Versatl

Versatl — Terms of Service

Effective Date: April 28, 2026

Last Updated: August 8, 2026

View previous versions

1. Agreement to Terms

By accessing or using Versatl ("Platform" or "Service"), operated by Mfini Inc. ("Mfini," "Company," "we," "us," or "our") at versatl.ai, you ("User," "you," or "your") agree to be bound by these Terms of Service ("Terms"). If you do not agree, do not use the Service.

Versatl is a product of Mfini Inc. These Terms constitute a legal agreement between you and Mfini Inc.

These Terms apply to all users, including consumers who hire agents, creators who publish agents on the marketplace, and administrators.

2. Description of Service

Versatl is an AI agent platform where users hire purpose-built AI agents to complete tasks and run recurring services. The Platform provides:

  • Tasks: One-shot requests submitted to AI agents (e.g., "write a blog post about sustainable fashion").
  • Subscriptions: Recurring scheduled tasks that run on a schedule you configure (e.g., "send me Dallas real estate news every morning").
  • Event Triggers: Webhook-fired automations — you connect an external event source (e.g., GitHub, Stripe, or Shopify) to an agent, and each incoming event runs the agent with instructions built from the event. Each fire is billed as a task run (Section 5.1).
  • Workflows: Multi-step orchestrations that chain multiple agents and tools, with deterministic control flow, scheduled or on-demand execution, and per-run billing.
  • Agent Marketplace: A catalog of first-party and third-party AI agents that users can discover, subscribe to, and configure.
  • Agent Builder: A workspace for users to define and publish their own AI agents and workflows declaratively, optionally listing them on the marketplace.
  • Orchestrator: A per-user master agent that acts as the conversational gateway across chat surfaces (Web, Telegram, Slack, Discord) and delegates work to specialist agents.
  • Proactive Assistant: An optional background assistant that, when you enable it, runs periodic reasoning cycles on your behalf — reviewing the goals, schedules, and connected data you have given it access to — and initiates tasks, messages, or digests within the autonomy level you configure. Actions the Proactive Assistant takes under your standing instructions and autonomy settings are use of the Service by you.
  • Platform-Managed Bots: Shared bots that the Platform operates on Telegram, Slack, and Discord for users who connect those messaging channels, enabling chat with the Platform without each user needing to register their own bot application.
  • MCP Server: A Model Context Protocol server operated by the Platform at mcp.versatl.ai (or its successor URL). When you opt in (Settings → MCP), external MCP-compatible clients you authenticate may invoke a user-controlled subset of your tools, agents, and memory on your behalf, subject to the master switch and the per-tool approval matrix you maintain.
  • Third-Party Integrations: Connections to external services that agents can act on your behalf through. The current inventory is maintained in the Connected Services Appendix at versatl.ai/legal/connected-services-appendix, which is incorporated by reference. Integrations may be added or removed; material changes are reflected in the appendix.
  • Memory System: Agents accumulate experience across your tasks to improve personalization over time.

3. Eligibility

You must be at least 18 years old and capable of forming a binding contract to use the Service. By using the Service, you represent that you meet these requirements. If you are using the Service on behalf of an organization, you represent that you have authority to bind that organization to these Terms.

4. Account Registration

4.1 Account Creation

You must create an account to use the Service. Authentication is managed through Clerk, our third-party authentication provider. You are responsible for maintaining the security of your account credentials.

4.2 API Keys

You may generate API keys (prefixed with vtl_) for programmatic access. API keys are shown only once at creation and cannot be recovered. You are responsible for securing your API keys and must revoke any compromised keys immediately.

4.3 Account Accuracy

You agree to provide accurate, current, and complete information during registration and to update your information as necessary. We reserve the right to suspend or terminate accounts with false or misleading information.

5. Billing

5.1 Pricing Model

Versatl is free to register and free to build private AI entities for personal, friends-only, or company-internal use. The Platform charges only for metered usage: the provider costs your runs consume on your behalf — LLM tokens (by agent runs and by the assistant and background features you use or enable) and any search or other third-party service calls those runs make on your behalf — agent fees set by publishers, and the platform fee described below.

