Terms of Use

Last verified: 2026-02-13 | Status: canonical

Last updated: February 13, 2026 Effective date: February 11, 2026

These Terms of Use ("Terms") are a binding agreement between you and a21e, Inc., or the a21e affiliate identified in an applicable order form ("a21e," "Company," "we," "us," or "our"). These Terms govern your access to and use of our websites, APIs, software, models, tools, integrations, and related services, including creator and enterprise offerings (collectively, the "Service").

By accessing or using the Service, creating an account, generating an API key, or clicking to accept, you agree to these Terms. If you do not agree, do not use the Service.

1. Definitions

"Account" means your user account for the Service.
"API" means our application programming interfaces and compatibility endpoints.
"Content" means data, prompts, text, files, media, and other materials submitted to or generated through the Service.
"Creator Content" means prompt or related content submitted by creators.
"Customer Content" means Content submitted by you or your users, including instructions, intents, uploaded files, and workflow data.
"Credits" means prepaid usage units where one credit equals one enhancement (a single prompt, skill, or artifact within an execution run). More complex tasks that involve multiple enhancements consume multiple credits.
"Order Form" means a mutually executed commercial order, SOW, or subscription document.
"Organization" means a workspace/team account with multiple users.
"Output" means responses or artifacts generated by the Service from your requests.

2. Eligibility and Authority

You represent that:

you are at least 18 years old and legally capable of entering a contract; and
if acting for a company or organization, you have authority to bind that entity.

If you are accepting on behalf of an entity, "you" means that entity.

3. The Service

3.1 Product Surface

The Service may include one or more of the following:

prompt discovery, catalog, and API delivery;
managed execution and bring-your-own-key (BYOK) execution modes;
workspace configuration including favorites, preferences, and project-level settings;
agentic tooling such as prompt optimization, website generation, code review, and IDE workflows;
creator onboarding, attribution, and payout-related workflows;
team, organization, and admin controls;
enterprise and agency-style offerings.

3.2 Evolving Features

We may add, modify, or discontinue features. Some features are beta, preview, or experimental and may be less stable.

3.3 Webhooks

You may register webhook endpoints to receive real-time event notifications (e.g., execution completion, review consensus, credit alerts). Webhook deliveries are best-effort with automatic retries (up to 5 attempts with exponential backoff). We do not guarantee delivery timing or order. Endpoints must respond within 10 seconds; non-2xx responses trigger retries. Webhook payloads are signed with HMAC-SHA256 using a per-endpoint secret; you are responsible for verifying signatures. We may throttle or disable endpoints that consistently fail.

3.4 Third-Party Dependencies

Parts of the Service rely on third-party services (for example identity, payments, cloud infrastructure, and model providers). Availability may depend on those providers.

4. Accounts, Security, and Organization Administration

You must:

provide accurate registration information;
safeguard credentials, API keys, and access tokens;
promptly notify us of unauthorized access; and
ensure your users comply with these Terms.

You are responsible for all activity under your Account and keys, including activity by Organization members and service accounts.

Organization admins may manage seats, billing, data visibility, and access controls within the Organization.

5. Acceptable Use

You will not, and will not permit others to:

violate law, regulation, or third-party rights;
use the Service to create, facilitate, or promote fraud, abuse, malware, spam, harassment, or unlawful surveillance;
attempt to bypass technical controls, security controls, or usage limits;
reverse engineer or extract underlying model prompts, system instructions, trade secrets, or non-public service internals except as expressly allowed by law;
use automated scraping or bulk extraction beyond documented APIs and permitted limits;
use the Service to build or operate a competing service that substantially replicates core protected functionality using non-public outputs or extracted internals;
submit content you do not have rights to process.

You must maintain reasonable human oversight for high-impact use cases (for example legal, medical, safety, hiring, credit, or other regulated decisions).

6. Customer Content and Output

6.1 Your Rights

As between you and a21e, you retain your rights in Customer Content, subject to the rights you grant below.

6.2 License to a21e

You grant a21e a worldwide, non-exclusive, royalty-free license to host, store, reproduce, modify (for formatting, safety, and technical operations), transmit, and otherwise process Customer Content to:

provide, maintain, secure, and improve the Service;
perform your requested workflows and deliver Output;
enforce policies and legal obligations;
provide support and troubleshooting.

6.3 Output

Subject to your compliance with these Terms and applicable law, we assign to you our rights, if any, in Output generated solely for your request, excluding:

Service IP;
Creator Content rights retained by creators;
third-party materials and open-source components under their own licenses; and
feedback, telemetry, and aggregated or de-identified analytics.

You are responsible for evaluating Output before use.

