AI Addendum

AI Addendum

Last Updated: November 21st, 2026

Version: 2026.11.21.1
Effective Date: November 21st, 2026

This AI Addendum (the “Addendum”) supplements the Countercyclical Terms of Service available at countercyclical.com/tos (the “Terms” or “Agreement”) between Countercyclical LLC (“Provider”) and the individual or legal entity identified as the customer in the applicable account or acceptance record (“Customer”).

This Addendum governs Customer’s use of Provider’s artificial intelligence features, including Agents, Orchestrations, supported integrations and automations, and the processing of Customer Content through those features. Capitalized terms not defined here have the meanings given in the Agreement or the applicable Data Processing Addendum.

1. Acceptance and Relationship to the Agreement

1.1 Acceptance

This Addendum applies to a Customer workspace when a user authorized to accept contractual terms on Customer’s behalf affirmatively accepts it through Provider’s designated acceptance process. Acceptance is workspace-wide and applies to the AI Services enabled for that workspace. Provider requires acceptance before enabling the applicable AI functionality.

A person accepting this Addendum on behalf of an organization represents that they have authority to bind that organization. Customer is responsible for ensuring that the person accepting has the necessary authority and for informing its authorized users (“Users”) of the terms governing their use of the AI Services.

Provider will maintain an acceptance record identifying the Customer and workspace, the accepting user, the accepted Addendum version, and the date and time of acceptance, together with the applicable acceptance method.

Workspace membership, use of a non-AI feature, submission of a prompt, upload of a file, or a Model-generated instruction does not independently constitute acceptance of this Addendum.

1.2 Contractual Hierarchy

This Addendum supplements the Agreement and governs AI-specific matters only to the extent consistent with the Agreement and the applicable Data Processing Addendum (“DPA”).

The following order of precedence applies:

  1. Mandatory applicable law controls to the extent its requirements cannot be waived or varied by agreement.

  2. A separately executed Order Form or amendment controls only those provisions it expressly and specifically overrides.

  3. The DPA controls the processing of Customer Personal Data by Provider on Customer’s behalf, including applicable processor obligations and related data-protection requirements.

  4. This Addendum supplements the DPA for AI-specific processing and use matters, but does not reduce or override DPA protections.

  5. The Terms govern the general commercial relationship, including fees, intellectual property, indemnification, limitations of liability, remedies, disputes, and other general contractual matters, except where expressly overridden by a separately executed agreement.

  6. The Privacy Policy describes Provider’s privacy practices and applicable individual privacy rights. It does not independently expand Provider’s contractual rights to Customer Content or amend the Agreement.

This Addendum does not create a separate liability cap, indemnity, warranty, or remedy that conflicts with or supplements the liability and remedy provisions of the Agreement. The Agreement’s applicable limitations and exclusions of liability apply to this Addendum.

1.3 Authorized Business Use

Customer may use the AI Services for authorized internal business research, due diligence, analysis, drafting, and workspace workflows, subject to the Agreement, this Addendum, applicable law, and Customer’s rights in the information it submits or authorizes Provider to retrieve.

Nothing in this Addendum prohibits a use of Agents, Orchestrations, APIs, MCP connections, schedules, Pipeline Stage Actions, or other supported automation that is otherwise authorized under the Agreement.

Customer Content does not become non-confidential, Provider-owned, or available for unrestricted reuse merely because it is submitted through an AI feature, prompt, comment, integration, or feedback mechanism.

2. Definitions and Service Scope

2.1 AI Services

“AI Services” means Provider’s enabled artificial intelligence and machine-learning features, including Agents conversations, contextual assistance, Orchestration design and execution, supported document or file processing, and AI work initiated through authorized schedules, comments, integrations, or Pipeline Stage Actions.

Availability depends on Customer’s subscription, purchased features, workspace settings, permissions, supported capabilities, applicable usage limits, and available credits.

2.2 Model

“Model” means an artificial intelligence or machine-learning model used to process Inputs or generate Outputs.

