ARC
Pricing
Discord
Talk to an engineerStart free
Start free
ARCby FenxLabsCollective Intelligence

Platform

  • How it works
  • Custom routing and governance
  • Use cases
  • Example architectures

Deployment

  • Hosted Platform
  • Managed Service
  • Self-hosted Licence
  • For ISVs
  • Open the portal

Services

  • Assessment and planning
  • Architecture and integration
  • Model Adaptation
  • EU AI Act readiness
  • Managed Service operations

Resources

  • Cost modeller
  • Compare approaches
  • Pricing
  • FAQ
  • Transparency
  • Get started
  • Discord community

Legal

  • Privacy notice
  • Terms and conditions
  • Sub-processors
  • Accessibility

For individuals

  • FenxChat

FenxLabs. KvK 91762782.

Herengracht 320, 1016 CE Amsterdam, Netherlands

+31 85 060 5273contact@fenxlabs.ai

© 2026 FenxLabs. All rights reserved.

Architecture and integration

Connect FenxARC™ to the systems you already run

A reviewed integration design that connects your applications and agents to approved models, tools and data through ARC, with every boundary and owner agreed before anything ships.

Discuss your requirementsSee where ARC sits in real environments

This engagement connects ARC, so where ARC runs must be settled first. Most engagements open with a paid proof of concept, deliberately small. Scoped to your requirements and quoted per engagement.

The target architecture, settled before anything is built

The engagement opens by writing the design down and reviewing it with the people who will run it.

  • The calling applications in scope
  • Where ARC sits in the request path
  • The model and tool endpoints it reaches
  • The identity and access path
  • The observability boundary
  • The approved data sources
  • Who operates each part afterwards
If the model mix or the infrastructure is not decided yet, that is assessment and planning, and it comes before this.
Plan the model mix before you commit

Connect applications and agents through ARC

ARC governs how users, applications and agents access models, tools and approved data sources. It does not govern an agent's reasoning, planning, application logic, runtime or business accountability.

Illustrative scenario

Without a control plane

Every team makes its own decisions

Application A

Provider
One, hard-wired
Fallback
None
Privacy
A rule in the code
Cost
No shared spending rule

Application B

Provider
A different one
Fallback
Retry the same one
Privacy
No defined rule
Cost
A cheaper default

Application C

Provider
Two, chosen by hand
Fallback
Manual escalation
Privacy
A separate rule again
Cost
No central limit

Four decisions, answered separately in each codebase. Reviewing or changing one means a release for every application that carries it.

One governed model layer

Central policy, applied consistently

  • Application A
  • Application B
  • Application C
ARCOne control plane

One policy layer

  • Access
  • Privacy, defined by you
  • Capability
  • Limits and budget
  • A direct route to an approved model
  • A composed route, with tools around the model call
  • A failover path to a destination you approved
Applications no longer carry the policy themselves. Update the controls in ARC and every connected application continues through the same governed endpoint.
  • OpenAI-compatible provider and private endpoints
  • Applications and API clients
  • Identity and access controls
  • Approved tools and functions
  • Observability and reporting interfaces
  • Connected vector storage on the Self-hosted Licence

Application logic, data pipelines and the wider estate stay with your own teams. The review follows one request end to end and names what leaves your boundary at each hop:

  • The path a request takes out of the calling application
  • Where classification and privacy handling execute
  • What each destination is sent
  • Where the result returns to ARC
See how ARC worksSee how a request is handled

One application first, on purpose

A first path is narrow on purpose. Policy and route behaviour are far easier to judge when a single application depends on them.

  1. 01One pathOne application, one list and a bounded set of models, connected end to end before anything else moves.
  2. 02ValidateWatch policy and route behaviour on that path, and confirm who owns each part while the scope is still small.
  3. 03WidenAdd the next application or list against a design that has already run.
A list groups the models, tools and rules approved for a team or use case.
Choose where ARC runs

An architecture your team can operate

The design transfers with the ownership split written beside it.

  • The approved target architecture
  • The integration points as built
  • Who owns each part in operation
  • The deployment decisions still open
See Managed Service operationsEmbed ARC in your software

Design the ARC integration

Thirty minutes with an engineer. Your numbers. A straight answer.

Discuss your requirementsSee where ARC sits in real environments