7. Creator Terms and Marketplace Terms

If you submit Creator Content:

you represent you have all rights needed to submit and license it;
you grant a21e the license necessary to host, deliver, process, evaluate, improve, and display metadata for the Creator Content;
creator economics, attribution, payouts, and eligibility are governed by the creator program terms, dashboard rules, or order terms then in effect;
we may moderate, suspend, or remove Creator Content for policy, legal, quality, trust, or security reasons.

Creator Content may be made available in different delivery modes (including direct content delivery and proxy-only execution) depending on product settings and policy.

7A. Auto-Improvement Program

The Service includes an optional auto-improvement program that allows the platform to generate and evaluate improved versions of your Creator Content. Participation is governed by the following terms:

Opt-in required. Auto-improvement is never automatic. You must explicitly enable it on a per-prompt basis. No candidate versions will be generated unless you opt in.
How it works. When enabled, the platform may generate candidate versions of your prompt, run canary evaluations (routing a small percentage of live traffic to the candidate), and promote the candidate to stable if quality metrics improve.
Revenue share. Versions improved by the platform may be subject to a different revenue share rate than your standard creator share. Your original human-authored versions remain at the standard rate. Specific percentages are governed by the creator program terms then in effect.
Creator control — opt out at any time. You may disable auto-improvement at any time. When you opt out, any running evaluations are aborted and your prompt reverts to the last human-authored stable version.
No content exposure. Prompt content is processed internally for improvement purposes only. It is never shared with third parties or returned in API responses beyond normal delivery.
Versioned and reversible. All changes are versioned. Stable versions are never modified in place. You can review, compare, and revert any auto-improved version from your dashboard.

8. Fees, Credits, Billing, and Taxes

8.1 Paid Services

Paid features may include subscriptions, prepaid credits, usage-based fees, and enterprise contracts.

8.2 Credits

Credits are consumable usage units where one credit equals one enhancement (a single prompt, skill, or artifact within an execution run). Credits are not legal tender.
Credits are non-refundable except as required by law or expressly stated in writing.
We may offer low-cost starter/test credit options and promotional credits under separate terms.
We may set expiration windows for promotional credits.
Certain Service features (such as website generation, code review, and agent execution) may consume multiple credits per use based on task complexity, as displayed to you before execution.
BYOK execution requires credits for a21e orchestration and platform intelligence. BYOK does not waive credit requirements.

8.3 Subscription Renewal

Recurring plans renew automatically until cancelled. You authorize us (or our payment processor) to charge applicable recurring fees.

8.3A Plan Changes and Credit Reset

When you upgrade or downgrade plans, your current-period balance remains available until the billing period ends. At renewal, your balance resets to the credits included with the renewed plan. Unused subscription credits do not roll over into the next period. Add-on credit packs also expire at the end of the active billing period unless explicitly stated otherwise in writing.

8.3B Subscription State

We track your subscription state (plan tier, billing period, and status) to provide plan-aware messaging, upgrade recommendations, and accurate billing information. Subscription state is updated automatically when you purchase, upgrade, renew, cancel, or encounter a payment issue. You can view your current plan status on the billing page.

8.4 Price Changes

We may change pricing with prior notice. Changes generally apply prospectively. Per-execution credit costs for certain features may vary based on task complexity and are displayed before each execution.

8.5 Taxes

Fees are exclusive of taxes unless stated otherwise. You are responsible for applicable taxes other than our net income taxes.

9. BYOK and Provider Terms

If you use BYOK mode:

you authorize us to transmit your requests to the provider you select;
you represent your use complies with that provider's terms and applicable law;
you are responsible for provider-side charges and account obligations;
unless explicitly stated otherwise, we do not persist raw BYOK secrets beyond transient processing needed to service your request.

9.1 BYOK Credit Enforcement

BYOK still requires credits for every execution run:

Provider costs and platform credits are separate. You pay your provider directly for model token usage and you consume a21e credits for platform execution.
Insufficient credits block execution. If your credit balance is insufficient, requests fail with an insufficient-credits error until you add credits or renew.
Applies to all run types. Single-prompt runs, streamed runs, and package executions all require credits.

9A. Provider Key Vault

The Service includes a Provider Key Vault that allows you to store and manage your LLM provider API keys for reuse across sessions. The following terms apply:

Encryption at rest. Provider keys are encrypted using AES-256-GCM before storage. Raw key values are never logged, displayed after initial submission, or returned in API responses.
Transient decryption. Keys are decrypted only in transient server memory during prompt execution and are not persisted in plaintext at any point.
Key hint only. After saving, only a masked hint (e.g. sk-...xyz) is visible. The full key cannot be retrieved.
User controls. You can add, label, set defaults, test, and delete your saved keys at any time through the API or product interface.
Your responsibility. You are responsible for the validity, permissions, and billing obligations associated with your provider keys. Saving a key in the vault does not transfer provider-side obligations to a21e.
Deletion. When you delete a saved key or your Account, the encrypted key data is permanently removed in accordance with our standard data retention practices.