“Training” means training, fine-tuning, distillation, reinforcement learning, or another use of data to create, modify, or improve a Model’s learned parameters.

2.3 Customer Content

“Customer Content” means information and materials submitted by or on behalf of Customer, or retrieved from sources Customer has authorized Provider to access, in connection with the Services. It includes private workspace information, files, prompts, conversation histories, instructions, selected skills and revisions, mentions, data blocks, connector records, source excerpts, and Outputs generated for Customer.

Copies, extracts, derived representations, and metadata that reveal or reproduce Customer Content are treated as Customer Content for purposes of the applicable confidentiality, privacy, and use restrictions.

Customer Content may include Customer Personal Data and Confidential Information, as those terms are defined in the Agreement or DPA. Those categories are not mutually exclusive.

Third-party content remains subject to the rights and restrictions of its respective owner.

2.4 Inputs and Outputs

“Input” means information supplied to an AI Service by or on behalf of Customer, including authorized context retrieved from Customer’s workspace or connected sources for the relevant operation.

“Output” means content generated by an AI Service in response to an Input, including answers, analyses, summaries, classifications, proposed actions, and draft artifacts.

2.5 Operational Metadata

“Operational Metadata” means service and usage records generated in connection with the operation of the AI Services, such as scoped identifiers, Model selection, timestamps, execution status, token or page counts, cost receipts, and security events.

Operational Metadata does not include raw document contents, prompts, conversation bodies, or response bodies merely because those materials are processed or referenced by an AI Service.

Operational Metadata that constitutes personal data, Customer Personal Data, or Confidential Information remains subject to the protections applicable to that information.

2.6 Availability and Deferred Features

A feature, integration, provider, template, or connector is not enabled or authorized solely because it appears in a catalog, interface, example, or technical configuration.

The initial release excludes uploaded AI file inputs, Slack file import, dictation, and xAI inference. Those capabilities may be introduced in the future subject to applicable product availability, contractual review, and any required updates to the DPA, subprocessor register, or other applicable documentation.

3. Permitted Processing and No Training

3.1 Limited Processing Authority

Customer instructs Provider to process Customer Content only as reasonably necessary to provide, secure, maintain, troubleshoot, and support the contracted Services; carry out Customer’s authorized instructions; and comply with applicable law.

Permitted processing may include retrieval, temporary caching, format conversion, document extraction where supported, Model inference, workflow execution, and delivery to a destination authorized by Customer.

This authority does not grant Provider a general right to disclose, sell, commercialize, or otherwise use Customer Content outside the purposes described in the Agreement, this Addendum, and the DPA.

Access to a source does not, by itself, authorize sending that source or its contents to another service, destination, or recipient.

3.2 No Training

Provider will not use Customer Content, Inputs, Outputs, or Customer feedback to Train any Model, whether for Provider, another customer, or a third party.

Provider will ensure that subprocessors receiving Customer Content on Provider’s behalf are subject to contractual restrictions prohibiting such Training for the content they process on Provider’s behalf.

No Customer Content, Input, Output, or Customer feedback is authorized as Training Data or for Training Purposes under this Addendum.

Use of an automatic Model-selection setting, submission of feedback, acceptance of a template, enabling an integration, or use of an AI feature does not authorize Training.

Any future arrangement permitting Training requires a separate, express written agreement between the parties and appropriate technical and contractual controls. No such authorization is granted by this Addendum.

3.3 Operational Metadata and Product Improvement

Provider may use Operational Metadata to operate, secure, maintain, troubleshoot, and improve the Services, subject to the Agreement, this Addendum, and applicable law.

For general product analytics or cross-customer improvement, Provider may use aggregated or statistical information only where it does not identify Customer or an individual, reveal Customer Content, or reasonably permit Customer Content to be reconstructed.

Provider will not attempt to reidentify information used in that manner.

This permission does not authorize Provider to create or use a shared corpus of Customer prompts, documents, or Outputs; use Customer Content for advertising; sell Customer Content; or use Customer Content for general Model benchmarking or evaluation outside the processing necessary to deliver and maintain Customer’s own Services.

