FluidGrids — AI Product Terms

Document ID: FG-LEGAL-012 Title: AI Product Terms / Data Training Terms Version: 1.0 Effective date: 5 August 2026 Last updated: 2026-08-05 Publisher: Algoshred Technologies Private Limited (Burdenoff Group) Canonical location: https://fluidgrids.com/ai-terms Status: Approved — India jurisdiction, 2026-08-05.

---

Summary

• What this covers. Any AI feature in the FluidGrids Service — the AI Agent node, NL→workflow generation, AI-generated nodes, AI-assisted suggestions in the canvas, AI-drafted node configuration, and AI-summarised logs.
• Assistive, not autonomous. FluidGrids' AI supports your decisions about workflows; it does not make them. Every run executes only when a human (or an external trigger you wired up — webhook, cron, form) explicitly starts it.
• Provider list. The default AI provider stack is OpenAI and Anthropic. The current enabled list lives at https://fluidgrids.com/subprocessors; material additions are notified 30 days in advance.
• No training on your data. Algoshred does not use your prompts, completions, workflow definitions, or run inputs/outputs to train, fine-tune, or evaluate any generally available AI model. This applies whether the AI is Algoshred-operated or customer-selected (BYOK).
• Opt-out. AI features are off by default at the workspace level. A workspace owner can disable AI entirely from workspace settings; members can simply choose not to use AI features.
• Human in the loop. You must keep a human in the loop for any decision that produces a legal or similarly significant effect on an individual (hiring, credit, healthcare, etc.).
• No warranty on outputs. AI outputs may be inaccurate, incomplete or biased and must be reviewed by a competent human before use.
• Governing law. Indian law, per §16 of the Terms of Service.

These AI Terms are incorporated into the Terms of Service (FG-LEGAL-PUB-002) and the MSA (for Enterprise customers). In case of conflict on AI-specific matters, these Terms prevail.

---

1. Overview of AI features

1.1 The Service includes AI-assisted capabilities to:

• generate draft workflow definitions from a natural-language description ("NL→workflow");

• suggest the next node, branch condition, or expression while you build a workflow ("AI Agent node" and inline suggestions);

• generate draft content for node configuration (prompt templates, regex, JSON shapes, code in code_tool nodes);

• summarise run logs, errors, and outputs for human review;

• surface workflow-analytics insights.

  1.2 These features are decision-support tools. They produce suggestions, drafts, summaries and explanations that a human must review before any production change to a workflow that runs in your environment.

  1.3 FluidGrids does not ship autonomous agents that execute real-world side effects (financial transactions, communications, infrastructure changes) without a human-triggered run. Every run executes only when a human (or an external system they wired up — webhook, cron, form) explicitly starts it.

2. How AI is provided

2.1 AI features are provided as an embedded capability of the Service. Algoshred selects and operates the underlying AI provider(s) and infrastructure; the AI provider acts as Algoshred's Subprocessor and is listed in the DPA's subprocessor annex (https://fluidgrids.com/subprocessors).

2.2 The default AI provider stack is OpenAI and Anthropic. Where an Enterprise Order Form expressly enables a "bring-your-own-key" (BYOK) or customer-selected-provider option, the chosen provider's terms additionally apply to that leg, and Customer is responsible for that provider relationship; absent such an option, Algoshred operates the AI provider relationship.

2.3 Additional AI providers (e.g., Google Gemini, Cohere, open-source / self-hosted models) may be added over time. The list of enabled providers is published at https://fluidgrids.com/subprocessors; material additions are notified per the notice clause in §14.

2.4 AI-assist usage may be metered per the FluidGrids Pricing Guide (FG-PRICING-001).

3. Customer Data and AI

3.1 Inputs to AI features (natural-language prompts, workflow snippets you attach, run excerpts you ask the AI to summarise, code you ask the AI to draft) and AI outputs (generated workflows, node configurations, summaries, recommendations) are Customer Data and are governed by the Terms of Service and the DPA.

3.2 Algoshred will not use prompts, completions, workflow definitions, run inputs/outputs, or any Customer-side AI content to train, fine-tune, evaluate, or otherwise improve any generally available machine-learning or large-language model. This applies whether the AI is Algoshred-operated or, where enabled, customer-selected (BYOK).

3.3 Customer represents that it has the right to submit the inputs to the AI features, including any third-party materials, code, or data subject to confidentiality or third-party agreements, and that it has the lawful basis and any notices/consents required to process Personal Data through AI features.

3.4 Customer should minimise personal data in free-text prompts and must not submit special-category data to AI features.

4. Human-in-the-loop and prohibited automated decisions

4.1 Customer shall keep a human in the loop for any decision that produces a legal or similarly significant effect on an individual — including hiring, compensation, credit, healthcare, or any decision governed by DPDP Act / GDPR Article 22 and applicable law.

4.2 Customer shall not use FluidGrids AI features to make such decisions on a solely-automated basis without meaningful human review and without the consent, notice and rights the law requires.

4.3 Customer shall not use AI features for unlawful discrimination, mass surveillance, social scoring, or to evade labour, employment, financial-services or data-protection law.

