Enterprise AI Platform
One Control Layer Above Every Model
An enterprise AI platform is the shared layer an organisation uses to run all of its AI work, from chat and agents to retrieval, model routing, fine-tuning and evaluation, with one identity model, one policy set and one audit trail instead of a separate stack per department. VDF AI is an enterprise AI platform you can run in managed cloud, private cloud, on-premises or air-gapped.
Six products, one governance layer, any model provider.
Buy it as a service, or install it where your data already lives.
Or start free in the cloud and move on-premises later.
The situation this solves
- Marketing has a chatbot, engineering has a coding assistant, finance has a pilot, and nobody can list them all
- Each tool has its own login, its own data path and its own idea of what a user may see
- The board wants one AI strategy and the CISO wants one place to switch it off
- Procurement is being asked to renew three overlapping AI subscriptions this quarter
What one platform replaces
- Departmental chatbots become VDF AI Chat with permission-aware retrieval
- Framework prototypes become governed agents and networks
- Per-vendor model contracts become routing across providers
- Ad-hoc fine-tuning becomes the Data Suite with audit-ready evaluation
- Scattered logs become one governance and audit layer
Twelve Tools
or One Platform
A tool per team
- Twelve logins, twelve admin consoles, twelve invoices
- Data entitlements re-implemented in each product
- No shared view of spend, quality or risk
- Each vendor's cloud becomes a data residency exception
- An audit request means twelve export formats
- Switching models means renegotiating everywhere
One enterprise AI platform
- Single sign-on, one admin plane, one contract
- Entitlements from your directory applied everywhere
- Cost, latency, quality and energy per workload in one view
- Deployment mode chosen per data class, not per vendor
- One audit vault, one evidence export
- Model changes are routing policy, not projects
Work. Data.
Control.
Where people and agents work
VDF AI Chat for questions over internal knowledge, VDF AI Agents for role-specific assistants with tools, VDF AI Networks for objectives that need several agents, and VDF Code inside the engineering workflow.
Where models meet your data
The Data Suite generates fine-tuning datasets from your own documents, fine-tunes and evaluates models on-premises, and keeps result history so a model change is traceable and reversible.
Where control lives
The Router chooses models by quality, cost, latency and energy within policy. Living Knowledge stores every execution with cryptographic provenance. Role-based access and approval gates apply across all of it.
Three ways to buy enterprise AI
| Criterion | Hyperscaler AI suite | Productivity copilot bundle | VDF AI enterprise platform |
|---|---|---|---|
| Runs outside the vendor's cloud | No | No | Yes, including air-gapped |
| Model providers supported | That cloud's catalogue | The bundle's models | Any provider plus open-weight models |
| Agents and orchestration | Building blocks | Fixed assistants | Governed agents and multi-agent networks |
| Retrieval over internal systems | Assemble yourself | Within the suite's data | Native connectors across ecosystems, permission-aware |
| Audit and evidence | Cloud logs | Suite logs | Immutable Vault with evidence export |
| Commercial model | Consumption | Per user plus per message | Per user in cloud, capacity licence on-premises |
The eight criteria an enterprise AI platform evaluation should score
Weight these before the demos start, or the demos will set the weights for you. The scoring method, proof-of-concept design and exit clauses are in the enterprise AI procurement guide, and the practical run-off format is in how to run a bake-off between enterprise AI platforms.
- Coverage of the AI estateDoes one platform serve chat, agents, retrieval, routing, fine-tuning and evaluation, or will you buy and govern six products?
- One identity and policy modelUsers, roles and data entitlements should come from your directory once and apply to every model and agent.
- Deployment options that match data classesPublic data in managed cloud, regulated data on-premises, classified data air-gapped, all on the same platform.
- Model independenceCommercial and open-weight models interchangeable per task, so a price rise or a policy change is a routing update, not a migration.
- Cost visibility per workloadToken, transaction, latency and energy metrics per team and per node, ready for chargeback.
- Audit that satisfies a regulatorImmutable records of prompts, retrievals, tool calls and outputs, exportable as an evidence pack.
- Exit and portabilityYour data, embeddings, fine-tuned weights and workflow definitions leave with you in open formats.
- Total cost over three yearsLicence plus infrastructure, integration and operations, compared honestly against per-message cloud bundles.
Explore the platform
Category pages for each capability, the product suite and the deployment options.
- Product suiteAgents, Networks, Chat, Router, Code and the Data Suite.
- AI agent platformBuild, orchestrate and govern agents.
- Agentic AI platformBounded autonomy for multi-step objectives.
- AI governance platformPolicy, approvals and audit for agents and models.
- Enterprise RAGPermission-aware retrieval over internal knowledge.
- On-premises AI platformDeployed inside your datacentre in two weeks.
- AI agents by industryBanking, insurance, healthcare, government, energy, telecom and more.
- PricingCloud plans and on-premises capacity licensing.
Questions about enterprise AI platforms
What is an enterprise AI platform?
An enterprise AI platform is the shared software layer an organisation uses to run all of its AI work, including chat assistants, AI agents, retrieval over internal knowledge, model routing, fine-tuning and evaluation, with one identity model, one policy set and one audit trail instead of a separate stack per department. VDF AI is an enterprise AI platform available as managed cloud or as software you run on your own infrastructure.
How is an enterprise AI platform different from Microsoft Copilot or a hyperscaler AI suite?
Copilot bundles assistants into one vendor's productivity suite and cloud. Hyperscaler suites give you model APIs and building blocks inside that cloud. An enterprise AI platform sits above providers: it lets you use Microsoft, Google, Anthropic, Mistral or open-weight models through one governed layer, and it can run outside any of their clouds. That independence is the reason regulated organisations choose it.
Which products make up the VDF AI platform?
VDF AI Agents for governed agent workspaces, VDF AI Networks for multi-agent orchestration, VDF AI Chat for private retrieval-augmented chat, VDF AI Router for cost and quality-aware model routing, VDF Code for on-premises coding assistance, and the Data Suite for fine-tuning data, model fine-tuning and evaluation. Living Knowledge stores every execution with provenance. All of them share one identity, policy and audit layer.
How should we evaluate enterprise AI platforms?
Run a bake-off on your own data with a scorecard that weights deployment options, model independence, governance evidence, connector depth, cost transparency and exit terms. Insist that vendors demonstrate on-premises or sovereign deployment rather than describe it. The procurement guide and RFP checklist on this site give the full method and the question set.
What does an enterprise AI platform cost?
VDF AI cloud plans are priced per user with a free starter tier. On-premises deployments use capacity licensing: one annual subscription for a governed capacity pool with unlimited users, so orchestration, routing, retrieval and audit are never metered separately. Pricing details and the reasoning behind the model are on the pricing page.
See the whole platform on your data
A thirty-minute walkthrough of chat, agents, networks, routing and audit running together, then a scoping call for cloud or on-premises.