Zylon
6 known investors
Zylon is a private, on-premise enterprise AI platform designed for organizations in regulated industries that need to deploy generative AI without sending sensitive data to public cloud providers. The platform enables secure document processing, knowledge retrieval, and workflow automation while maintaining full data ownership and regulatory compliance.
Also known as PrivateGPT (open-source project by Zylon) · zylon-ai
Investors · 6
Also in the syndicate · 4
Company profile
researched Aug 2026Zylon is a European enterprise software company that develops private artificial intelligence infrastructure for organizations operating in regulated environments. Its platform lets institutions deploy generative AI inside their own data centers, private clouds or air-gapped networks so that sensitive data is not sent to public cloud providers. The product supports secure document processing, knowledge retrieval and workflow automation while the customer retains control of data, models and deployment environment [4].
The platform is organized into three components. Zylon AI Core is the self-contained foundation, bundling local LLMs, vector databases and GPU orchestration, and deploys to cloud VPCs, on-premise servers or fully air-gapped environments. Zylon Workspace is the end-user interface, offering an AI assistant, document creation, knowledge-base access, collaborative projects and connectors to existing systems. The Zylon API Gateway exposes OpenAI-compatible endpoints with authentication, logging, rate limiting and observability, and supports integration with tools such as n8n, LangChain and Claude Code; a one-click n8n deployment runs on the same server for building multi-step agents and automations [0][3]. Documentation describes single-command installation in under three hours, plus separately documented auditability, traceability, evidence-preservation and AI governance capabilities [3]. A packaged offering marketed as "Zylon in a Box" ships a pre-configured on-premise server that runs locally with no cloud dependency [0].
Zylon is the commercial evolution of PrivateGPT, the open-source project created by the same team. PrivateGPT is described as the open-source API layer that turns local models into production AI applications, following the Claude API as its reference specification, and connecting to any OpenAI-compatible inference server such as Ollama, llama.cpp or vLLM; capabilities include a messages API with streaming and token counting, file/artifact ingestion, retrieval with citations and agentic RAG, built-in and custom tools, MCP connectors, database and CSV querying, and embeddings. The repository states that PrivateGPT powers Zylon [2][5].
Founding story
According to a third-party profile, Zylon was founded in 2023 by Ivan Martinez Toro and Daniel Gallego Vico, motivated by the gap between enterprise demand for AI and the inability of regulated organizations to use cloud LLMs without ceding data control. The company's origin is tied directly to PrivateGPT, launched in May 2023 as a proof of concept for chatting with documents fully offline, which went viral on GitHub and demonstrated demand for private, local AI [2][5][7].
Business model
Zylon licenses a self-hosted enterprise AI platform installed inside the customer's own infrastructure (on-premise servers, private cloud VPCs on AWS, Azure or GCP, or air-gapped data centers), with solution architects assisting the customer's IT team during implementation. It also offers a hardware-plus-software bundle, "Zylon in a Box", consisting of a pre-configured on-premise server. The open-source PrivateGPT project serves as the technical core and as a developer-facing funnel, with Zylon positioned as the enterprise, end-to-end product built on top of it [0][4][5][7].
Marketing materials describe fixed, predictable platform pricing with unlimited AI usage rather than per-token or usage-based charging, positioned explicitly against pay-per-token cloud AI services. Specific price points are not disclosed in the sources [0][3].
Traction
Traction claims center on the PrivateGPT open-source project, which reports about 57,400 GitHub stars and 7,600 forks, with the company citing 57K+ stars, 8k+ derived projects and a 5k+ developer community. Endorsements are shown from LangChain, LlamaIndex and Qdrant. On the commercial side, the website names credit unions (including Redwood Credit Union and Bellwether Community Credit Union) as financial-services users; no revenue or customer-count figures are disclosed beyond a third-party estimate of under $5M revenue [0][2][5][7].
Latest developments
Current website messaging highlights "Zylon in a Box", a pre-configured on-premise server for plug-and-play private AI. The PrivateGPT repository has been rebuilt as PrivateGPT 1.0, an open-source API layer aligned with the Claude/Anthropic API that works with any OpenAI-compatible inference server and adds built-in database and CSV querying, MCP connectors, custom tools and integrations with Claude Desktop/Cowork, Claude for Microsoft 365, OpenCode, n8n and VS Code/Cline. Zylon documentation now covers a separate auditability and AI governance track [0][2][3].
▸Full profile — market position, technology, go-to-market, geography, history, risks & controversies
Market position
Positions itself as a private, on-premise alternative to public cloud assistants such as ChatGPT and Microsoft Copilot and to internally built AI stacks, competing on air-gapped deployment, audit and governance, model choice, predictable cost and deployment speed (about one week, or a claimed three hours for installation, versus 12-18 months for in-house builds). A third-party profile characterizes the company as a small firm (fewer than 25 employees, under $5M revenue, under $5M raised) whose credibility rests largely on the PrivateGPT open-source pedigree [0][3][7].
