OnVeris

THE ONVERIS PLATFORM

Private AI
Connected to the way
you work

Give your people AI that works with your documents, business data and everyday systems. Keep model inference within infrastructure you control.

OnVeris Chat showing available assistants and model controls
OnVeris Chat - a familiar workspace for your team. View full size

FROM INFRASTRUCTURE TO EVERYDAY WORK

One platform: Three connected parts

Run your models, manage your organisation and give people a familiar way to use AI.

ONVERIS GATEWAY

Your control centre

Manage users, assistants, API keys, knowledge bases and connectors from a central administration interface. Route requests to your connected nodes.

ONVERIS NODE

Your inference engine

Host and run AI models on your own hardware. Connect multiple nodes to the Gateway as your model choices and capacity requirements develop.

ONVERIS CHAT

Your team's workspace

Make assistants available in a browser, on Windows, macOS and Linux, and on iPhone, iPad and Android. Your applications can use the API too.

PLATFORM CAPABILITIES

From a capable model to a capable assistant

Build the experience around the task. Keep relevant knowledge close, and make the right tools available.

Purpose-specific assistants

Choose instructions, models, knowledge and tools for each assistant. Create a focused experience for the people who need it.

Document knowledge bases

Import reference material and make it searchable. Attach selected knowledge bases to an assistant to ground answers in your content.

Connector and tool controls

Select the connectors and individual tools an assistant may use. Configure permission prompts for supported actions in chat.

Choice of models

Run locally hosted models on your Nodes. Assign models to assistants and configure fallback models where appropriate.

OpenAI-compatible API

Connect supported applications through familiar API patterns for chat, responses and embeddings. Manage access using API keys.

Desktop, web and mobile

Give staff a consistent place to chat. Desktop connectors can also reach permitted files and tools on the person's own computer.

CONNECTED TO YOUR BUSINESS

Your business systems: One useful conversation

Bring live business data into the conversation with a growing catalogue of connectors. Give each assistant the specific tools it needs, from customer records and accounts to HR and everyday collaboration.

HubSpot
CRM & sales

HubSpot

Look up contacts, companies, deals and support tickets. Give a sales assistant the customer context it needs.

Xero
Accounts & finance

Xero

Read invoices, bills and financial reports. Create draft invoices for a person to review in Xero.

BreatheHR
HR & people

BreatheHR

Find employee information and leave records. Use the available tools to book or review leave requests.

SharePoint Outlook Teams
Documents & communication

Microsoft 365

Search SharePoint and OneDrive files, read Outlook email and work with Teams chats through dedicated connectors.

Stripe Trello
Payments & projects

Stripe & Trello

Look up customer and payment information in Stripe, or work with boards, cards and checklists in Trello.

OnVeris custom connectors
Your own systems

Custom connectors & MCP

Add predefined tools for internal services, connect an MCP server, or build an OnVeris connector for your own workflow.

Connector capabilities vary by service. Administrators configure credentials and access. Connections to cloud services use those services; local AI inference remains in your chosen infrastructure.

CONTROL WHAT YOUR AI CAN DO

Useful access: Deliberate boundaries

Decide which people and applications can use an assistant, then select the knowledge and tools it can reach. Review shared credentials and connector access as part of your deployment.

Read about assistant controls ↗

Choose tools precisely

Grant a whole connector or selected tools. Use read-only capabilities when an assistant only needs to find information.

Ask before supported actions

Set permission prompts for chosen tools in chat. Tools that require approval are not offered through unattended API requests or scheduled tasks.

Keep deployment in your hands

Manage your Gateway, Nodes and locally hosted models. Connecting to an external service is an explicit choice made through a configured connector.

VISIBILITY ACROSS YOUR DEPLOYMENT

Manage the models: Understand the workload

See connected Nodes, inspect hardware and model executors, and explore usage and performance through the Gateway.

OnVeris Gateway Nodes page showing a connected Node and usage statistics
Nodes - connected inference capacity and usage at a glance. View full size
Node hardware monitoring with CPU, memory, GPU and model executors
Node details - hardware monitoring and model executors. View full size
OnVeris analytics with usage metrics, filters and requests-over-time chart
Analytics - explore requests, model activity and usage over time. Values shown are from the supplied example deployment. View full size

MAKE IT FIT YOUR BUSINESS

Connect what you have: Build what you need

OnVeris connectors

Install from the marketplace, import a connector package or create your own predefined tools in Python for an internal service.

How connectors work ↗

MCP connections

Connect tools provided by an MCP server alongside OnVeris connectors, with the authentication and access options that server supports.

Explore connector options ↗

Existing applications

Use the OpenAI-compatible API to bring private model inference and configured assistants into supported applications and workflows.

Explore the API →

START WITH CONTROL

Roll out useful AI: At your own pace

Start with a focused assistant, a suitable model and a defined set of users. Add knowledge, connectors and more capacity as your needs grow.

  1. 1

    Choose your environment

    Deploy the Gateway and Nodes within infrastructure you control.

  2. 2

    Define the assistant

    Set its instructions, model, knowledge sources and available tools.

  3. 3

    Give your team access

    Share the assistant with selected users or applications, then expand from there.