StartStep AI Features and Responsible Use Notice

Version: V1.16 Effective: This version takes effect on the date it is published in the Rules Centre.

Effective Date: This version takes effect on the date it is published in the Rules Centre.
Operator: Noviq Pte. Ltd.
UEN: 202607454K
Registered Office: 68 Circular Road, #02-01, Singapore 049422
Official Domain: startstep.ai
Customer Support: support@startstep.ai
AI Safety and Content Reports: support@startstep.ai

Important Notice

This AI Features and Responsible Use Notice (the “AI Notice”) forms part of the StartStep AI Rules and applies whenever you access an artificial-intelligence, machine-learning, automation, search, file-analysis, workflow, Agent or AI-assisted support feature supplied through StartStep AI.

AI systems are probabilistic. They may produce inaccurate, incomplete, outdated, biased, offensive, unsafe, non-original or legally restricted material, may misunderstand your instructions and may present invented information as fact. Output is not guaranteed to be correct, unique, lawful, fit for purpose or free of third-party rights. You must review Output and use appropriate human judgment before relying on it, publishing it or acting on it.

StartStep AI is not a substitute for a qualified professional, emergency service or human decision-maker. Unless we expressly identify a separately purchased human service and its scope, AI interactions do not create a lawyer-client, doctor-patient, accountant-client, fiduciary, employment, agency, partnership or other professional relationship.

You must be at least 18 years old to use an authenticated StartStep AI Service. StartStep AI is not designed as a persistent emotional companion or virtual intimate-relationship service. Initial-release Agents are not authorised to autonomously execute high-risk or irreversible actions.

1. Scope, Acceptance and Order of Precedence

1.1 Scope

This AI Notice applies to AI dialogue and generation, content transformation, file parsing, personal knowledge assistance, web-assisted research, Project research, templates, workflows, Agents, model-selection interfaces, AI customer support and other AI-assisted functionality actually displayed in the Service.

A feature described here may be unavailable in a particular territory, channel, plan or account. Its description is not a promise of launch, continued availability, minimum performance or compatibility.

1.2 Acceptance

You accept this AI Notice when you accept the International Terms of Service, enable or use an affected feature, or complete a feature-specific confirmation. If applicable law requires separate consent or a more prominent warning for a particular use, we will request it separately.

1.3 Related Rules

This AI Notice must be read with the StartStep AI International Terms of Service, applicable Regional Supplement, Global Privacy Notice, Compute Credits Rules, Content and Community Rules, Project rules, paid-service rules, and any feature notice or permission screen.

If documents conflict, the order stated in the International Rules Framework applies. Mandatory law prevails to the extent it cannot lawfully be varied. This AI Notice does not create an additional warranty, service level, refund right or liability cap beyond the rights and obligations stated in the controlling Rules or mandatory law.

1.4 Interface disclosures

The interface may describe a feature, model route, charging basis, context limit, connected tool, permission, source coverage or output format. A specific, current interface disclosure controls over a general description here only for that transaction or feature and only to the extent of the inconsistency.

2. What the AI Service Is and Is Not

2.1 Probabilistic operation

AI features predict, retrieve, rank, transform or generate material using statistical and computational processes. The same or similar Input may produce different Output. Apparent confidence, citations, formatting, fluent language or a human-like tone does not prove accuracy.

2.2 No personhood or independent authority

An AI feature, AI support assistant, workflow or Agent is software, not a person, employee, licensed professional or independent contracting party. It cannot bind Noviq, waive the Rules, approve an exception, promise a refund, modify a subscription, accept legal service or make a final determination of your statutory rights unless an authorised system flow or human representative expressly confirms that result.

2.3 No guarantee of originality or clearance

Output may resemble material produced for other users or material already in existence. Noviq does not guarantee that Output is copyrightable, patentable, registrable, confidential, exclusive, non-infringing or suitable for commercial use. You are responsible for obtaining any clearance, consent, licence, attribution or professional review required for your intended use.

2.4 No general monitoring promise

We may use automated and human review for safety, security, abuse prevention, support and compliance, but we do not promise to inspect every Input, Output, linked source or use of a feature. The absence of a warning, block or moderation action does not mean content is safe, lawful, accurate or approved.

3. Models, Routing and Service Providers

StartStep AI uses mainstream, trusted model and infrastructure providers selected and connected by Noviq. The actual providers, models, processing locations, retention settings and permitted fallback routes remain subject to production selection, contracting, technical verification and the disclosures required by the Privacy Documents.

