Assessment
The cap comes off, the meter goes in, the rules go on.
A share of your annual AI spend, net. The minimum is a floor, not a fixed fee.
from €24,000
For companies that want to open up AI without handing over money or data. The gateway goes into your account, budgets and rules go in, the dashboard goes on.
How it runs
Which applications, keys and agents exist, which data they see, and which class that data falls into. One day with IT and the business.
Installed on your infrastructure. Connected to your providers: Bedrock, EU inference, frontier with zero retention. Identity through your existing directory.
A budget per team and application. Routing by data class. Pseudonymisation before the model, filtering before the output. Tool permissions per agent. An audit entry per call.
Cost and hours per task, cost per outcome, adoption per department, rules and violations. Runbooks, operator training, handover to your team.
Management sees per team what AI costs and what it delivers. IT sees which data goes where and what agents may do.
Scope
Included are the inventory, the installation, the provider connections, budgets and rules, the dashboard, runbooks, operator training and the handover. The Cost Picture is credited against the price.
Not included are process redesign, prototypes and new AI features. That comes afterwards, once the gateway shows where it pays, and is then offered separately at a fixed price.
Operations
Our own gateway, our own numbers, our own agents under the same rules. What we build into your setup is what we use every day.
Questions
An LLM gateway is the one place every AI call from people, tools and agents passes through before it reaches a provider. Plexify installs it in the client's own account, because that is where it is counted and where the rules apply: a budget per team, routing by data class, pseudonymisation, tool permissions per agent, an audit entry per call. No second vendor sees the prompts.
A cap per head limits everyone equally without knowing who creates value. Plexify derives budgets per team and application from measured usage: those who move a lot get room, those who do nothing stop costing. The gateway enforces the budgets and shows in the dashboard where they bite. The cap is switched off afterwards.
The provider's region alone does not answer that. Plexify's gateway decides before the provider: the data class picks the model, internal data stays with EU inference or on the company's own hardware, and frontier models are used only with a zero-retention agreement and documented. Pseudonymisation runs before the model, role-based filtering before the output.
Exactly what its task needs, and not one path more. Plexify grants tool permissions per agent on the principle of least privilege: which directories it may read, which files it may write, which commands it may run. An agent with full access is not a productivity tool but an open window. Every call is in the audit trail.
About six weeks, depending on the number of providers and teams. Plexify starts with one day of inventory with IT and the business, installs and connects in weeks two and three, sets budgets and rules in weeks four and five, and hands over in week six with dashboard, runbooks and training. After that the client's team runs the system itself.
Contact
Fixed price, a clear result, no obligation.
An answer within 24 hours.
