For managed service providers

The Hiveframe agentic AI platform for MSPs.

Hiveframe connects AI agents to your clients' CRM, email, accounting, and databases. Each client gets its own isolated AWS environment running our gateway, with its own encryption key and secrets, a token check on every call, and logs that never store the data. We build and manage the agents. You keep the client.

See how it's built
Overview of Hiveframe for one client. Your MSP sends service account credentials one way into the client's own environment, where they are stored with that client's key. The client's staff and AI agents reach the Hiveframe MCP gateway inside that environment through a token check on every call. The gateway reaches the client's systems, such as Microsoft 365, QuickBooks, and a SQL database, through service accounts you create and can disable. Every tool call is written to a log that records the action, not the data. Other clients each have their own separate environment. ONE ENVIRONMENT PER TENANT every client gets its own YOUR MSP WRITE-ONLY SECRETS credentials go in, never back out SECRETS · OWN KEY HIVEFRAME MCP GATEWAY TOOL-CALL LOG CLIENT A CLIENT STAFF in their chat app AI AGENTS on a schedule TOKEN CHECKED ON EVERY CALL their identity provider decides who CLIENT SYSTEMS M365 QUICKBOOKS SQL DB through service accounts you create and can disable LOGS RECORD THE ACTION, NOT THE DATA which tool, who, when, result Overview of Hiveframe for one client. Your MSP sends service account credentials one way into the client's own environment, where they are stored with that client's key. The client's staff and AI agents reach the Hiveframe MCP gateway inside that environment through a token check on every call. The gateway reaches the client's systems, such as Microsoft 365, QuickBooks, and a SQL database, through service accounts you create and can disable. Every tool call is written to a log that records the action, not the data. Other clients each have their own separate environment. YOUR MSP WRITE-ONLY SECRETS LOGS RECORD THE ACTION, NOT THE DATA SECRETS LOG HIVEFRAME MCP GATEWAY ONE ENVIRONMENT PER TENANT TOKEN CHECK EVERY CALL STAFF AGENTS CLIENT SYSTEMS through service accounts you create and can disable

The platform

Every tenant gets its own AWS account, with our gateway inside.

Hiveframe is the whole platform, the AWS accounts we build and maintain and the MCP gateway running inside each one. Every tenant, meaning one of your clients, gets a brand-new AWS account of its own, separate from every other tenant's in every way. That client's AI agents do all their work through the gateway in it.

Your MSP
You create a scoped service account in each of the client's systems and send its credential to Hiveframe through a write-only path.
Hiveframe
The platform. One AWS account per tenant, built and maintained by us, with our MCP gateway running inside it. It holds that tenant's secrets, encrypted with that tenant's own key. Nothing in it is shared with any other tenant.
Agents
Built and run by us. They call tools only through the gateway in their own tenant's account. The token is checked on every call, and every call is logged.
Client business systems
Tools reach the client's systems only as the service accounts you created. Disable one and every tool that uses it stops.
Four sections, left to right. Your MSP: the service accounts you create, whose credentials go to Hiveframe. Hiveframe: one hexagon per tenant, each a separate AWS account with its own MCP gateway inside, and an open slot for the next tenant. Agents: drawn as small bee marks, each working only through its own tenant's gateway. Client business systems: each tenant's own systems, reached only as your service accounts. YOUR MSP HIVEFRAME AGENTS CLIENT BUSINESS SYSTEMS SERVICE ACCOUNTS TENANT A AWS ACCOUNT GATEWAY TENANT A SYSTEMS CRM · EMAIL TENANT B AWS ACCOUNT GATEWAY TENANT B SYSTEMS M365 · QUICKBOOKS TENANT C AWS ACCOUNT GATEWAY TENANT C SYSTEMS SQL DB · FILES OPEN SLOT Four sections, left to right. Your MSP: the service accounts you create, whose credentials go to Hiveframe. Hiveframe: one hexagon per tenant, each a separate AWS account with its own MCP gateway inside, and an open slot for the next tenant. Agents: drawn as small bee marks, each working only through its own tenant's gateway. Client business systems: each tenant's own systems, reached only as your service accounts. YOUR MSP SERVICE ACCOUNTS HIVEFRAME TENANT A AWS ACCOUNT GATEWAY TENANT A SYSTEMS TENANT B AWS ACCOUNT GATEWAY TENANT B SYSTEMS TENANT C AWS ACCOUNT GATEWAY TENANT C SYSTEMS AGENTS CLIENT BUSINESS SYSTEMS

What every tenant account has

Read the security details

Its own environment
A separate AWS account per tenant, on serverless infrastructure. Nothing is shared between tenants, so a problem in one can't reach another.
Its own key and secrets
Service account credentials are encrypted with a key for that tenant alone. The path you send them through is write-only. The model never sees them and they never appear in logs.
A token check on every call
The gateway validates the caller's token and checks its groups before any tool runs. The client's identity provider decides who can use which tools.
Logs that never store the data
Every tool call is logged: which tool, who triggered it, when, and whether it succeeded. The records a tool returns are not kept.
A clean exit
Give 30 days' notice. We export that tenant's configuration and logs, you disable the service accounts, and nothing of ours can reach the client's systems.

