C

Contractbook

AI Import reads existing contracts, extracts structured fields and helps turn a collection of files into a usable agreement database. The Contract Assistant adds plain language questions, summaries, risk identification and clause revisions. Both operate inside a wider application with template drafting, collaboration, electronic signatures and ongoing contract management. Templates and configured workflows still handle repeat agreements independently of the assistant.

Customers can combine the extracted dates and obligations with reminders, while CRM data populates new agreements. Essential includes the assistant; higher plans add broader data extraction, custom fields, automation and team administration around these AI functions.

AI Import connects extracted data to the text it came from. Selecting a purple field in the sidebar moves the reader to its exact location in the original agreement and highlights the passage. Users can inspect individual values or verify all fields together, and the repository marks completed verification. Import first processes the uploaded files and extracts their contract data, with OCR included for relevant files.

This is source based verification of customer agreements, rather than a legal citator. No test set, measured accuracy result or explicit unsupported answer behavior accompanies the Contract Assistant's broader clause recommendations.

An AI import can be canceled during upload, OCR or extraction, and the canceled file is discarded. After processing, a reviewer checks the proposed data against highlighted source text and marks the fields verified. That verification can become the trigger for a reminder or a move into a shared space. Users need the relevant company permissions to import files and manage custom contract types. Template workflows also support approval steps before signing.

These controls give teams review and cancellation surfaces, but the assistant has no stated confidence cutoff or defined escalation path for an answer that cannot be supported.

Bizzkit's compliance officer describes saving 15 to 20 hours extracting data from about 50 legacy data processing agreements. That is a named example of AI extraction in use, with the time estimate supplied by the customer. Lenus separately describes reducing contract preparation from 15 minutes to five through templates and integrations with Salesforce, Greenhouse and HiBob. Those automation results are not presented as a measurement of the newer Contract Assistant.

GANNI and Pave also discuss repository and extraction benefits. No dated comparative study isolates the assistant's effects on review quality or legal outcomes.

Section 9.12 of the service terms bars AI model training on retained platform data unless the customer gives consent in writing or through the application. Confidential information is limited to service delivery and remains protected after termination. AI imports start in the user's private contract view, while sharing depends on the application's space permissions. Google Cloud hosts platform data and Azure hosts the language model, both with EU locations in the processor list.

Privilege and work product handling are not addressed directly, and the terms do not specify what the model provider retains from individual prompts or responses.

The service is restricted to business activity and is not offered to consumers. Section 8 describes the supplied templates as drafting inspiration rather than legal advice, and the company profile makes the same tooling distinction for the product. Legal teams can control templates, approvals and shared spaces while other employees generate routine agreements. The customer remains responsible for using suitable agreements and deciding whether electronic signatures are appropriate.

No jurisdiction specific competence guidance or named professional supervision standard accompanies the assistant. Its explanation and rewriting features therefore sit inside the customer's own legal review and contracting responsibilities.

Contractbook's trust center describes security policies, change management and an incident response program. Management reviews that security program at least annually, and production data is restricted in development and testing environments. The AI features separately support source highlights, user verification and consent for model training. None of these materials sets out an AI release evaluation process, a responsible AI framework or results showing uneven performance across contract types or user groups.

The page titled AI Policy Page is a company and product profile for automated readers, rather than a policy assigning accountability for the behavior of the language model.

Customers can remove or archive their own data through the application and request removal at termination. Signature verification information has separate retention rules and generally requires all signers' consent before deletion. The privacy policy gives a general service data retention period with legal exceptions, while the processor list identifies providers, roles and locations. Space permissions, encrypted storage, logging and a trust center incident plan add operational controls.

Confidentiality breaches require prompt notice under the terms. A specific retention window for assistant prompts and outputs, together with a deletion schedule at the model host, is not stated.

The service agreement limits liability to fees paid in the 12 months before the first claim arose. It excludes lost profit, business opportunity, third party liabilities, data loss and several other categories of loss, and supplies the service as is. Death or personal injury caused by negligence, fraud and fraudulent misrepresentation are outside the stated limitations. The customer, rather than the vendor, gives the agreement's indemnity.

Contractbook promises customary insurance without naming its amount or scope. These provisions do not provide a specific warranty or recovery route for a mistaken AI clause explanation, extraction or revision.

The HubSpot integration moves deal and company data into contract templates and can synchronize negotiated values back after signing. Setup requires defined template fields, the appropriate Contractbook template or company permissions, and specified HubSpot CRM, product library and marketplace rights. The administrator connects the accounts through Integrations and accepts the requested permissions, then configures triggers and actions.

HubSpot quotes can become attachments, and its widget displays contract status and activity. Disconnecting unpublishes related automations without deleting generated files. Salesforce, API access, Zapier and Make extend other workflows, with availability and charges varying by plan.

Contractbook runs as an online contract workspace with organization accounts, shared spaces and individual access rights. The subprocessor list places Google Cloud platform hosting and Azure language model hosting in the EU. It also identifies United States services for email, messaging, monitoring, traffic management and integrations, so the EU hosting entries do not describe every supporting data flow. The privacy policy permits transfers outside the EU and EEA with safeguards.

No country selection, customer hosted installation or private tenancy option is specified, and the available material does not allocate storage and processing locations separately for each subscription plan.

The security page states SOC 2 compliance and describes recurring penetration tests, encrypted credentials, verified accounts and two factor signing. A Secureframe trust center offers a SOC 2 report through a request control and lists security policies covering access, change management, risk and incident response. Its public material does not identify the assessment period, report type, auditor or detailed system scope.

The separate Openli profile could not be accessed, so any material there remains unknown. Customer signatures and identity controls support contract execution, while the SOC 2 claim addresses service controls rather than measured accuracy of the AI assistant.

Microsoft Azure hosts the large language model used for the product's AI functions, with the processor list assigning that service an EU location. Google Cloud is separately responsible for storage and platform hosting. The application combines document import, OCR, field extraction and a conversational contract assistant, but the specific model name and version are not identified. The terms require customer consent before retained platform data is used for model training.

A general notice rule for contractual updates does not provide a specific commitment to notify customers about model replacements or changes to the inference supply chain.

Essential, Centralize and Accelerate include one, five and ten users, with annual signed contract allowances of 100, 350 and 500. Template limits are five, 15 and 25. The current pricing page shows no base subscription amount; a company profile updated in March 2026 gives a starting rate of €89 per month. The pricing FAQ lists five extra users at €50 per month, or €600 annually. Automation and provisioning are optional for Centralize and included in Accelerate, while several integrations cost extra. Custom setup uses a quote, and the terms provide for upgrades or additional charges when usage limits are exceeded.

Small and midsize sales, legal, HR and operations teams use the platform for standard recurring agreements such as NDAs, sales contracts and employment agreements. The Lenus case describes legal counsel setting templates and boundaries for colleagues across several markets. Bizzkit uses extraction for legacy data processing agreements, while GANNI and Pave describe contract visibility and renewal work. The company focuses on European markets and also offers industry pages for government, healthcare, construction and procurement.

Its own profile excludes legal advice and simple one off signing as its intended role, without supplying a jurisdiction by jurisdiction support matrix.