Product demo
MERIDVAR in five minutes
How one record and one gate take an organisation from not knowing which AI it uses to deciding, enforcing and proving what is allowed, on its own infrastructure.
5 minutes 4 seconds, English narration with captions. The product screens shown are illustrative.
Chapters
Jump to a stage
Each chapter follows one stage of the product, from discovery to the audit package and the spend ledger.
Transcript
The full narration
For readers who prefer text, and for anyone using a screen reader.
0:00 The problem
AI entered the enterprise from every direction at once. Employees use consumer AI. Developers wire in model APIs. Business units buy AI inside SaaS products. Agents act through MCP servers. Each decision was local. The exposure is not. No single team can say what is in use, who approved it, or what it costs.
0:25 What MERIDVAR is
MERIDVAR is the independent control layer for enterprise AI. It turns a fragmented AI estate into something you can discover, govern, secure, and prove. It becomes the AI system of record for risk, security, compliance, approvals, policy, evidence, and spend, without replacing your existing AI stack. One record, five connected stages, running on your own infrastructure.
0:55 Discover
Start with discovery. MERIDVAR builds an AI inventory from systems it observes in traffic, and systems a person declares. Every entry carries its provenance, owner, vendor, models, and lifecycle status. Shadow AI is found by comparing traffic with your list of sanctioned providers. For traffic that never touched the gateway, upload proxy or DNS logs. Agents are registered with signed identities, and each one shows the MCP servers and tools it can reach. An observation never becomes a declaration on its own. A person confirms it.
1:40 Assess and approve
Next, governance. Each AI system, vendor, agent, and MCP server is assessed with a versioned questionnaire. Published rules classify risk from Low to Critical, with a reason for every factor. The same facts always give the same level, and unknown is never read as safe. A High classification requires a privacy review, a security review, and an approved vendor before approval can be requested. Approval is decided by an administrator who did not make the request, on a pinned basis, with conditions and an end date. Vendors carry their own classification and approval.
2:25 Control and prove
Policy is enforced before the model is called. Traffic routed through the admission gateway passes a seventeen stage pipeline: identity, shadow AI, prompt DLP, threat detection, data residency, budgets, and approved models. Here, a prompt carries a national identification number. Prompt DLP refuses it. Every later stage is skipped, the provider is never called, and only a masked sample is stored. Each decision is written to an append only, hash chained evidence log. Control status for the EU AI Act, NIST AI RMF, and ISO 42001 is computed from the running system, and every AI system and vendor has a downloadable audit package.
3:15 Spend
The same record carries the money. Every governed request is written to a ledger with its tokens, the model served, and the actual cost, by team and by model. Budgets are checked before the call, not after the invoice. A team over a hard cap is refused at admission. The request never reaches the provider, and nothing is billed. Approved model lists decide which models each team may call. Past and ungoverned consumption can be imported with provider administration keys. Realised and modelled savings are kept apart, so finance sees the method as well as the number.
4:00 One record for every team
One record, read by every team that answers for AI. The CISO sees what AI exists, what it can reach, and what was refused. AI governance, privacy, and risk see every decision, with who made it, when, and on what basis. IT gives teams an OpenAI compatible endpoint, with a defined route to approval. Finance sees spend by team and by model, beside the decision that allowed it. And when the facts change, an approval reads review required.
4:35 What makes it different
AI gateways and runtime control planes govern only the traffic that passes through them. MERIDVAR governs AI regardless of where it runs, and ties the record to a gate that can refuse. Self hosted. Single tenant. No vendor cloud. MERIDVAR. The independent control layer for enterprise AI. Request a demo at meridvar.com.
Go further
Read the detail behind each chapter
See MERIDVAR refuse a request on your own network
A working session with the founder, on a host you control.
What happens next
- You send the formYour request goes straight to the founder.
- We reply by emailWe agree a time and what you want to see.
- We run a working sessionWe install the appliance with you, route a test request through it and read the evidence it leaves.