3.1 Model selection and routing

StartStep AI may connect separately to multiple model, search, hosting, speech, image, safety or infrastructure providers. Depending on the task, account, location, availability, risk and cost, we may:

  1. select or route a request to one or more models or tools;
  2. divide a request into subtasks handled by different providers;
  3. use fallback models, cached results, classifiers or safety layers;
  4. update, replace, tune or discontinue a model, version or route; and
  5. apply different limits, context windows, speeds or Compute Credit charges.

A model label may identify a product family, capability tier or route rather than a permanently fixed underlying version. If an interface expressly promises a particular user-selected model for a transaction, we will use that selection or provide an appropriate notice if a material fallback is required. No provider is Noviq’s agent for making commitments to you.

3.2 Provider identification

We disclose service providers and data-processing information in the Privacy Documents to the extent required by applicable law and appropriate to the relevant processing. This AI Notice does not promise that every model name, subprocessor, version, routing weight or confidential infrastructure arrangement will be publicly listed here.

3.3 Provider terms and technical limits

Third-party models, search indexes, connected services and app stores may impose technical, geographic, content or usage restrictions. We may enforce those restrictions through the Service. A provider outage, policy change, rate limit or withdrawal may affect availability without creating liability beyond the controlling Rules and applicable law.

3.4 Data handling

Requests may be sent to the production providers identified for the relevant processing under the Privacy Documents. The private-content training rule in Global Privacy Notice Section 5.2 applies across AI routes. Zero retention, a specific processing region or other additional controls apply only where expressly offered and technically supported; they are distinct from no-training controls. Noviq will not use routing or a provider's independent terms to defeat an applicable published commitment or lawful privacy choice.

4. Inputs, Outputs and Intellectual Property

4.1 Your responsibility for Inputs

You must have all rights, permissions and lawful bases needed to provide an Input and instruct us and our providers to process it. Do not submit trade secrets, privileged material, identity documents, financial credentials, health information, precise location, biometric data, children’s data or another person’s sensitive information unless the feature expressly requests it and you are legally authorised to provide it.

You are responsible for checking files for hidden sheets, comments, metadata, tracked changes, embedded objects, malicious code and unintended personal or confidential information before upload.

4.2 Limited service-processing licence

You retain the rights you have in private Input. Providing Input does not by itself transfer ownership to Noviq. You grant Noviq and its service providers the limited, worldwide, non-exclusive, royalty-free licence and permission reasonably necessary to host, copy, transmit, format, analyse, transform and otherwise process the Input to provide, secure, maintain and improve the requested Service, enforce the Rules and comply with law.

This service-processing licence lasts for as long as reasonably needed for those purposes, subject to the retention practices in the Privacy Documents and technical backup cycles. It does not convert private Input into a Project Asset.

4.3 Output rights

Output ownership, assignment and embedded-material licences are governed by Section 5.3 of the International Terms. User-specific Output is distinct from the Project Body, Noviq technology and third-party content. A Project or commissioned deliverable remains subject to its expressly accepted specific licence. Output may be non-exclusive, unprotected or subject to third-party rights; Noviq does not guarantee registration, exclusivity or non-infringement.

4.4 Similar and non-exclusive Output

Because AI systems can generate similar results for different users, other users may receive identical or similar Output. Similarity alone does not create confidentiality, exclusivity, ownership or compensation rights against Noviq or another user.

4.5 Project contribution is separate

Creating or testing content in a private workspace does not make it a platform-owned Project Asset. If you submit material for Project publication, the separate Project contribution confirmation and Project rules govern source warranties, the transfer or licence of transferable commercial rights, platform processing, Compute Credit rewards and the absence of sales royalties or cash withdrawal rights.

4.6 Feedback

If you voluntarily provide product feedback, suggestions or error reports, you grant Noviq a worldwide, perpetual, irrevocable, transferable, sublicensable, royalty-free right to use them without restriction or payment, excluding personal data that remains governed by the Privacy Documents.

5. Feature-Specific Notices

5.1 Dialogue, writing and transformation

Text generation, translation, summarisation, rewriting and classification may omit qualifications, misstate facts, alter meaning or reproduce common phrases. Compare summaries and translations with the source, verify names, dates, figures, citations and quotations, and do not present generated content as independently verified.

5.2 Images, audio, video and synthetic media

Where available, generated or edited media may create unrealistic details, false depictions, misleading context or similarities to real people, brands and works. You must obtain required rights and consents, clearly disclose synthetic or materially altered media where law, platform rules or context requires, and preserve any provenance signal, label or watermark we apply.