Customer-specific support access must be necessary for the relevant support or operational purpose, appropriately access-controlled, and subject to applicable confidentiality obligations.

3.4 Provider and Subprocessor Configurations

Provider will use reviewed service accounts and configurations consistent with the restrictions in this Addendum and the DPA.

Automatic Model routing, fallback, and provider selection remain subject to the same applicable processing, confidentiality, privacy, authorization, no-Training, and regional restrictions as manual provider selection.

A no-Training commitment does not, by itself, mean that a provider retains no data, performs no automated safety or abuse-prevention processing, or permits no authorized operational access. Provider-specific processing, retention, deletion, and access arrangements are addressed through the applicable DPA and subprocessor register.

4. Rights and Confidentiality

4.1 Ownership

As between Provider and Customer, Customer retains its rights in Inputs and Customer Content.

To the extent Provider obtains any rights in an Output that are assignable, Provider assigns those rights to Customer. This assignment does not transfer rights belonging to a third party or give Customer rights in Provider’s underlying Models, software, platform, generic templates, or pre-existing materials.

AI-generated content may be similar to content generated for other users and may not qualify for intellectual-property protection. Nothing in this provision permits Provider to disclose another customer’s Confidential Information or Customer Content.

4.2 Confidentiality

Provider will protect nonpublic Customer Content in accordance with the confidentiality obligations in the Agreement and DPA.

Provider may disclose Customer Content only to personnel and permitted subprocessors who have a service-related need to know and are subject to appropriate confidentiality obligations; to recipients authorized by Customer; or where disclosure is required by applicable law.

The confidentiality exclusions in the Agreement apply only to the extent Provider can demonstrate that the relevant information qualifies for an applicable exclusion. Personal-data and data-protection obligations continue independently of any confidentiality exclusion.

4.3 Compelled Disclosure

Where legally permitted, Provider will notify Customer of a compulsory disclosure request before making the disclosure, reasonably assist Customer in seeking appropriate protection, and disclose only the information legally required.

Provider will not identify Customer, disclose Customer’s private research, or use Customer Content in marketing materials without separate authorization.

4.4 Third-Party Source Rights

Customer must have the rights and lawful authority necessary to submit Inputs and direct the uses it authorizes.

Provider’s inclusion of a citation, company logo, enrichment, filing, or source link does not grant Customer a license to redistribute third-party material.

Customer remains responsible for complying with applicable market-data, document, connector, confidentiality, intellectual-property, and publication restrictions.

5. Governance and Automated Actions

5.1 Authorization Boundaries

AI work is subject to authenticated user and workspace identity, applicable resource and nested-resource permissions, workspace and team policies, authorized providers and tools, and required approvals.

Selected skills, enriched chips, mentions, data blocks, uploaded text, retrieved pages, and tool responses may provide instructions or context within those boundaries. None independently grants access, changes permissions, or authorizes disclosure.

Provider will apply authorization controls at relevant retrieval, execution, and delivery boundaries in accordance with the functionality and safeguards of the Services.

These controls do not guarantee that every malicious instruction, prompt-injection attempt, inference error, or unauthorized instruction will be detected.

5.2 Saved Automations

An authorized User may configure a schedule, Pipeline Stage Action, or other supported automation to initiate work without a new prompt or confirmation for each occurrence.

Customer must review the trigger, execution identity, scope, destinations, tools, permissions, and applicable spend settings before enabling an automation.

A configured trigger does not bypass current permissions, privacy restrictions, or an approval required by the applicable action mode or workflow. Customer is responsible for keeping execution owners, credentials, permissions, and authorizations current.

5.3 Human Review

Customer is responsible for applying meaningful human review appropriate to the use and its consequences, particularly for investment conclusions, client communications, publication, and changes to external systems.