9B. Skills Marketplace

The Service includes a Skills ecosystem that allows reusable, composable capabilities ("Skills") to be registered, installed, and invoked. The following terms apply to your use of Skills:

Permission model. Skills declare the permissions they require (such as file access, network requests, email sending, data export, secret access, or LLM calls). You are shown the full list of declared permissions at install time and must approve them before the Skill can be invoked on your behalf. High-risk permissions (file write, email send, data export, secret access) require explicit confirmation before each first invocation.
Sandbox and resource limits. Skill execution is subject to resource limits including execution timeouts, network domain allowlists, and scoped file system access. Skills execute in isolated contexts with no shared mutable state between them. These boundaries are enforced by the runtime and cannot be overridden by Skill code.
Immutability and signing. Once a Skill version is published, it is immutable — the package contents cannot be changed. Published Skills are cryptographically signed (HMAC-SHA256) for integrity verification. The runtime verifies the signature before execution.
Security scanning. All Skills submitted for publication undergo automated security scanning, including secret detection (API keys, tokens, credentials), permission validation, and prompt injection pattern analysis. Skills that fail security checks are rejected and cannot be published.
Audit logging. All Skill invocations are logged for audit purposes, including the Skill identifier, version, permissions used, execution duration, and outcome status. These logs are available to workspace administrators and may be used for security review and compliance purposes.
Prompt injection mitigations. The Service enforces instruction/data separation for Skill-injected content. Skill-provided data is clearly delimited from system instructions and cannot override system prompts. High-impact actions triggered by Skills require explicit user confirmation even when the Skill has the declared permission.
No warranty on third-party Skills. Skills published by third parties (when available) are provided as-is. We review Skills for security compliance but do not guarantee their accuracy, reliability, or fitness for any particular purpose. You are responsible for evaluating Skills before installation and use.
Removal and suspension. We may remove, suspend, or deprecate any Skill for policy, legal, quality, trust, or security reasons at any time.

9C. Memory Layer

The Service includes a memory layer that stores contextual information to improve AI prompt execution. The following terms apply to your use of memory features:

Server-side storage. Memory data is stored server-side and associated with your Account. Memories persist across sessions and are used to augment AI prompt execution with relevant context, reducing the need to restate project decisions, constraints, and preferences.
User controls. You can create, view, update, and delete your memories at any time through the API or product interface. You can also mute, archive, or pin individual memories to control how they influence future AI interactions.
Prompt augmentation. When you execute prompts, the Service may retrieve relevant memories and inject them as additional context to improve output quality. Memory-influenced runs include an explanation payload indicating which memories were used and why.
Sensitivity classification. The Service automatically classifies memory content for sensitivity. Auto-capture is blocked for content classified as security credentials, legal privileged material, or personally identifiable information (PII). Blocked capture attempts are logged with reason codes. You remain responsible for not manually storing content that violates applicable law or these Terms.
Quality feedback. You can provide feedback on memory quality, including positive, negative, irrelevant, and outdated signals. This feedback is used to tune retrieval relevance and may affect memory confidence scores over time.
Organization-scoped memories. If you use the Service within an Organization, certain memories may be scoped to the Organization and shared with other Organization members. Organization administrators can set policies that govern memory retrieval and usage within the Organization.
Hard-enforced policies. Organization administrators and the Service may designate certain memories or policies as hard-enforced. Hard-enforced memories cannot be overridden or bypassed by individual users. Soft-guidance memories may be overridden with an explicit logged reason.
Data retention. Memory data follows the same data retention and deletion policies as other user data under these Terms and our Privacy Policy. When you delete your Account, memory data associated with your Account is deleted in accordance with our standard data retention practices.

9D. StylePrint

The Service includes a StylePrint feature that extracts abstract visual design properties from public URLs. The following terms apply:

What is extracted. StylePrint extracts abstract visual properties only: colors, font families, spacing scales, border radii, shadow definitions, and animation timings. These are delivered as CSS custom properties and structured JSON.
What is NOT extracted. StylePrint does not extract logos, icons, images, copy text, layout code, or unique UI components. Commercial fonts are detected and flagged with a license recommendation; the system suggests system-ui fallbacks when commercial fonts are identified.
Attribution. The source URL is stored and attributed in the extraction record (e.g., "Inspired by [url]").
Legal disclaimer. Every extraction includes a disclaimer in the database record and API response. Design tokens are abstract properties and do not constitute copying of a website's protected expression.
User responsibility. You are responsible for verifying that extracted design tokens do not infringe third-party intellectual property rights before using them in production. a21e does not warrant that extracted tokens are free of third-party rights.
Credits. Each StylePrint extraction consumes 5 credits.