You must not use the Service to create or distribute deceptive impersonation, non-consensual intimate imagery, sexual content involving minors, forged evidence, unlawful election deception, fraudulent endorsement or content intended to cause a person reasonably to believe a fabricated event occurred when that use is unlawful or materially harmful.

5.3 File analysis and personal knowledge assistance

File parsing may fail to detect text, formulas, images, permissions, embedded data, scanned pages or context. It may corrupt formatting or infer relationships that are not present. Maintain your own source copy and verify extracted data and calculations. StartStep AI is not a records-management, archival or backup service unless we expressly agree otherwise.

5.4 Web-assisted search and citations

Search-assisted Output may rely on incomplete, indexed, cached, machine-translated, paywalled or third-party material. A citation may not support the proposition for which it is offered, and a source may have changed since retrieval. Open and review important sources yourself. Third-party website terms, access controls and rights continue to apply.

5.5 Project research and opportunity analysis

Research, market, product, employment, business, revenue or opportunity analysis is informational and may be speculative. Rankings, scores, forecasts and suggested opportunities are not guarantees of demand, earnings, employment, financing, investment performance or regulatory approval. Conduct independent diligence before committing time, money or reputation.

5.6 Templates and workflows

Templates and workflows are starting points, not completed professional work. You must test them in your actual environment, review permissions and dependencies, and adapt them to applicable law, industry obligations, security standards and contractual requirements.

5.7 Agents and connected tools

An Agent may misunderstand a goal, follow malicious instructions contained in external content, use stale context, repeat an action, exceed an intended scope or disclose information to a connected service. You must:

  1. grant only the minimum permissions and data access needed;
  2. review the tool, destination, recipient, amount and content before confirmation;
  3. monitor ongoing tasks and promptly revoke unnecessary access;
  4. use test environments and backups where appropriate; and
  5. keep a human approval checkpoint before any material external effect.

At initial release, StartStep AI does not authorise Agents to autonomously initiate payments, sign contracts, submit binding legal or government filings, make high-impact eligibility decisions, send unsolicited messages at scale, irreversibly delete material, deploy to a production environment with high privilege, or take comparable high-risk actions. We may block, pause or require confirmation for an action even if you previously granted permission.

5.8 AI customer support

We may use AI for immediate customer-support responses. The interface will identify or reasonably indicate when you are interacting with AI where required by law or where omission would be misleading. AI support may explain published Rules and collect information, but it cannot finally waive charges, determine statutory rights, reject a legally valid complaint or make a binding exception unless an authorised system flow or human representative confirms the outcome.

You may request human review where the AI cannot resolve a material billing, account, safety, privacy or legal-rights issue. Human review is subject to reasonable authentication, available support channels and applicable law; it is not a promise of instantaneous response.

Each account can send a limited number of messages to AI customer support per day. When the limit is reached, AI support asks you to come back later, and you can still ask for a human. AI customer support answers questions about using StartStep AI, the Rules and your orders. It cannot receive files.

5.9 Separately paid official human services

An Official Done-for-You Service, if offered, is a separately agreed production, configuration, organisation, processing or implementation service. Its scope, required User materials, deliverables, delivery period, price, payment, revisions, acceptance and cancellation terms are set out in the quote and the services agreement that you and Noviq sign online, as described in the Paid Services and Automatic Renewal Terms Section 6. It is not a real-time one-to-one consultation, coaching or advisory service and does not include regulated professional work unless a separate compliant agreement expressly states otherwise.

The availability of Official Done-for-You Services does not make other AI Output human-reviewed. AI customer support is not the same as a separately agreed Done-for-You Service. AI customer support may explain service pages and help you describe your needs, but it does not set or change a price, issue or approve a quote, sign an agreement for Noviq, confirm a payment or approve a refund. These are done by staff whom Noviq has authorised or, for payment status, by the payment system.

5.10 Emotional interaction boundary

StartStep AI may generate empathetic or conversational language, but it is not a human relationship, therapist, crisis counsellor or persistent emotional companion. Do not rely on it for emotional exclusivity, validation of delusions, crisis intervention or decisions involving self-harm or harm to others. In an emergency or immediate danger, contact local emergency services or an appropriate qualified person. StartStep AI may not know your identity or location and cannot guarantee intervention.