Stated differentiators are 100% on-premise or air-gapped operation with no external data egress; a complete, self-contained stack (models, vector search, ingestion, gateway, workspace) rather than a wrapper; freedom to use any model with no vendor lock-in; fixed, unlimited-usage pricing instead of per-token billing; rapid single-command deployment; documented auditability and AI governance; white-label and customization options; and pre-built connectors to on-premise data sources including banking cores [0][3][7].
Technology
A full-stack, self-contained AI infrastructure rather than a wrapper over a hosted model API: locally served LLMs, vector search/databases, document ingestion, GPU orchestration, an OpenAI- and Anthropic-compatible API gateway, and a collaborative workspace. Connectors cover banking cores (Symitar, Corelation, Fiserv), SharePoint, Confluence, PostgreSQL, Salesforce, file systems and S3. The underlying PrivateGPT layer supports retrieval with citations, agentic RAG, built-in and custom tools, MCP connectors, text-to-SQL and tabular analysis, embeddings, structured outputs and extended thinking, and works with any OpenAI-compatible inference server; prompt caching and OAuth/organizations are listed as unsupported [0][2][3].
Go-to-market
Direct enterprise sales driven by demo requests and free architecture consultation sessions with engineers, covering GPU/server architecture, model shortlists, RAG/agent blueprints and governance checklists for regulated deployments. Developer-led awareness comes from the PrivateGPT open-source repository and community, and adoption is eased by pre-built connectors to systems customers already run [0][5].
Enterprises and public bodies in regulated industries that handle sensitive or confidential data and must satisfy internal security policies, industry regulation and data sovereignty requirements. Named verticals include financial services (credit unions, banks, insurers, investment firms), government and public sector, healthcare, and defense and critical infrastructure; other referenced segments are engineering firms and manufacturers. The website lists credit unions including Redwood Credit Union and Bellwether Community Credit Union among financial-services references [0][4].
Geography
European company; third-party sources place the headquarters in the Madrid region of Spain (Valdemorillo / Madrid). The platform is described as serving enterprises across the globe, and deployments run inside customer infrastructure in cloud VPCs (AWS, Azure, GCP), on-premise data centers or air-gapped environments [2][4][6][7].
History
PrivateGPT was released in May 2023 and became one of the most-starred AI repositories of the year, twice reaching the top cross-category position on GitHub. The team then built Zylon as the commercial, enterprise-ready evolution, arguing that PrivateGPT solved privacy but not usability for non-technical workplaces. Zylon was founded in 2023 and raised a seed round announced in February 2024. PrivateGPT continued as an actively maintained open-source project, later rebuilt as PrivateGPT 1.0, an API layer following the Claude API spec, and it remains the core that powers Zylon. The company has since added a pre-configured hardware offering, "Zylon in a Box" [0][2][5][6][7].
Risks & controversies
Public funding data is inconsistent: one source reports a $3.2M seed on 13 February 2024 with Felicis, Zypsy, LifeX Ventures and several angels, while another reports a $3.0M seed in February 2024 with SNR and Talis Capital. Third-party aggregator data is also unreliable in places (one profile lists an employee range of 1001-5000 while its own narrative states fewer than 25 employees, and it references unrelated companies named Zylon Technologies in India and Zylon Software in Australia). Product claims such as deployment in one week versus three hours are inconsistent within the company's own site, and most performance, security and customer claims are self-reported and unaudited [0][6][7].
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Competitors · 6
by search overlap
Companies competing with Zylon for the same Google search keywords, organic and paid, via search-intersection analysis.
Timeline · 4
launches, deals, and filingsPrivateGPT was rebuilt as an open-source API layer for private AI applications, following the Claude/Anthropic API spec, connecting to any OpenAI-compatible inference server and providing RAG with citations, tools, MCP connectors, text-to-SQL and embeddings.
Zylon markets a pre-configured on-premise server shipped ready to run locally with no cloud dependency, labeled as new on the company website.
Madrid-based Zylon raised a $3.2 million seed round; reported investors include Felicis, Zypsy, LifeX Ventures and angels Hans Christian Zappel, Kirill Tashilov and Travis McPeak. A separate third-party profile records the same February 2024 seed round as $3.0M with investors SNR and Talis Capital.
$3.2M source ↗
PrivateGPT was released in May 2023 as a proof of concept allowing users to chat with documents fully offline; it went viral on GitHub and reached the top cross-category position twice that year. It later became the core technology powering the Zylon platform.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
In the news
▸Research sources · 8
primary sources listed
- Zylonzylon.ai · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Zylon do?
- Zylon sells an on-premise, air-gap-capable private generative AI platform for regulated industries, built on its open-source PrivateGPT project.
- Who are Zylon's investors?
- Zylon's investors include Felicis Ventures, LifeX Ventures.