5. Deployer obligations: bias, fairness, transparency

5.1 Where applicable law treats an AI-assisted workflow as a high-risk AI system or an automated decision tool — for example the EU AI Act and various US state AI laws — Customer is the deployer and is responsible for:

• conducting or commissioning any required bias / disparate-impact audit;

• providing any required notice to data subjects / end users that AI is used and how;

• honouring any opt-out, human-review or explanation rights the law grants individuals;

• maintaining records of its AI use.

  5.2 Algoshred will provide reasonable information about how the relevant AI features function to support Customer's deployer obligations, but Algoshred does not assume the deployer's legal duties.

  5.3 AI outputs may be inaccurate, incomplete or biased and must be reviewed by a competent human before use, particularly when used downstream in regulated decisions.

6. AI opt-out

6.1 AI features are off by default for workspaces that have not enabled them. A workspace owner (or an Administrator with the appropriate RBAC role) can disable AI features entirely from workspace settings; disabling turns off NL→workflow generation, the AI Agent node, AI-generated nodes, and AI-summarised logs.

6.2 Where AI features are enabled at the workspace level, individual members can still choose not to use them (no AI suggestions inserted, no prompts sent). Members may also contact their workspace owner to request a workspace-level disable.

6.3 If you prefer an AI-on-by-default posture with no opt-out, contact [email protected] to discuss an Enterprise Order Form that opts in for the entire tenant.

7. Logging and retention

7.1 By default, AI metadata (timestamps, feature, model, token counts, success/failure, IDs) is logged; AI input/output content is stored per Customer's tenant configuration (default: 30 days; configurable down to "do not retain prompt/completion bodies").

7.2 Customer may configure shorter retention via account settings, or request deletion of specific AI content under the DPA.

7.3 For Enterprise Customers, retention windows are stated on the Order Form.

8. Use restrictions specific to AI

In addition to the Acceptable Use Policy (https://fluidgrids.com/aup), Customer shall not (and shall not allow Authorized Users to):

• use AI features to make automated decisions with legal or similarly significant effect without human review and required consents/notices (see §4);
• use AI features to generate content that violates law, infringes third-party rights, or is discriminatory, harassing or deceptive;
• use AI features for biometric identification, mass surveillance, or social scoring of individuals;
• generate deepfakes of identifiable individuals without consent, or content designed to defraud or impersonate;
• attempt to extract Algoshred's system prompts, internal model orchestration logic, or guardrails to evade safety controls or build a competing service;
• rely on AI outputs without verifying them where accuracy matters (production deployments, regulated decisions, compliance).

9. No warranty for AI outputs

9.1 AI outputs may be inaccurate, incomplete, biased or offensive. Algoshred makes no warranty as to accuracy, reliability, fairness, originality, or fitness of AI outputs for any purpose, including any production workflow decision.

9.2 Customer is solely responsible for reviewing and using AI outputs and for the actions its workflows take.

9.3 AI outputs may be similar to outputs received by other users; Customer does not have exclusivity over AI-generated content.

9.4 Where Customer publishes AI outputs (e.g., AI-drafted node configurations shared with end users), Customer is responsible for compliance with transparency, labelling and disclosure obligations imposed by law (e.g., EU AI Act transparency requirements, MeitY / IT Rules disclosures applicable in India).

10. Intellectual property in AI outputs

10.1 As between Customer and Algoshred (and subject to any enabled provider's terms), Customer owns the AI outputs generated through its use of the Service, to the extent rights can be granted, and accepts the risk that AI-generated content may not be protected by copyright (uncertain in India / US as of the date of these Terms).

10.2 Algoshred makes no representation or warranty that AI outputs are non-infringing; the IP indemnity in the Terms of Service / MSA does not cover AI outputs.

11. Eligibility

AI features are used by Customer's Authorized Users in a business workflow context. FluidGrids is a 13+ platform (see Privacy Policy §13); Customer shall not configure AI features to evaluate individuals in a manner prohibited by applicable child-labour or data-protection law.

12. Personal data in prompts

If Customer's inputs contain Personal Data (e.g., names, emails, identifiers in run logs), processing remains governed by the DPA. Customer should minimise Personal Data in prompts and must not submit special-category data.

13. Suspension

Algoshred may suspend AI features for a Customer or account where suspected abuse, unlawful automated decision-making, or violation of the AUP is detected, and will notify Customer.

14. Updates

Algoshred may update these AI Terms with at least 30 days' notice for material changes (provider list additions, opt-out mechanism changes, retention changes). Continued use after the effective date is acceptance. For Enterprise Customers, in-force Order Forms preserve material terms until renewal.

15. Governing law

These AI Terms are governed by Indian law, per §16 of the Terms of Service (or §15 of the MSA for Enterprise Customers).

---

16. Contact

For questions about these AI Terms, contact [email protected].

Algoshred Technologies Pvt Ltd
No.43, Veeramani Nagar, 2nd Cross Street,
Nanmangalam, Chennai - 600117, Tamil Nadu, India
Email: [email protected]

---

Document ID: FG-LEGAL-012 Owner: Legal / Algoshred Review cycle: every 12 months, on material change, or when adding a new AI provider.