5.11 AI Team stage runs

When you ask the AI Team to complete a stage, it runs the steps one after another without stopping at each decision. Where a step needs a choice from you, the AI picks the recommended option, marks it as chosen by AI and carries on. Where information is missing, it marks what is needed instead of making up facts or figures, and any assumed figure is marked as an assumption. At the end, you get a review list of these choices, gaps and assumptions, where you can accept or change each item, or accept them all. Changing an item re-runs only the later steps it affects, and those steps are charged for the processing they use. The AI Team does not send anything externally, publish, pay or delete on your behalf. It places drafts in the review list for you to act on.

6. Compute Credits and AI Metering

6.1 Internal service units

Compute Credits are internal service-use units. They are not cash, deposits, stored value, cryptocurrency, securities, wages or a guarantee of access to a fixed amount of computing power. They cannot be withdrawn, transferred or traded except where StartStep AI expressly enables a limited function under the Compute Credits Rules.

6.2 Metering

AI use may consume Compute Credits according to model route, input and output size, media duration or resolution, tool calls, search, storage, retries, priority, risk review and other displayed factors. A single user request may involve multiple technical calls. A fixed charge is shown on the action button. Other requests are charged for the processing actually used, without an estimate or a maximum shown in advance, as set out in Compute Credits Rules Sections 7.2 and 7.4.

We may hold Credits before each model call and settle the actual usage after processing. A task that is cancelled, blocked, interrupted or unsuccessful may still consume Credits for work already performed, subject to the Compute Credits Rules, any specific interface disclosure and mandatory law. A run that fails for a reason on our side is not charged, as set out in Compute Credits Rules Section 8.5.

6.3 Changes

We may change future metering rates, included allowances, model availability and promotion rules. We will not retroactively change the charge for a completed request. Rules for expiry, promotional Credits, refunds and correction of manifest metering errors are stated in the Compute Credits Rules and applicable paid-service rules.

7. Responsible Use Obligations

7.1 General duty

You must use AI features lawfully, in good faith, within the permission and purpose communicated through the Service, and with reasonable safeguards proportionate to the foreseeable impact.

7.2 Prohibited uses

You must not use or attempt to use the Service to:

  1. violate law, court order, sanctions, export control or another person’s enforceable rights;
  2. exploit, sexualise, groom, traffic or endanger a minor, or create child sexual abuse material;
  3. create or distribute non-consensual intimate imagery, unlawful deepfakes, deceptive impersonation, forged identity material or fraudulent evidence;
  4. commit fraud, phishing, credential theft, identity theft, extortion, scams, money laundering or deceptive commercial practices;
  5. develop, deploy or facilitate malware, destructive code, unauthorised access, denial-of-service activity or attacks on systems, while lawful authorised security testing remains subject to our security rules;
  6. facilitate unlawful weapons activity, violent wrongdoing, terrorism, self-harm encouragement or instructions intended to cause serious physical harm;
  7. unlawfully collect, infer, expose, sell or misuse personal data, precise location, communications, biometric identifiers or confidential information;
  8. make or materially facilitate unlawful discrimination or a high-impact decision about employment, housing, education, credit, insurance, healthcare, essential services or legal status without required authority, testing, notice, human review and appeal;
  9. generate spam, fake engagement, undisclosed paid endorsements, manipulative reviews or unsolicited communications at scale;
  10. infringe intellectual property, privacy, publicity, confidentiality or database rights, or remove rights-management information;
  11. falsely represent AI Output as human-created, independently verified or officially endorsed where that representation is deceptive or unlawful;
  12. manipulate a person through deception, coercion, emotional dependency or exploitation of a known vulnerability in a way likely to cause material harm;
  13. evade safety filters, access controls, geographic restrictions, rate limits, payment controls, sanctions screening, moderation or verification;
  14. extract or disclose system prompts, model weights, non-public datasets, credentials, secrets or security mechanisms, or reverse engineer the Service except to the limited extent a restriction is prohibited by law;
  15. scrape, harvest or systematically reproduce the Service, Project Assets or Outputs to build a substitute service, train or improve a competing model, or create an unauthorised dataset;
  16. benchmark or publish test results in a misleading manner or in breach of confidentiality, security or access restrictions; or
  17. cause another person or automated system to do any of the above.

7.3 No circumvention by indirect instruction

These restrictions apply whether conduct occurs through direct prompts, encoded instructions, uploaded files, connected tools, multi-step workflows, another account or a third party. You may not use benign-looking intermediate requests to assemble a prohibited outcome.

8. High-Risk and Professional Contexts

8.1 Human verification required

You must not rely solely on Output for a decision that may materially affect a person’s rights, safety, health, liberty, livelihood, access to essential services or legal position. Use qualified human review, reliable source data, testing, documentation, required notices and a meaningful route to contest the decision.