Draft creation is distinct from approval, publication, or execution. Provider will maintain approval gates made part of the applicable platform functionality. This Addendum does not authorize their removal or require a new approval for every otherwise permitted read or automated operation.

Customer’s review obligations do not excuse Provider from its own contractual obligations.

5.4 Cancellation, Retries, and External Effects

Stopping a run or revoking access may prevent further work at subsequent control boundaries, but may not recall a provider request already accepted, undo an external change already made, or eliminate a charge already incurred.

Some event deliveries may occur at least once. A delivery receipt does not establish that a recipient read a message or that downstream work completed.

Customer must review uncertain external outcomes before repeating work. Duplicate-prevention controls do not guarantee universal exactly-once execution.

6. Files and External Services

6.1 Ordinary File Storage

Uploading an ordinary attachment stores it within the file’s authorized user, workspace, resource, and nested-resource scope, as applicable.

Ordinary attachment storage does not automatically make a file shared workspace knowledge, select it for an AI run, submit it to a document processor, or incur AI document-processing charges.

Provider may perform storage-related validation and security processing without treating the file as an AI Input.

6.2 AI File Selection and Processing

Where supported and enabled, document processing requires selection of an eligible file, the necessary plan and workspace settings, and an authorized processing operation.

A shared Orchestration does not, by itself, expose its creator’s private files to other Users.

Processing may produce extracted text, page references, and other derivatives for authorized AI work. Format support for ordinary storage does not imply support for AI processing. Applicable file size, page, count, and other admission limits continue to apply.

6.3 Separate Copies and Retention

Ordinary file originals, AI-related projections, extracted content, conversation history, and generated resources may have distinct lifecycles.

Removing an AI selection or projection does not necessarily delete its ordinary original, previously extracted content, or independently saved generated artifacts.

Deleting a source does not automatically erase every independently retained result. Retained results remain subject to applicable confidentiality, authorization, retention, and deletion obligations.

6.4 Provider Copies and Customer-Selected Services

An explicit copy to a connected data source is a transfer to that destination and is not necessarily a continuously synchronized or bidirectional relationship.

Later renaming, replacement, or deletion of the platform original does not automatically change or delete a copy held by the destination service.

Customer is responsible for the destination it selects and the permissions it grants to that destination.

Customer-selected external services, including remote MCP tools and agents, may receive information necessary for an operation Customer authorizes. Those services may have their own terms, privacy practices, security measures, and retention policies.

The fact that a service is described as an integration does not determine or remove Provider’s legal responsibilities. Provider remains responsible for its own processing, access and disclosure controls, and obligations concerning its subprocessors.

7. Personal Data and Security

7.1 Processing of Customer Personal Data

To the extent Provider processes Customer Personal Data on Customer’s behalf through the AI Services, Provider acts as Customer’s processor or service provider, as applicable, and processes that data in accordance with the DPA, documented Customer instructions, this Addendum, and applicable law.

Customer determines its purposes, authorized data, lawful bases, notices, and instructions, subject to applicable law.

Account, billing, fraud-prevention, and other information Provider processes for its own independent purposes remains subject to the Agreement, Provider’s Privacy Policy, and applicable law. Describing information as Operational Metadata does not remove any applicable privacy obligations.

7.2 Data-Protection Restrictions

Provider will comply with the processor and service-provider obligations applicable to it under the DPA and applicable law.

Nothing in this Addendum authorizes Provider to sell Customer Personal Data, share it for cross-context behavioral advertising, or retain, use, disclose, or combine it outside the permitted relationship and purposes, except as permitted by applicable law and the Agreement.

The DPA governs applicable assistance, verification, cooperation, and other processor obligations concerning Customer Personal Data.

7.3 Security Measures

Provider’s technical and organizational security measures for the AI Services are those described in the DPA, including its applicable security schedule, as updated in accordance with the DPA and Agreement.

This Addendum does not create a separate or additional set of security certifications, technical controls, audit commitments, or recovery guarantees beyond those expressly stated in the DPA or another executed agreement.

