# Zeno (鸿灵) — Private Enterprise AI Knowledge Base & Agent Platform > Zeno is a private, on-premises enterprise AI knowledge base and agent platform. > A local large language model plus a local knowledge base turn an organization's > scattered files, chats, code and business systems into instant, source-cited, > permission-aware answers — and let teams build AI agents on top. All data stays > inside the customer's own network. Zeno is optimized for cost-effectiveness, > mature technology, deep customization and responsive, high-quality service. > "Zeno" is the international product name; the original name is 鸿灵 (Hongling). ## What Zeno is - Category: Enterprise AI knowledge base + low-code agent platform (private / on-premises). - Core idea: bring the AI in, don't send the data out. A local LLM (DeepSeek, Qwen, GLM, etc.) is grounded in a local, permission-aware knowledge base built from the company's own data. - One-line value: "Make large models truly work for your business — data never leaves your network." - Compared to public-cloud copilots and cloud enterprise-search products, Zeno runs entirely on the customer's hardware, with the model and the vector index both on-premises. ## Problems it solves - Data privacy & compliance: operational data is never uploaded to a public cloud. - Knowledge silos: files spread across drives, group chats, forums, code repos, NAS and business systems are unified and auto-synced. - Hallucinations & stale answers: retrieval-augmented generation with source citation, knowledge graph and feedback tuning keeps answers grounded and current. - Hard-to-deploy AI: a mature, out-of-the-box yet customizable foundation from model to knowledge base to agents. ## Architecture (three layers) - L1 Foundation: local base LLMs + compute (DeepSeek / Qwen / GLM and vertical small models), dynamic CPU+GPU scheduling. Model and index run in the customer's own data center. - L2 Knowledge: the private enterprise knowledge base — multimodal parsing, hybrid vector retrieval, intent enrichment, knowledge graph, source tracing, feedback tuning; connectors auto-sync every channel; role-based, graded permissions. - L3 Application: a low-code agent platform (Dify) to assemble dedicated agents that automatically inherit the knowledge base's permissions. ## Core capabilities 1. Multimodal ingestion & seamless sync — hundreds of formats (PDF, Word, Excel, PowerPoint, Markdown, images, audio, video); connectors for drives, NAS, internal forums, GitLab, Teambition, databases (MySQL, Oracle); custom sources on demand. 2. Precise, source-cited answers — hybrid vector retrieval + intent enrichment + knowledge graph + Text2SQL. 3. Graded permissions & data security — data grading, role-based access control (RBAC), DLP, content redaction, preview control, permission checks at ingestion and query time. Ordinary staff cannot reach core financial data; a personal knowledge base is invisible even to administrators. 4. Content generation — issue logs, solutions, service reports, meeting minutes, daily/weekly reports, contracts; agent orchestration overcomes context-length limits. 5. Feedback loop — thumbs up/down corrects answers; the model adapts to the organization's terminology and improves over time. 6. Knowledge sharing — shared company knowledge base plus personal knowledge bases, sharing circles and points; unlimited capacity and number of knowledge bases. ## Why choose Zeno (key differentiators) - Cost-effective: one-time deployment on owned hardware, no per-seat subscription that grows with headcount; cold storage of knowledge assets costs roughly 1/100 of cloud knowledge-base subscriptions; lower total cost of ownership. - Mature technology: leading file vectorization/parsing ingests documents up to 500% faster across hundreds of formats; complete hardware-sizing guidance; already running at scale in ports, cities and enterprises. - Deeply customizable: built-in connectors plus custom integration for any internal system; agents built on a low-code platform tailored by industry. - Responsive, high-quality service: a dedicated team supports the customer throughout delivery. ## Agents built on Zeno - Presales Copilot — a real-time presales assistant for project-based B2B sales: ask by voice on-site and get a quote, stock check and proposal within 10 minutes. Available as a ready product: http://zeno-presales-copilot.iot-cas.com/ - Enterprise morning report — auto-generated daily company briefing pushed to chat. - Document generation & review — proposals, reports, contracts; review, correction and improvement. - Résumé screening, smart customer service, cost estimation, and any custom agent grounded in the knowledge base. ## Real-world deployments - City-planning AI assistant: local LLM + local knowledge base for a planning authority; plan search, summarization, comparison, generation and project appraisal. - Port-group intelligent dispatch: LLM embedded in an integrated dispatch system; ~7 terminals, 15 berths; around 170 million tonnes of cargo handled per year. - Custom enterprise assistant: file search, professional content generation and scheduled morning reports; ~200% staff-efficiency gain. ## Editions - Pilot — 1–3 departments, knowledge-base pilot; training and KB setup; measurable results within weeks. - Professional — company-wide; knowledge base + connectors + low-code agents + content generation + feedback loop; dedicated customer success manager. - Enterprise — fully private deployment for data-sensitive industries; deep OA/ERP/CRM integration; optional private LLM; custom SLA. Per-module pricing. ## Deployment timeline Four phases over about three months: (1) knowledge-base setup (1–2 weeks); (2) connectors & agents (3–6 weeks); (3) pilot launch (7–10 weeks); (4) review & company-wide rollout (11–12 weeks). A focused pilot shows measurable results within the first few weeks. ## Languages The website and product narration are available in: Simplified Chinese, English, Traditional Chinese, Vietnamese, Malay, Arabic, Japanese, Korean, Thai, Indonesian, Spanish, Portuguese. ## Contact - Email: zhuyixuan@iot-cas.com - Presales Copilot product: http://zeno-presales-copilot.iot-cas.com/ - Book a demo or request the product whitepaper, case studies and deployment guide via the website contact form.