When you run an agent, the Platform bills the following line items:

  • LLM token cost — the cost of the underlying language-model tokens consumed by the run. If you have provided your own LLM API key for the relevant provider ("BYOM"), this amount is paid by you to that provider directly and the Platform does not charge you for it. If you have not provided your own key, the Platform charges you at the provider's published per-token rate, without markup.
  • Search and service cost — some runs make calls to third-party services on your behalf (for example, web search). The Platform charges you the provider's rate for those calls, without markup.
  • Platform fee — a percentage of the total provider cost described above: the LLM token cost plus any search and service costs, calculated against the published rates of the relevant providers (for LLM tokens: Anthropic, OpenAI, Google, or OpenRouter). The current Platform Fee rate is published on the Platform's Pricing Page at versatl.ai/pricing, which is incorporated into these Terms by reference; the rate in effect at the time of a run applies to that run. The fee is charged identically to platform-key and BYOM usage; for BYOM the fee is computed against the same published LLM rates even though you paid the provider directly. Prompt-cache savings reduce both the LLM cost and the platform fee proportionally. This fee covers compute, storage, networking, and other operational overhead. Changes to the Platform Fee rate are made as described in Section 16: increases take effect no earlier than 30 days after we notify you, apply prospectively only, and are reflected on the Pricing Page; decreases may take effect immediately.

Each agent in the marketplace is additionally priced by its publisher (or by the Platform for first-party agents) using one of the following models:

  • Per-task: a fixed amount charged when you run the agent. Runs initiated by your event triggers charge the same per-task fee per fire. Runs executed under an active subscription's schedule do not incur the per-task fee (the subscription-period fee covers them); other runs of the same agent — direct or trigger-initiated — do. Per-task and per-run fees are automatically returned if the run terminally fails or is cancelled (Section 5.6).
  • Subscription tier: a fixed monthly (or other-period) amount that authorizes recurring scheduled runs.
  • Per-run (workflows): a fixed amount charged for each run of a published workflow, with the same fee-return treatment as per-task fees (Section 5.6).

Many platform-built agents charge no agent fee at all; you pay only the metered usage described above.

5.1.1 Signup Credit and Welcome Credit Offer

Signup credit. New accounts receive a one-time promotional signup credit (currently $5) so you can try the Platform's agents before adding funds. The signup credit has no cash value, is non-refundable, is never paid out, and is not contingent on any deposit — it is not reversed under any circumstance, including refunds of later deposits. We may change the signup credit amount, or discontinue it, for newly created accounts at any time; credits already granted are unaffected.

Welcome credit. Get $25 in promotional credit when you turn on Auto Top-Up and add $25 or more to your account balance. When you accept the offer, you authorize a one-time charge of at least $25 (or the higher amount you select) to your saved payment method; on success, your paid amount plus the $25 promotional credit is added to your balance and Auto Top-Up is enabled. For example, depositing the $25 minimum results in $50 in your balance ($25 paid + $25 promotional credit); depositing $100 results in $125 in your balance ($100 paid + $25 promotional credit). Both the paid amount and the $25 promotional credit are non-refundable per Section 5.5.

Your account balance is internally tracked as two components — paid funds and promotional credit — that debit together at an equal rate as you use the Service. For each unit of balance consumed, one half is drawn from paid funds and one half from promotional credit, until either component reaches zero. Once one component reaches zero, subsequent charges are drawn entirely from whichever component still has a positive balance. The total balance shown to you is always the sum of the two components.

The welcome credit is contingent on the matching paid deposit being retained. If any portion of that paid deposit is refunded under Section 5.5, any unused welcome credit is reversed at the same time. The signup credit is not affected by such a reversal.

Unspent balance remains available indefinitely. The offer is one-time per account and may be withdrawn or modified at any time. You may turn on Auto Top-Up without claiming the offer; the offer remains available to claim later as long as we are still running it.

5.2 Account Balance

Paid agents are charged against your account balance, a prepaid credit you maintain on the Platform. You fund your balance through our payment processor, Stripe: adding funds charges a payment method saved to your account (saving your first payment method uses a Stripe-hosted setup page). When you run a paid agent, the price of the run is deducted from your balance. Subscription fees are deducted at the start of each billing period. Pausing a subscription stops its scheduled runs immediately and suspends billing: you are not charged while paused, and the unused portion of your current billing period is preserved and resumes when you unpause. Cancelling a paid subscription that still has prepaid time takes effect at the end of that time: the scheduled runs you have already paid for continue, no further period fees are charged, and the subscription then ends. This applies whether the subscription is running or paused — cancelling a paused subscription resumes it so the time preserved by the pause is actually delivered to you, and it then ends at that point rather than being forfeited. Cancelling a subscription with no prepaid time remaining — a free subscription, one whose period has already ended, or one in a payment-failed state — takes effect immediately. Because compute charges post after the underlying work completes, a charge may take your balance briefly below zero; any negative amount is an amount you owe and is recovered from your balance when it is next funded.

Starting new work requires a positive balance: when your balance is at or below zero, new tasks, runs, and scheduled or trigger-fired executions are refused until you add funds (accounts with Auto Top-Up enabled and a payment method on file are not refused, since the next charge replenishes the balance automatically). Scheduled and trigger-fired runs refused for this reason appear in your task history as failed runs with the reason shown.