Provider does not guarantee error-free security or immunity from prompt injection, malicious instructions, vulnerabilities, or other attacks.

7.4 Security Incidents

Provider’s incident-notification and cooperation obligations concerning Customer Personal Data are governed by the DPA.

If Provider becomes aware of a breach of Provider's security resulting in unauthorized access to, disclosure of, alteration of, or destruction of Customer Content processed through the AI Services, other than an incident involving Customer Personal Data governed by the DPA, Provider will notify Customer without undue delay after becoming aware of the incident, based on the information reasonably available at the time. Provider may provide an initial notification before the investigation is complete and supplement it as additional material information becomes available.

Provider will reasonably cooperate with Customer concerning containment, investigation, and remediation, subject to applicable law and the Agreement.

No separate universal 72-hour contractual notification deadline is created by this Addendum. Any non-waivable notification deadline imposed by applicable law remains controlling.

7.5 Assurance

Provider’s compliance information, audit cooperation, and other assurance obligations concerning the processing of Customer Personal Data are governed by the DPA.

Nothing in this Addendum represents that Provider holds a SOC 2 report, is certified to a particular standard, operates a regulated books-and-records archive, or satisfies a Customer’s industry-specific compliance obligations unless expressly documented in a separate written agreement.

7.6 Additional Data Terms

The DPA supplements the protections in this Addendum. Where a proposed use requires additional contractual terms, transfer safeguards, or other protections, the parties must put them in place before the relevant processing begins.

Nothing in this Addendum waives applicable privacy, data-protection, or other legal obligations.

8. Subprocessors and Processing Locations

The authoritative register of Provider's Subprocessors and other third-party providers authorized to receive Customer Content, including their relevant service purposes, roles, and processing information, is maintained in the Countercyclical Trust Center:

Countercyclical Trust Center

The DPA governs the authorization, notice, objection, and related procedures applicable to Subprocessors and other authorized Customer Content recipients.

8.1 Authorized Providers

Provider may engage third-party providers to support the AI Services, subject to the DPA, this Addendum, and the provider register maintained in Provider's Trust Center:

Countercyclical Trust Center

The register is the authoritative and definitive record of Provider's Subprocessors and other third-party providers authorized to receive Customer Content. It identifies the role and relevant service or purpose of each listed provider.

Only providers identified in the register as authorized to receive Customer Content may receive Customer Content in connection with the AI Services. A provider authorized to receive Customer Content does not necessarily constitute a Subprocessor under the DPA unless it Processes Customer Personal Data on Provider's behalf.

Provider will complete appropriate contractual and operational review before permitting a provider to receive Customer Content. Provider remains responsible for its obligations concerning Subprocessors as provided in the DPA and Agreement.

8.2 Provider Changes and Objections

Provider will give Customer at least fifteen (15) days' advance notice before adding a new provider authorized to receive Customer Content or materially changing an existing provider's authorized access to Customer Content, in accordance with the notice process in the DPA.

Customer may object within ten (10) days after receiving notice, provided the objection identifies documented data-protection or information-security grounds.

The parties will work in good faith to resolve a timely objection, including by considering reasonable alternative providers or other appropriate measures.

If the parties cannot resolve the objection, the remedies and procedures in the DPA apply, including, where applicable, refraining from the disputed processing or disclosure, implementing alternative measures, or discontinuing the affected feature with the applicable pro-rata refund.

The DPA governs urgent provider changes, notice exceptions, objection procedures, and related remedies.

8.3 Processing Locations

The configured processing regions for the AI Services’ infrastructure are in the United States, where regional configuration is available.

This statement concerns configured infrastructure processing regions. It does not represent that all Provider personnel, contractors, or authorized support personnel accessing or supporting the Services are physically located in the United States.

Processing locations, remote access, applicable international transfers, and related safeguards are further described in the DPA and the subprocessor register.

Provider will not represent that all processing, access, or support occurs exclusively in the United States unless that commitment is expressly stated in an applicable executed agreement.

