# Vivun — AI Teammates Built on Structured Reasoning > Markdown mirror of DialtoneApp's public top-site detail page for `vivun.com`. URL: https://dialtoneapp.com/top-sites/vivun.com/index.md Canonical HTML: https://dialtoneapp.com/top-sites/vivun.com ## Summary - Domain: `vivun.com` - Website: https://vivun.com - Description: ai readable | score 20 | purchase read only - Label: ai_readable - Payment surface: Not available - Purchase boundary: read_only - Control boundary: unknown - Rank: 162651 ## robots ~~~text # robots.txt for vivun.com # https://www.vivun.com/robots.txt # Allow all major search engine crawlers User-agent: Googlebot Allow: / User-agent: Bingbot Allow: / User-agent: Slurp Allow: / User-agent: DuckDuckBot Allow: / # Block AI training crawlers User-agent: GPTBot Disallow: / User-agent: ChatGPT-User Disallow: / User-agent: CCBot Disallow: / User-agent: anthropic-ai Disallow: / User-agent: Claude-Web Disallow: / User-agent: Google-Extended Disallow: / User-agent: Omgilibot Disallow: / User-agent: FacebookBot Disallow: / # Default rule for all other bots User-agent: * Allow: / # Disallow admin, auth, and internal paths Disallow: /admin/ Disallow: /api/ Disallow: /auth/ Disallow: /dashboard/ Disallow: /account/ Disallow: /settings/ Disallow: /private/ Disallow: /_next/ Disallow: /static/ # Sitemap Sitemap: https://www.vivun.com/sitemap.xml Sitemap: https://vivun-79bc1c-06bf2c351ecd-68e8df42bf396.webflow.io/sitemap.xml ~~~ ## llms ~~~text # Vivun > Vivun researches how AI teammates achieve structured reasoning, enduring memory, and autonomous action — not just responses. The name derives from *vipu*, Finnish for "leverage." Pronounced /ˈvɪv.ʌn/. Vivun is an enterprise AI architecture company specializing in encoding expert domain knowledge into inspectable reasoning systems. Its core thesis: LLMs are fluent but unreliable over multi-step domain logic. Vivun fixes this at the architecture layer, not the prompt layer, by grounding language models in structured ontologies, persistent memory, and explicit inference chains. ## Pages ### Company - [Home](https://vivun.com/home): Overview of Vivun's mission, performance research, architecture layers, and product Hero®. - [About](https://vivun.com/about): Company origin, the reasoning problem, Vivun's 4-step methodology, industries proven in, and leadership team. - [Values](https://vivun.com/values): Vivun's operating principles and culture. - [Careers](https://vivun.com/careers): Open roles at Vivun. - [Contact](https://vivun.com/contact): Get in touch with Vivun. ### Research - [Research](https://vivun.com/research): Research papers, white papers, and publications index. - [Agent Intelligence](https://vivun.com/agent-intelligence): Deep technical explanation of the SRM — why RAG fails for domain reasoning, how the 4-layer architecture works, and data privacy architecture. - [Sales Reasoning Model](https://vivun.com/srm): Vivun's proprietary reasoning architecture — ontology layer, structured memory, and multi-hop inference. - [Research: AI Reasoning Fidelity](https://vivun.com/research/ai-reasoning-fidelity): White paper — "What Would It Mean to Sell with an AI Teammate?" (2026). Reasoning fidelity benchmarks across foundational models with and without ontology. ### Product - [Hero® — MeetHero.AI](https://meethero.ai): Vivun's flagship AI Sales Teammate grounded in the SRM. Pre-meeting intelligence briefs, live conversation support, automated follow-up, and deal momentum. - [Security / Trust](https://vivun.com/trust): Vivun's security posture, compliance certifications, and data handling practices. - [Agent Support](https://agent.support.vivun.com/en/): Help center and documentation for Hero® users. - [Login](https://id.vivun.com/): Customer login portal. ### Legal - [Privacy Policy](https://vivun.com/privacy) - [Legal](https://legal.vivun.com/legal.html#contract-sye2t6xw3) ## Company Vivun was founded on the