There's a lot of noise about AI right now, and most of it comes back to the same worry. AI feels like something that happens to your business rather than something your business directs. 

We think that worry is fair. Not because AI is dangerous by default, but because AI without control can be unpredictable, and unpredictable tools don't get adopted. They get quietly abandoned or used in ways nobody signed off on. 

Here's what we believe at Hatz. Confidence comes from control. Productivity follows confidence. You can't skip the first step and expect the other two to show up. 

AI for your business, delivered the way trusted technology should be 

Hatz is built for the people who use it every day. Small and mid-sized businesses, the teams inside them, and the work they need to get done. 

But we made a deliberate choice about how it reaches them. Hatz is delivered through managed service providers, the same partners businesses already trust to run their email, identity, devices, backups, and security. 

That's not a sales channel decision. It's a control decision. 

Every other piece of important technology in your business arrived with checks and balances. Someone configured it, secured it, set the policy, and kept an eye on it. AI deserves the same treatment. MSPs already know your environment, your compliance needs, and which departments handle sensitive data. Putting AI inside that managed layer means you get the productivity while someone accountable holds the controls. 

The three things you need to control in AI 

Every AI governance conversation comes back to three questions. Where does my data go? What is this going to cost? Who is using it, and for what? 

We organize the Hatz platform around those exact questions. Security, spend, and scope. 

Security controls: your data stays yours 

The fastest way to lose confidence in AI is to wonder where your information went after you hit enter. Hatz is built so you never have to. 

• No training on your data. Your prompts, files, and outputs are never used to train AI models. 

• Segregated storage. Data lives in secured data centers, logically separated by tenant, organization, and user. 

• Inference isolation. The model layer is separated from storage, so an LLM has no persistent access to past interactions unless a user brings them in. 

• Identity controls. MFA and SSO enforcement, so AI access follows the same rules as the rest of your stack. 

• Audit logs and compliance exports. See who did what, when, and export it for review. 

• Independent validation. SOC 2 Type I and Type II, regular third-party penetration testing, and external security reviews built into development. 

• 
  

 Full reports and documentation live in the Hatz Trust Center at trust.hatz.ai. 

Spend controls: predictable cost and more from every credit 

Most AI cost problems aren't caused by AI being expensive. They're caused by nobody setting a policy and nobody watching until the bill arrives. 

Controlling cost 

• Monthly allowance and Extra Usage caps. Every business has a defined credit budget. Your MSP decides whether you stop there or continue to a hard ceiling they set. Never an accident. 

• Per-user credit limits. Roles carry a monthly ceiling per user. Different teams, different ceilings. 

• Model policy. Set the default model and disable the ones you don't need, by tenant and by role. Each layer can only narrow the one above it. 

• Alerts and visibility. Notifications fire as limits approach, and dashboards break down usage by user, model, app, workflow, and agent, with CSV export and an API. 

Optimizing value

• Auto model selection. Auto reads every request and routes it to the right model for the task, so simple questions stop running on premium models. No extra charge. 

• Auto modes you control. Your MSP chooses which modes each role can use. Lite for everyday work, Performance for balanced tasks, Turbo only where it's genuinely needed. 

• Model multipliers. Every model shows its credit cost relative to the flagship, so allow decisions are informed, not guessed. 

• Same quality, fraction of the cost. For most day-to-day work, Auto Lite delivers the same or similar output for [less than a quarter of the credits]. 

• AI spend that behaves like any other managed service line. Predictable, reviewable, efficient. 

The result is AI spend that behaves like any other managed service line. Predictable, reviewable, and efficient.

Scope controls: AI where you want it and nowhere else 

The third question is the one most platforms skip. Not just who can log in, but what they can do once they're there. 

• Custom roles. Define exactly who can chat, build in Workshop, create workflows, publish to the community, manage phone agents, or administer settings. 

• Feature toggles by role. Memory, personalization, subagents, and artifact authoring, each on or off per team. 

• Integration and custom MCP access. Deny specific integrations or block custom MCP servers for a role. Anything denied at the organization level is locked below it. 

• Groups and sharing. Sales gets the sales tools, finance gets theirs. 

• Tenant templates. Your MSP bakes all of this into a template so you're configured from day one, not after the first surprise. 

When people know the boundaries, they use AI more, not less.

Control is where confidence starts 

We're not trying to make AI feel less powerful. We're trying to make it feel like yours. 

Security, spend, and scope aren't features bolted onto a chat window. They're the reason Hatz is shaped the way it is, and delivering it through your MSP means a real person who understands your business is holding the controls. 

That's our promise. AI you can direct, measure, and trust, delivered by the partner who already keeps your business running. 

Ask your MSP about Hatz. If you're an MSP, your Hatz account team can walk you through setting these controls up for every client you serve.