8.4 Provider-Specific Processing and Retention

Provider-specific data handling, including applicable retention, deletion, operational access, safety or abuse-prevention processing, and onward subprocessors, is addressed through the DPA and the applicable subprocessor register.

A no-Training commitment does not constitute a zero-data-retention commitment. No blanket zero-retention representation is made by this Addendum.

Customer-configurable Model preferences or routing settings do not override the restrictions in the Agreement, DPA, this Addendum, or the applicable subprocessor register.

9. Research Use and Acceptable Use

9.1 Research and Analytical Limitations

The AI Services support investment research, due diligence, evidence analysis, drafting, and authorized workspace operations.

Outputs may be inaccurate, incomplete, stale, biased, unsupported, or incorrectly attributed. Customer must verify material facts, calculations, citations, dates, assumptions, and source rights before relying on or distributing an Output.

Source attribution, a completed status, or a generated citation is not a warranty of correctness, completeness, or exhaustive coverage.

9.2 No Professional or Discretionary Authority

Unless separately agreed in writing, Provider supplies research and workflow tools, not personalized investment, legal, tax, accounting, or other regulated professional advice.

Provider does not accept discretionary authority to trade securities, transfer assets, make binding investment decisions, or exercise authority over Customer’s accounts.

Customer remains responsible for its professional and fiduciary duties, supervision, client disclosures, conflicts, material nonpublic information controls, and required books and records.

The Services are not Customer’s designated regulatory books-and-records archive. Customer must preserve required records in its approved archive before applicable deletion deadlines. Platform history and audit views do not substitute for that archive.

No term waives a non-waivable statutory right or duty or determines Provider’s legal status solely by disclaimer.

9.3 Acceptable Use

Customer will not use the AI Services to:

  • Bypass security, permissions, approvals, metering, or provider safeguards;

  • Unlawfully disclose confidential, personal, or otherwise protected information;

  • Infringe intellectual-property or other third-party rights;

  • Impersonate another person or misrepresent the origin of information;

  • Fabricate evidence, citations, source material, or provenance;

  • Falsely represent AI-generated material as wholly human-created where disclosure is required by law; or

  • Conduct an activity prohibited by the Agreement or applicable law.

Customer must make disclosures required by applicable law.

Solely automated decisions producing legal or similarly significant effects on individuals, autonomous trading or asset transfers, and other specially regulated high-impact uses are outside the authorized scope unless expressly approved under an appropriate separate written agreement.

9.4 Restricted Information

Customer may submit ordinary business personal data and confidential investment information when lawfully authorized and within the agreed processing scope.

Unless Provider expressly agrees otherwise in writing and establishes the necessary safeguards, Customer must not submit protected health information requiring a business associate agreement, full payment-card security data, authentication secrets, sensitive government identifiers, or specially regulated biometric or similarly sensitive information.

Credentials must be supplied through supported secure connection mechanisms, not through ordinary prompts or documents.

These restrictions do not represent that the Services automatically identify or prevent every restricted submission.

These restrictions are in addition to, and do not limit or replace, the restrictions on sensitive, regulated, or otherwise restricted information set forth in Section 12.1 of the Terms of Service and Section 12 of the DPA, which continue to apply to Customer's use of the AI Services.

9.5 Geographic Availability

The initial release of the AI Services is not offered for use from the United Kingdom or European Union. Customer must not circumvent Provider’s applicable geographic access restrictions.

This availability restriction does not determine whether any particular processing activity is subject to a privacy or data-protection law.

10. Usage and Charges

AI work may consume credits for Models, research tools, document processing, and other paid operations disclosed in the applicable pricing or billing materials.

Charges depend on actual supported usage and applicable published or agreed pricing, not on a guarantee that a particular operation will produce a useful result.

Multi-step work and distinct paid retries may incur additional usage, including where a run is stopped or fails. The same recorded provider request or durable usage receipt will not be charged twice merely because it is replayed.

