Let your teams use Claude Code and Claude Desktop on your most sensitive work, with access, spend and business rules under your control and a record of every request. Runs in your own AWS account on Amazon Bedrock.
For Claude Code and Claude Desktop on Amazon Bedrock.
Both are reasonable. TrustOps gives you the proof, so Claude can reach the work that matters most.
We can't let it near the sensitive repos.
Requests go straight to a vendor endpoint outside your boundary.
Who spent all those tokens?
One cloud bill, no owner per person or team, and no cap until after the fact.
It doesn't know our rules.
Built-in safety covers harmful content. Your confidential matters are a different list.
If something goes wrong, can we explain it?
No record of who asked what, which rule applied and what came back.
Claude knows what's unsafe. It doesn't know what's off-limits at your company.
Guardrails stop harmful content. TrustOps enforces your business rules, including on questions that would be perfectly reasonable anywhere else.
Every Claude request passes two checkpoints before it reaches a model.
Your VPC
Traced end to end with TrustHouse Policy Lineage, AWS CloudTrail and Amazon CloudWatch.
Control over who can use Claude, which models and tools they get, and what they can spend.
Control over what is safe to send, with your business policies, masking and a record of every request.
Claude apps gateway decides who can use Claude. TrustHouse decides what is safe to send.
One console for your admins, deployed and run by TrustHouse.
What you get each time someone presses Enter.
Your people use their work identity, and nothing long-lived sits on their machines.
Sign in with gateway
you@yourcompany.com
Platform engineeringEach group gets the models, tools and connectors you approve, and nothing else.
First matching group applies. Everyone else catches the rest.
IllustrativeSet limits by organisation, team or person, and watch spend against them as it happens.
Spend counters estimate cost from token counts at list price. They work as a real-time brake. Your AWS bill stays the record.
Organisation default
Group limit
Per-person override
highest precedence
Spend vs cap, this month
Every prompt and response is checked against your policies, so sensitive matters are masked, blocked or answered with a safe reply.
Bedrock Guardrails, IAM and CloudTrail stay in place, and TrustOps adds to them.
Amazon Bedrock
For any answer Claude gave, or didn't give, you can show who asked, which policy applied and why.
Two requests your staff might reasonably make, with and without your policies in the path.
“I heard one of our competitors is under a regulatory investigation over customer data. What's going on?”
I’m not able to comment on that. It involves an ongoing legal or regulatory matter. For questions like this, please contact Legal or Corporate Communications.
TrustHouse | Active litigation and legal hold | input and output | safe fallback
Illustrative. Example policies from the TrustHouse policy library. Your policies use your wording.
Roll Claude out team by team, with one sign-in and only the tools each group needs.
Claude runs in your AWS account, your business policies apply before it answers, and every request leaves a record.
Spend caps and clear attribution by person and team, so costs stay predictable.
The same Claude Code and Claude Desktop, with one work sign-in and no change to how they work.
Each step ends with a check your security team can see.
Connect your identity provider, choose the AWS account and network path, and switch on audit logging.
We deploy both gateways in your account and set up your groups, policies and spend caps.
Turn on Bedrock Guardrails, cost allocation, budgets and dashboards.
Move your current Claude users across, review the results with security, then widen access to sensitive work.
A leading enterprise moved from Claude Direct to Claude on Amazon Bedrock, with TrustHouse providing the trust layer from day one. Their teams kept the Claude experience they already relied on.
Under 1 week
to make the switch
No change
in the user experience
Your data, cloud and IP stay yours.
No. They keep Claude Code and Claude Desktop and sign in once with their work credentials. From then on, their requests route through your gateways.
Both gateways run in your AWS account, and Claude runs on Amazon Bedrock. Traffic between them stays on a private connection.
No. The Claude apps gateway is Anthropic's self-hosted gateway. Claude TrustOps is a TrustHouse service that deploys, configures and runs it, and adds TrustHouse policy enforcement and lineage on top. TrustHouse is not affiliated with or endorsed by Anthropic.
Every request is checked against your policies, which adds some latency compared with going direct. Typical overhead: to be confirmed.
Keep them. Guardrails handle content filtering and sensitive information. TrustOps adds the parts Guardrails are not built for: who can use Claude, what each group can do and spend, policies written in your business language, and a record of every request.
Claude Code and Claude Desktop.
Yes, and we recommend it. Start with one group, review the results with security, then widen access.
Per managed user, with Claude usage on Bedrock passed through at applicable rates. See pricing for details.
Your data, cloud and IP stay yours.
Inference runs on Amazon Bedrock in your AWS account.
Caps before spend runs away, with visibility by person and team.
Your business policies enforced on every request.
A record of every answer, and why.