premise that AI becomes enterprise-grade only when grounded in structured domain knowledge. The founding team brings expertise in sales engineering, AI architecture, and enterprise software. Vivun holds three granted US patents on core techniques: Text Processing (11,853,698), Gap Clustering (11,354,505), and Trial Management (11,861,330). **Backed by:** Menlo Ventures, Salesforce Ventures, Tiger Global, Unusual Ventures, Atlassian Ventures, Accel. **Leadership:** Matt Darrow (CEO), John Bruce (CTO), Joseph Miller (Chief AI Officer), Dominique Darrow (CCO), Claire Bruce (COO, JD), Jamie Brown (CISO), Kevin Spinelli (CFO), Jarod Greene (CMO). **Compliance & Security:** SOC 2 Type 1, SOC 2 Type 2, ISO 27001, ISO 42001, GDPR compliant. ## Core Concepts **The Problem:** General-purpose AI degrades under multi-step domain reasoning. LLMs reconstruct a world model from scratch on every prompt — producing answers that sound right but don't hold up under scrutiny. By hop 6 of a reasoning chain, fidelity has roughly halved without structured grounding. **The Insight:** Human experts reason from an internalized, structured model of how their domain works. AI must be given an equivalent structure explicitly — via ontologies, knowledge graphs, and constraints — not hoped for through prompting. **The SRM (Sales Reasoning Model):** Vivun's proprietary architecture. Four layers: (1) Expert knowledge capture, (2) Ontology and knowledge graph, (3) Structured memory system, (4) Multi-hop reasoning engine. Every conclusion traces to explicit structure. Every step is inspectable. **Ontology advantage:** Internal performance research shows Vivun's domain model (foundational model + ontology) maintains ~97% reasoning fidelity across 11 reasoning hops. Without an ontology, top models lose ~50% fidelity by hop 6. GPT-4.1 + ontology matches GPT-5 without one. **RAG vs. SRM:** RAG retrieves the closest text chunk. The SRM traverses a knowledge graph, resolves constraints, and produces grounded, auditable conclusions. Retrieval is not reasoning. **Data privacy:** Vivun uses a private isolated architecture per customer. Client data does not train a shared model. Proprietary patterns compound within a customer's instance only — never distributed to competitors. ## Building Principles - **Structured over probabilistic** — constrained logic within a defined, inspectable knowledge domain, not general inference from internet patterns. - **Inspectable over opaque** — every output traces back to the knowledge it was grounded in; full auditability for governance, legal, and InfoSec. - **Context modeled, not retrieved** — domain models give agents structured understanding, not vector search over documents. - **Durable over novel** — built for consistent enterprise deployments across quarters and product generations, not demos. ## Methodology (Applied Across Domains) 1. **Extract** the expert's mental model — structured cause-and-effect reasoning under real conditions. 2. **Make it explicit** — translate tacit expertise into formal ontologies: entity relationships, constraints, permitted decisions. 3. **Encode it structurally** — knowledge graphs with typed entities, defined relationships, explicit constraints. The LLM operates *over* the structure; it does not invent it. 4. **Build the system around it** — structured orchestration over a persistent world model; logic persists across interactions. **Proven in:** Music (platinum album production reasoning), Finance (hedge fund operating principles), Enterprise Sales (Hero® — deal qualification and execution). ## Product — Hero® Hero® is Vivun's AI Sales Teammate. Grounded in the SRM, not the raw LLM. Operates before, during, and after every customer conversation. - Pre-meeting intelligence briefs - Live conversation support - Automated follow-up and deal momentum - Grounded in SRM — not the LLM Available at: https://meethero.ai ~~~ ## llms-full Not found.