Introducing

Meet your new AI Companion

Built into Complete Control.

Same Complete Control you know, with a new pair of eyes. Upload your contracts and AI Companion reads them for you, pulling out what matters and answering your questions directly, so you don't have to search and dig through large volumes of agreements and documents yourself.

What it does

Three things it's good at

Not a separate tool to learn — AI Companion lives where you already work in Complete Control.

Smart ingestion

Upload a contract. AI Companion extracts key fields in seconds, each one backed by a reference to the right page in the document.

Portfolio intelligence

Ask about your whole portfolio and get answers based on your own data, not a generic guess.

Ask the contract

Ask plain-language questions about a single agreement. No more hunting through PDFs.

See it in action

AI Companion, as it works today

A short walkthrough of today's functionality. Updated as AI Companion develops further.

For pilot customers

You're here because you signed up

You're getting early access to AI Companion because you told us you wanted to test it and give us feedback along the way. Thank you — that's how we make it better.

Rollout
All pilot customers test on their own account in Complete Control. Everyone with a user gets access to AI Companion, there's no separate sign-up per person.
Feedback loop
House of Control facilitates a feedback session during the pilot, a meeting of up to 30 minutes. Afterwards, every pilot customer is followed up with what we found and how it shaped the product.
Summary of findings
After the pilot, we'll send out a summary of what we found — key takeaways, tips, and feature requests that came out of the pilot customers' feedback, along with an update on what happens next.

Plan

Test phase: 5–30 October — pilot customers use AI Companion in their daily work, before we gather feedback together.

  1. 5 October

    First pilot customers go live

  2. 19–30 October

    30-minute feedback session with House of Control

  3. 30 October

    Pilot phase ends. Goal: every pilot customer is onboarded and continues as one of our very first AI Companion customers

  4. Mid-November

    Rollout begins for first movers among the remaining customers

Questions and answers

Security and data handling

What it is

What is AI Companion, and what doesn't it replace?

AI Companion is the AI functionality built into Complete Control. It helps you import, find and understand information in your contracts faster. 

Can AI Companion change data in Complete Control on its own?

No. It can read information you already have access to and suggest text or values, but it can't itself save, update or overwrite data. You always decide whether to use a suggestion.

Can we control who has access to it?

Access follows Complete Control's existing authentication and access controls — AI Companion isn't a way around them. Availability of the feature itself can only be configured per customer account — not per individual user.

What happens if it gets something wrong?

Tell your contact person at House of Control. Given the nature of AI agents as non deterministic, AI Companion is meant as support, not the sole source of truth — double-check important decisions against the original document. AI can make mistakes, the human operator is responsible for the correctness of data imported data into Complete Control.

Data and security

What data does it have access to?

Whatever is needed to answer what you ask: the text you write, content and attachments you already have access to in Complete Control, and other relevant information made available for that specific request. It follows your existing access — it never gives you access to anything you don't already have.

Is our data used to train AI models?

No. AWS confirms that neither AWS nor third-party model providers use input or output from Amazon Bedrock to train or improve the underlying models. Documents, questions and answers are never used as training data.

Is data sent to Anthropic?

No. AI Companion uses Claude Haiku 4.5, built by Anthropic, but the model runs through Amazon Bedrock. AWS does not share customer input or output with Anthropic, so Anthropic never receives your prompts, document content or answers.

Does data ever leave the AWS environment?

Not as part of the AI processing. Data sent to the language model is processed through Amazon Bedrock within AWS's EU environment. There's no separate transfer to Anthropic or any other external party.

Is Anthropic a new sub-processor?

Not in practice. Since AWS doesn't share customer data with Anthropic through Bedrock, Anthropic never receives it and doesn't act as a separate recipient. AWS carries out the technical processing and is covered by our existing data processing agreement.

Where are prompts and answers stored?

In Complete Control's own database, within your organization's own tenant — logically separated from other customers' data. Amazon Bedrock doesn't retain prompts and answers as a permanent dataset by default, and Anthropic never receives a copy.

How long are prompts and answers stored?

Prompts are stored in Complete Control for up to 7 days, then deleted from the regular AI history. The rest of your data follows Complete Control's normal retention and deletion rules.

Who can read prompts and answers?

Only what follows from Complete Control's normal access controls within your organization's tenant. Other customers can't see your AI history, and Anthropic has no access through Amazon Bedrock. Internal access to production data is restricted through established access controls.

Is one customer's data ever used to answer another customer's questions?

No. Each request only uses context from your own organization's tenant. One customer's documents, prompts or answers are never used as context for another customer.

Is data encrypted?

Yes, both in transit and at rest. Communication between Complete Control and Amazon Bedrock is encrypted with TLS, and data stored in the AWS environment is protected by Complete Control's standard encryption at rest.

What security certifications are in place?

Amazon Bedrock is covered by several of AWS's established security and compliance programs, including SOC 1/2/3, ISO 27001, 27017, 27018, 27701, 22301, 20000, and CSA STAR Level 2. Bedrock can also be used in GDPR-compliant solutions.

What happens in the event of a security incident?

A security incident related to AI Companion is handled through the same established processes as the rest of Complete Control, via Visma's cloud delivery model. Affected customers are notified in line with the agreement and applicable privacy regulation.

Agreement and ending the pilot

Is this covered by our existing agreement?

Our latest Data Processing Agreement (DPA) covers AI Companion, including the AWS sub-processor entry for managed AI infrastructure processed within EU regions. Customers on an earlier version need to sign the latest one before access is granted.

Do we need to sign an updated DPA or terms before activation?

The latest version of our DPA (Version 2026:1, published 2026-09-17) must be signed by the Customer before access to AI Companion is granted.

What happens to our data if we leave the pilot or end the agreement?

Prompts are deleted no later than the regular 7-day retention period, even if you stop using AI Companion. There's no separate copy at Anthropic that needs deleting. The rest of your data in Complete Control follows the same retention rules as the rest of your agreement.

After the pilot

Rolling out to the full customer base

What we learn during the pilot shapes how AI Companion rolls out further. This section fills in once the framework is ready.

Access

After the pilot phase, we'll start a controlled rollout to existing customers who want AI Companion and have it included in their agreement.

Pilot results

We'll use feedback from the pilot to improve AI Companion and to shape the roadmap for further development. We'll also ask some pilot customers for permission to use quotes from the pilot phase in marketing.