Available estimates and reservations are subject to their disclosed limits. Customer should configure appropriate budgets and review usage and uncertain outcomes before restarting work.

Exhausted credits may stop further paid work. Purchasing credits does not, by itself, authorize restarting an interrupted run.

This Addendum does not create or change any auto-recharge authorization, payment method, credit price, refund right, or subscription term. Those matters remain governed by the applicable billing agreement and Customer’s separately authorized settings.

11. Retention, Deletion, and Service Changes

11.1 General Retention and Deletion

Provider will handle the retention, return, export, and deletion of Customer Content processed through the AI Services in accordance with this Addendum, the DPA, the Agreement, and the applicable retention, return, export, and deletion mechanisms made available through the Services.

Access revocation, marking content as deleted, and physical erasure are distinct operations. A visible change in the application does not necessarily mean that every associated execution record, derived file, provider-side record, log, or backup has been physically erased.

Lawful preservation obligations and valid legal holds may delay deletion. Retained copies remain subject to applicable access restrictions, confidentiality obligations, and prohibitions on Training or unrelated use.

11.2 Agent Conversation and Execution Content Retention

Agent conversation content and associated execution content are retained in accordance with the retention policy configured for the applicable Customer workspace.

The default retention period is ninety (90) days. Workspace administrators may configure retention periods ranging from seven (7) days to three thousand six hundred fifty (3,650) days, or select indefinite retention, subject to the functionality made available in the Services.

Provider operates an automated daily cleanup process designed to remove Agent conversation and execution content that has reached the end of its applicable configured retention period.

Deletion of an Agent conversation is distinct from deletion of its underlying execution content. When Customer or an authorized User deletes a conversation, the conversation is marked as deleted and removed from ordinary conversation access. Associated execution content is not independently purged solely because the conversation was marked deleted. Instead, that execution content remains subject to the applicable workspace Agents retention policy.

Where a workspace is configured for indefinite retention, the daily retention cleanup process does not automatically purge execution content solely because its associated conversation has been deleted.

Customer is responsible for selecting and maintaining an appropriate workspace retention configuration. Customer must not treat deletion of a conversation from the Services as confirmation that all associated execution content has been immediately or independently erased.

Other AI-related data, including generated resources, intermediate files, provider-side records, logs, and backups, may be subject to separate retention and deletion mechanisms as described in this Addendum, the DPA, and the applicable subprocessor terms.

11.3 Other Data and Copies

Ordinary workspace files, generated resources, independent resource copies, provider-side records, operational logs, security records, billing records, and backups may have separate retention and deletion lifecycles.

Deletion of one category does not necessarily delete another category or a copy held by a separately authorized destination.

Provider’s DPA describes applicable active-system deletion, attachment cleanup, backup, and residual-record practices. Provider does not promise immediate or synchronous deletion of every copy across every system or subprocessor.

11.4 Service Changes

Provider may restrict or modify an AI feature to address security, legal, provider-availability, or service-integrity concerns, subject to the Agreement and applicable law.

A restriction or discontinuation does not remove existing confidentiality, data-protection, deletion, or billing-reconciliation obligations.

A material expansion of permitted data use, Training rights, or disclosure purposes requires express agreement. Continued use of the Services or a routine documentation update does not grant those rights.

Other material changes will be handled through the amendment and acceptance procedures in the Agreement and applicable law.

12. Warranties, Liability, and Contact

12.1 Agreement Terms Apply

Except for express obligations in this Addendum, applicable law, or another executed agreement, the warranties, remedies, indemnities, exclusions, and limitations of liability in the Agreement apply to this Addendum and the AI Services.

This Addendum does not create an additional AI-output infringement indemnity, a guarantee that Outputs will be exclusive or protectable, or a separate liability cap.

An AI accuracy or risk disclaimer does not cancel Provider’s express confidentiality, security, no-Training, or data-processing obligations under the Agreement and DPA.

Any negotiated AI-specific indemnity or liability treatment must be documented expressly in a separately executed agreement.

Nothing excludes responsibility that cannot lawfully be excluded.

