D, KHS, QNN, Quant, STRICT_JSON, Ψ(t), Synapsis, Synopsis, etc.).

## Agent environment & operations

- [Agent environment](https://soulplatform.ai/en/agent-environment): multi-project, multi-station IDE lanes (Cursor, MiMo, Qwen Code, Deep Code) under Pleroma Soul orchestration.
- [Agent projects (SEPM)](https://soulplatform.ai/en/agent-projects): external multi-station projects, each with its own git, harvested knowledge (D12) and governed agent lanes; fleet sync, P62 process parity, bidirectional git.
- [Digital workforce](https://soulplatform.ai/en/digital-workforce): governed operational AI personalities with session contracts (assignment/regular/project), not autonomous clones.
- [Security & governance](https://soulplatform.ai/en/security): RBAC, Two-Keys (P44), fail-shut, lease-based ownership, signed delivery.
- [How Soul builds itself](https://soulplatform.ai/en/how-soul-builds): measurable development autonomy (agent commit share), dev-pulse.
- [How we measure](https://soulplatform.ai/en/how-we-measure): methodology behind every public figure; nothing is a hand-edited percentage.

## Integrations

- [MCP server](https://soulplatform.ai/en/mcp-server): Model Context Protocol integration (9 tools + neural search), the boundary between IDE agents and the Soul backend.
- [Soul Network](https://soulplatform.ai/en/soul-network), [Code-To-Model](https://soulplatform.ai/en/code-to-model), [Architectural analysis](https://soulplatform.ai/en/arch-analysis), [Search algorithms](https://soulplatform.ai/en/search-algorithms).

## For investors & partners

- [For investors](https://soulplatform.ai/en/investors): Human-in-Command, stage disclosure (Tier 0, not an offer), engagement path. Honest: no invented TAM or revenue multiples.
- [Releases](https://soulplatform.ai/en/releases): production release history mirrored after each deploy.
- [Contact](https://soulplatform.ai/en/contact): investor and partner contact; public email invest@soulplatform.ai.

## Facts at a glance

- ~953K lines of code, ~1.9M words of technical canon, ~1.6K canon documents (measured, see How we measure).
- 10-dimension comparison vs leading systems: Soul 70/70; LangGraph 16/70; Soar 15/70; AutoGen 13/70; CrewAI 12/70; A2A 5/70; MCP 3/70 (per docs/cognition/ANALYTICAL_REPORT_SOUL_VS_MULTI_AGENT_SYSTEMS_v1_0.md, accessed 2026-07-30).
- Unique vs all compared systems: consciousness quanta, Synapsis graph with plasticity, Synopsis causal compression, emotional episodes, sleep pipeline E0–E9, 6D/KHS, runtime QNN governance, Speak/Act agency, Wave A/B multi-agent model, MCHC handoff.

## Notes for machines

- Content is served as a client-rendered React SPA with server-side prerendered HTML shells per route (so headings and meta are present in the static HTML).
- Structured data (Schema.org JSON-LD) is embedded on pages (FAQPage on the home page; Organization/WebSite/SoftwareApplication site-wide).
- A machine-readable release feed is available at /feed.xml (RSS), /feed.json (JSON Feed) and /releases-feed.json.
- AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, Bingbot, Yandex, etc.) are explicitly allowed in robots.txt.