8.2 Professional information

AI Output concerning law, tax, accounting, medicine, mental health, finance, investment, insurance, employment, immigration, safety, engineering or regulated activity is general information only. It may not reflect current law, local rules, your facts or professional standards. Consult an appropriately qualified professional before acting.

8.3 Code and technical instructions

Generated code, commands, configurations and security advice may contain vulnerabilities, destructive instructions, licence conflicts or hidden dependencies. Review, scan and test them in an isolated environment. Do not execute with administrative or production privileges until a competent person has verified the action and a recovery plan exists.

8.4 No regulated or high-impact automation at launch

Unless StartStep AI later launches a feature with specific disclosures and safeguards, the Service is not offered to make fully automated decisions or execute regulated high-impact activities. A user’s attempt to repurpose a general feature for such an activity is outside the intended Service and remains the user’s responsibility.

9. Safety Controls, Review and Enforcement

9.1 Controls

We may use input and output filters, abuse classifiers, rate limits, permission gates, content provenance, temporary holds, transaction review, model-routing restrictions, automated monitoring and human review to protect users, Noviq, providers and the public.

9.2 Platform discretion

To the extent permitted by law, we may refuse a request, modify an Output, disable a tool, require verification, restrict a model, suspend a workflow, preserve evidence, remove content, limit an account or terminate access where we reasonably consider it necessary for safety, security, legal compliance, provider obligations, service integrity or enforcement of the Rules.

We are not required to disclose confidential detection methods, model prompts, risk thresholds, reports that would identify another person, privileged material or information whose disclosure would undermine safety or security.

9.3 Reporting and review

You may report potentially unlawful or harmful AI content, an account action or a material support error through the published reporting channel. Include the relevant account, date, feature, Output or transaction identifier and a concise explanation. Do not resend unlawful intimate or child-abuse material; provide identifiers and follow the reporting instructions.

Where required by applicable law or offered under our Rules, you may request review of a moderation or account decision. We may require authentication and may decline repetitive, abusive, manifestly unfounded or technically impossible requests to the extent permitted by law.

9.4 Cooperation and preservation

We may preserve relevant records, cooperate with lawful requests and make reports to competent authorities or safety organisations where required or permitted by law. Nothing here requires Noviq to monitor generally or to take an action prohibited by law.

10. Transparency, Labels and Public Use

10.1 AI disclosure

You must disclose the use of AI or synthetic media where required by law, an applicable platform rule, professional duty or the context needed to avoid misleading another person. You must not remove or circumvent a watermark, label, metadata field or provenance signal applied by StartStep AI or a provider. Images generated through StartStep AI may carry provenance data, such as C2PA Content Credentials or an invisible watermark, added by StartStep AI or a model provider.

10.2 Your public representations

Do not state or imply that StartStep AI, Noviq, a model provider or a Project Host endorses you, your Output or your product unless authorised in writing. Do not use AI to fabricate reviews, credentials, evidence, customer results or affiliation.

10.3 Platform records

We may record model route, tool calls, confirmation events, safety actions, Credit usage, version information and other operational data reasonably needed to provide and protect the Service, resolve disputes and comply with law, subject to the Privacy Documents.

11. Availability, Changes and Data Preservation

11.1 No permanent model or feature commitment

We may add, modify, limit, test, suspend or discontinue a model, provider, tool, Agent capability, limit, plan inclusion or output format. We may do so without advance notice where reasonably necessary for security, legal compliance, provider availability, urgent risk or service integrity. For a material adverse change to a paid Service, the paid-service rules and mandatory consumer law govern notice and remedies.

11.2 Beta and experimental features

A feature marked beta, preview, experimental, test or similar may be less reliable, may change rapidly and may be withdrawn. Unless mandatory law provides otherwise, it is supplied without a service-level commitment and should not be used for critical production dependency.

11.3 Retention and backups

Conversation history, uploaded files, tool state, web results or Agent memory may have limits and may be deleted, truncated or become unavailable. You must keep independent copies of material you need. Retention and deletion of personal data are governed by the Privacy Documents; technical backups may persist for limited cycles.

11.4 Security limitations

No system is completely secure. You must use appropriate access controls, encryption, redaction, data minimisation and backups for your risk. Do not rely on a consumer-facing AI workspace as the sole repository for confidential or mission-critical information.