12.2 Contact

Questions, legal notices, data-protection requests, and incident reports concerning this Addendum may be sent to:

Privacy: privacy@countercyclical.io
Security: security@countercyclical.io

Formal notices remain subject to the applicable notice procedures in the Agreement and DPA.


Schedule 1 — AI Processing and Retention Particulars

1. Purpose and Nature of Processing

The AI Services process Customer Content to provide authorized workspace functionality, including retrieval of selected context, requested Model inference, research and tool operations, creation and delivery of authorized Outputs, and the operation, security, maintenance, troubleshooting, support, auditing, and accounting of those services.

Processing may include collection, storage, organization, retrieval, transformation, transmission to authorized subprocessors or destinations, and deletion.

Training is excluded.

2. Categories of Individuals

Depending on Customer’s use, processing may concern:

  • Customer’s Users and authorized personnel;

  • Business contacts;

  • Customers and clients;

  • Investors and counterparties; and

  • Other individuals referenced in Customer’s authorized workspace records or source materials.

3. Categories of Information

Depending on Customer’s use, information may include:

  • Names and business contact details;

  • Workspace membership, user identifiers, and permissions;

  • Research, investment, and due diligence records;

  • Confidential business and investment information;

  • Supported files and source materials;

  • Prompts, instructions, and conversation content;

  • Inputs and Outputs; and

  • Service and usage metadata.

Restricted information described in Section 9.4 is excluded unless separately authorized in writing and supported by appropriate safeguards.

4. Processing Duration

Processing continues during the applicable service period and any applicable retention, deletion, or lawful preservation period described in this Addendum and the DPA.

5. Agent Conversation and Execution Content

Item

Default retention

Workspace-configurable retention

Automated cleanup

Explicit conversation deletion

Indefinite retention

Other associated data

Retention and deletion rule

Ninety (90) days.

Seven (7) to three thousand six hundred fifty (3,650) days, or indefinite retention, subject to available functionality.

An automated daily cleanup process is designed to remove Agent conversation and execution content that has reached the end of the applicable configured retention period.

Marks the conversation as deleted and removes it from ordinary conversation access. Associated execution content remains subject to the applicable Agents retention policy and is not independently purged solely because the conversation was deleted.

The daily retention cleanup process does not automatically purge execution content solely because its associated conversation was deleted.

Generated resources, intermediate files, provider-side records, logs, and backups may follow separate retention and deletion mechanisms.

6. Other AI-Related Data

Other categories of AI-related data may include:

  • Ordinary workspace file originals and attachments;

  • Extracted content and file derivatives;

  • Generated native resources and independently saved copies;

  • Provider requests, responses, caches, and safety or abuse-prevention records;

  • Operational, audit, security, and billing records;

  • Backup and disaster-recovery copies; and

  • Copies transferred to Customer-selected external services.

These categories may have separate retention and deletion mechanisms. The applicable DPA, subprocessor register, provider terms, and destination-service terms describe the relevant practices and limitations.

7. Processing Locations and Safeguards

The configured processing regions for AI Services infrastructure are in the United States where regional configuration is available. Remote personnel access may occur from other countries, as described in the DPA.

Subprocessor-specific locations, onward processing, and applicable transfer safeguards are identified in the DPA and the authoritative subprocessor register.

The technical and organizational safeguards applicable to the AI Services are those described in the DPA and its security schedule.

8. Subprocessors

The authoritative list of Provider subprocessors, including relevant service purposes and processing information, is maintained in the Countercyclical Trust Center:

https://security.countercyclical.io?itemUid=e3fae2ca-94a9-416b-b577-5c90e382df57

The DPA governs subprocessor authorization, notice, objections, and related procedures.

Learn More from AI

Manage Account

© Countercyclical 2026 · All Rights Reserved

Your Privacy Choices

Countercyclical

Investment research, reimagined.

Learn More from AI

© Countercyclical 2026 · All Rights Reserved

Your Privacy Choices