What it runs

The AI agents and tools we build and manage for each tenant.

We tie the gateway into the client's systems, build each agent for a specific workflow, and manage it under a monthly care plan. Governance comes with it. The client's identity provider decides who can use which tools, access is limited to what each job needs, every action is logged, and you can shut any agent off right away.

Agentic AI doesn't just answer questions. It takes the next step in the software the business already uses.

Semi-autonomous
A person works in a chat client the business already uses, such as Claude or ChatGPT, and it connects to the tenant's gateway.
Fully autonomous
A schedule or system event starts the agent. Each one gets its own service account, scoped to its one job.

Three examples

  1. Customers

    CRM · Email

    ProblemA long-time customer emails, upset that an order is late.

    AgentThe agent finds them in the CRM, checks the order and their account, and drafts a reply with the real ship date and a goodwill credit. Nothing goes out until you approve it, and the case is logged.

  2. Operations

    Inventory · Projects

    ProblemA supplier pushes a part back two weeks.

    AgentThe agent finds every open job that uses the part, proposes new dates, and lists the customers whose jobs slip with a drafted note to each. You review the schedule changes and the notes before anything is updated or sent.

  3. Finances

    Accounting · Billing

    ProblemThirty invoices are past due at month end.

    AgentEvery Monday the agent pulls overdue invoices, checks each customer's open jobs and recent emails, and drafts a reminder that fits. Customers with an open complaint are held back. You get one report with every draft ready to approve.

How it plugs in

Little to install. Little to change on the client's side.

A new tenant needs its own AWS account, which we build, a service account for each system, and the client's existing sign-in. That is the whole footprint.

We connect cloud systems, databases, files, and legacy software. Where the vendor's own connector fits the job, we use it behind our gateway. Where it doesn't, we build a custom tool.

One service account per system
You, or the client's IT, create a scoped service account in each system the tools need. We never ask for admin accounts.
Nothing installed for cloud systems
Tools reach cloud systems through each system's API, the way any cloud app does.
One connector for on-prem systems
A database or file share on the client's network needs one small tunnel connector. It connects outbound only, so nobody opens an inbound firewall port.
The sign-in they already have
People sign in through the client's identity provider, such as Microsoft Entra ID, Okta, or Google Workspace.
One environment per tenant, built from code
The same Terraform stack builds every tenant's AWS account.
per-tenant stack
hiveframe-tenant-infra
Terraform stack. Creates one tenant's environment.
hiveframe-mcp-gateway
The gateway app. Lambda functions, deployed once per tenant.
request path
HTTPS -> API Gateway at the tenant's endpoint -> Lambda -> tool -> client system

How you offer it

Under your brand, or by referral. You keep the client.

We sell through MSPs. You own the client relationship, and we work behind you, white-label by default.

  1. Client business Keeps running the software it already uses
  2. You Own the relationship, create the service accounts, approve the plan
  3. Hiveframe Systems Builds, hosts, and manages the platform, the agents, and the tools

Bringing us your first client

Discovery and delivery are fully remote. We don't do on-site visits.

How partnering works

  1. 01

    You bring us in

    Refer the client, or resell the work under your name.

    Refer or resell
  2. 02

    Fit and discovery calls

    A 20-minute fit call, then a 1-3 hour walkthrough of the client's systems.

    Free
  3. 03

    Plan and bid

    A written plan with a flat price. You approve it before we build.

    Written, flat price
  4. 04

    Build and connect

    You create the service accounts, and we build, host, and connect the tools.

    We do the work
  5. 05

    Monthly care plan

    Hosting, monitoring, fixes, and tool changes as the business changes.

    Required, hosted by us

Pricing

Per agent. A flat-price build, then a monthly care plan.

The build price is set in a written plan after discovery. The care plan is required because we host the tools. Partner pricing is on request, and it leaves room for you to mark it up.

See what each tier includes

Contact

Book a free 20-minute partner call.

Email us to set a time. We'll cover which clients fit and whether you'd resell or refer.

Every build has a 30-day money-back guarantee.