12. Disclaimers and Allocation of Responsibility

12.1 Incorporation of Terms

The warranties, disclaimers, exclusions and limitations of liability in the International Terms of Service and applicable Regional Supplement apply to AI features. This AI Notice does not establish a separate or cumulative liability cap.

12.2 Non-excludable rights

Nothing in this AI Notice excludes, restricts or waives a warranty, remedy, liability or consumer right that applicable law does not permit us to exclude, restrict or waive. References to user responsibility do not transfer to you a legal duty that mandatory law places exclusively on Noviq.

12.3 User-controlled use

Subject to non-excludable law, you are responsible for the purpose, context, Input, instructions, human review, permissions, recipients, publication and external actions you control. Noviq is not responsible for a decision or representation you make contrary to an express warning, applicable law or the intended scope of the Service.

12.4 Third-party services

A third-party model, search source, website, connected account or tool remains subject to its own availability, rights and terms. Noviq does not adopt third-party content merely by retrieving, linking, summarising or processing it.

13. Territory-Specific Safeguards

13.1 General rule

This section highlights initial product boundaries; it is not an exhaustive statement of local law. Applicable Regional Supplements and mandatory law continue to apply. A regulatory proposal, voluntary framework or future law is not incorporated as a contractual promise merely because Noviq considers it in product governance.

13.2 Singapore

For Singapore users, personal-data processing relating to AI is governed by the Global Privacy Notice, Singapore privacy provisions and the Personal Data Protection Act 2012 where applicable. Noviq may use risk scoping, human approval checkpoints, technical controls and incident processes appropriate to the feature. References to regulatory governance frameworks describe risk-management considerations and do not create a warranty that every recommendation is contractually binding.

13.3 United Kingdom

For United Kingdom consumers, the use of AI does not reduce mandatory consumer rights. AI support may not finally deny a valid cancellation, refund, complaint or statutory remedy. We will not knowingly present an AI agent as a human where doing so would be misleading, and material decisions about consumer rights remain subject to an authorised process or human review where required by law.

13.4 United States

For United States users, applicable federal and state laws concerning unfair or deceptive practices, privacy, biometrics, synthetic media, discrimination and automated decisions continue to apply. Initial-release StartStep AI is not offered as a system for fully automated high-impact decisions concerning employment, housing, education, credit, insurance, healthcare, essential services or legal status. Any state-specific rights are described in the applicable United States schedules and privacy notices.

14. Changes to this AI Notice

We may update this AI Notice for changes in law, models, providers, features, risks or business operations. We will provide the notice and obtain renewed acceptance required by applicable law or the International Terms of Service for a material change. Changes do not retroactively alter a completed transaction or take away a non-waivable accrued right.

The version shown in the Service at the time of the relevant activity applies, subject to the International Rules Framework and mandatory law.

15. Contact

For general questions: support@startstep.ai
For AI safety or harmful-content reports: support@startstep.ai
For privacy requests: privacy@startstep.ai
For legal notices: legal@startstep.ai

Postal address: Noviq Pte. Ltd., 68 Circular Road, #02-01, Singapore 049422.

We may require reasonable information to authenticate a request, locate the relevant record and prevent fraud.

Appendix A — Initial-Release Feature Boundary

The following table is an operational summary and is subject to the controlling Rules and actual interface.

AreaInitial-release position
Eligible usersIndividuals aged 18 or older in Full Version or permitted Global Lite areas; Restricted Version limits apply
ChannelsStartStep AI website/web app, iOS app and Android app as actually released
AI functionsDialogue and generation, file and knowledge assistance, web-assisted research, Project research, templates, workflows, Agents and AI support as displayed
Model architectureMainstream, trusted model and tool providers, each connected directly; the providers in use and the data they process are listed in the Third-Party Service Provider and Subprocessor List, which is updated before a new route is enabled; routing, fallback and replacement permitted
High-risk executionNo autonomous payments, signatures, binding filings, high-impact eligibility decisions, irreversible deletion or high-privilege production deployment
Customer supportAI first response with human escalation for material unresolved matters
Official human-assisted servicesSeparately paid Official Done-for-You production, configuration, organisation, processing or implementation, if displayed
Emotional useNo persistent emotional companion or virtual intimate-relationship product
Project contributionSeparate affirmative Project contribution confirmation required; private AI use alone does not transfer ownership
CreditsInternal Compute Credits only; no cash redemption, trading or fixed monetary value
Enterprise functionsNo team workspace, multi-seat enterprise knowledge base, public API, white-label resale, client embedding or private deployment unless later expressly launched

Back to Rules