9E. Multi-Model Deliberation and Self-Improvement

The Service includes multi-model deliberation and self-improvement capabilities that may be used as part of agent workflows:

Multi-model deliberation. The Service may route your requests to multiple AI models simultaneously for plan generation, critique, and consensus-building. Each model call consumes credits according to the model tier multiplier. You may configure the panel of models and number of critique rounds.
Execution of deliberated plans. Winning plans selected through the consensus process may be executed automatically (creating code changes, branches, or pull requests) only after explicit approval. Human approval gates are enabled by default.
Self-assessment and improvement. Agent domains may periodically assess their own performance, acquire new knowledge from external sources, and propose behavioral improvements. Significant improvements (above a configurable threshold) require human review before activation.
Knowledge persistence. Knowledge acquired through self-improvement is stored within the platform and scoped to your organization. Knowledge does not transfer across organizations.
No autonomous self-modification. Agents do not modify their own core code or system prompts without human review gates. All proposed improvements are versioned, reversible, and auditable.
Credit consumption. Deliberation sessions and improvement cycles consume credits based on the number of model calls made. Economy-tier models are used for internal operations (assessment, reflection) where possible to minimize cost.

10. Privacy and Data Processing

Our Privacy Policy explains how we collect, use, disclose, retain, and protect personal data. If there is a conflict between these Terms and an executed Data Processing Addendum (DPA), the DPA controls for covered processing.

11. Confidentiality

You may receive non-public information about the Service. You will not disclose or misuse it except as needed to use the Service under these Terms.

If you and a21e have a separate NDA, that NDA controls for covered confidential information.

12. Intellectual Property

The Service, including software, architecture, branding, workflows, and non-public methods, is owned by a21e or its licensors and protected by law.

No rights are granted except the limited rights expressly stated in these Terms.

13. Feedback

If you provide suggestions, comments, or feedback, you grant us a worldwide, perpetual, irrevocable, sublicensable, royalty-free license to use it for any lawful purpose without compensation.

14. Suspension and Termination

We may suspend or terminate access immediately if reasonably necessary for security, abuse prevention, legal compliance, non-payment, or material breach.

You may stop using the Service at any time. Termination does not relieve payment obligations accrued before termination.

Sections that by nature should survive will survive, including payment obligations, IP, disclaimers, limitation of liability, indemnity, and dispute provisions.

15. Disclaimers

THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE." TO THE MAXIMUM EXTENT PERMITTED BY LAW, WE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE, AND NON-INFRINGEMENT.

AI/agent outputs may be incorrect, incomplete, biased, or unsuitable for your use case. You are responsible for review and final decisions.

16. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY LAW:

WE WILL NOT BE LIABLE FOR INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, EXEMPLARY, OR PUNITIVE DAMAGES, OR LOSS OF PROFITS, REVENUE, DATA, OR GOODWILL.
OUR AGGREGATE LIABILITY FOR ALL CLAIMS ARISING OUT OF OR RELATED TO THE SERVICE WILL NOT EXCEED THE GREATER OF:
- THE AMOUNT PAID BY YOU TO a21e FOR THE SERVICE IN THE 12 MONTHS BEFORE THE CLAIM; OR
- USD $100.

These limits apply regardless of legal theory and even if a remedy fails of its essential purpose.

17. Indemnification

You will defend, indemnify, and hold harmless a21e and its affiliates, officers, directors, employees, and agents from claims, damages, losses, liabilities, costs, and expenses (including reasonable attorneys' fees) arising from:

your use of the Service;
your Customer Content or Creator Content;
your violation of these Terms or applicable law; or
your infringement or misappropriation of rights of another.

18. Export Controls and Sanctions

You may not use the Service in violation of export control, sanctions, or trade laws. You represent you are not prohibited from using the Service under applicable sanctions laws.

19. Changes to Terms

We may update these Terms. For material changes, we will provide reasonable notice (for example via email or in-product notice). Continued use after the effective date means you accept the updated Terms.

20. Governing Law and Dispute Resolution

Unless otherwise stated in an Order Form:

governing law is the laws of the State of Delaware, excluding conflict-of-law rules;
disputes will be resolved in the state or federal courts located in Delaware, unless arbitration below applies.

For U.S. customers where permitted by law, either party may elect binding arbitration on an individual basis, and each party waives class action participation. Either party may seek injunctive relief in court for intellectual property misuse, confidentiality breaches, or security abuse.

21. General Terms

Entire agreement: These Terms, Privacy Policy, and any applicable Order Form/DPA are the complete agreement for the Service.
Severability: If any provision is unenforceable, the remaining provisions remain in effect.
No waiver: Failure to enforce a provision is not a waiver.
Assignment: You may not assign these Terms without consent. We may assign in connection with merger, acquisition, financing, or asset transfer.
Force majeure: Neither party is liable for delays caused by events beyond reasonable control.

22. Contact

Legal: legal@a21e.com
Privacy: privacy@a21e.com
Support: support@a21e.com