You may optionally enable Auto Top-Up: when your balance falls below a threshold you set, we charge a saved payment method on file to bring it back up. Auto Top-Up requires you to first save a payment method via Stripe.

5.3 Operational Limits

The Platform enforces operational limits to maintain service quality, including per-account rate limits, request-size limits, and a platform-level per-task hard cap. These are not pricing tiers; they are abuse-prevention mechanisms and may change with reasonable notice.

5.3.1 Spend Caps

Two spend caps bound the charges on your account. The per-task safety cap is a platform-level limit (currently $10 per task) on a single run's compute and is not user-configurable; it halts runaway runs on the best-effort basis described in Section 5.6 — the run finalizes with any partial output delivered, and you are charged for the compute actually consumed up to the halt. The daily budget cap bounds your total daily charges — compute and agent, subscription, and workflow fees alike, across direct, trigger-fired, scheduled, and background activity (subscription fees counted spread across their billing period; see Section 5.6); you may change or disable it from the Billing page, subject to a platform maximum. Reaching the cap pauses new direct submissions (including workflow runs), trigger fires, and assistant-initiated runs — chat delegations, task retries, and runs invoked through the MCP integration; scheduled and background activity is additionally governed by its own controls, including the Proactive Assistant's separate daily budget. Charges from all of these still count toward the daily total. Both caps operate on the best-effort basis described in Section 5.6; the values in effect for your account are shown on the Billing page.

5.4 Payment Processing

All customer payments are processed through Stripe. By topping up your balance or enabling Auto Top-Up, you authorize Mfini Inc. (via Stripe) to charge your selected payment method. Mfini Inc. is the merchant of record for all customer payments. You may manage saved payment methods through the Platform's billing settings or the Stripe Customer Portal.

5.4.1 Sales Tax

Where applicable, the Platform collects and remits sales tax on top-ups and other balance charges. Tax is calculated by the Platform using tax-rate data obtained from a third-party rate provider, based on the billing address associated with your saved payment method (or the postal code on file where a full address is unavailable). Tax appears as a separate line item on the itemized receipts the Platform issues; it is added to — not deducted from — the amount you selected. The Platform is registered to collect and remit sales tax in jurisdictions where it is required to do so — currently the State of Texas (United States). As the Platform registers in additional jurisdictions, tax will be collected on charges to customers in those jurisdictions as well.

5.5 Refunds and Disputes

Amounts added to your account balance (via top-up or Auto Top-Up) are non-refundable. Unspent balance remains available indefinitely until you use it. We do not reverse card charges for unspent top-ups except in the following cases:

  • Subscription periods: subscription fees are charged in advance for the upcoming period. If the Platform is unable to deliver the contracted scheduled runs for that period (e.g., the underlying agent has been removed from the marketplace), you may request a pro-rated refund of the unused portion to your account balance.
  • Billing errors or unauthorized charges: mistaken or unauthorized charges (duplicate webhooks, system errors, fraud) are refunded to the original payment method on request, limited to the amount of the disputed charge not already refunded and, in all cases, capped at your total unspent paid balance at the time of the refund (promotional credit is never refunded).
  • Statutory rights: where applicable consumer law requires a refund (for example, the EU 14-day right of withdrawal for individual consumers on their first paid transaction), we comply with that obligation. Refunds under this provision are limited to the amount of the transaction not already refunded and, in all cases, capped at your total unspent paid balance at the time of the refund (promotional credit is never refunded).
  • Discretionary Platform refunds: at the Platform's sole discretion, on a case-by-case basis, we may issue a refund of some or all of an agent-run charge (e.g., where a Platform-side incident materially prevented the run from executing meaningfully). Discretionary refunds are not automatic and are not required; contact support to request review.

When we refund any portion of a paid top-up that was matched by a welcome credit, any unused welcome credit is reversed at the same time, regardless of the size of the refund (see Section 5.1.1). The signup credit is not affected by such a reversal.

Card disputes: if you dispute a charge through your card issuer, Stripe will pull the disputed funds from our account during the investigation. We may suspend your Platform account during an open dispute, and we may pass through Stripe-imposed dispute fees if the dispute is found in our favor and the dispute appears to have been filed in bad faith.

Marketplace publisher payouts: when a customer charge is reversed (refunded or charged back), the corresponding publisher revenue is also reversed under the terms of the Marketplace Publisher Agreement (§4.9). This is an internal accounting matter and does not affect your refund as a customer.

5.6 Compute-Based Billing and Safety Caps

Charges follow provider usage. Agent runs on the Platform consume compute at third-party model providers (Anthropic, OpenAI, Google, OpenRouter, and others). You are charged the provider's per-token rate for the tokens actually consumed on your behalf, plus the Platform fee described in Section 5.1. Because our third-party providers charge us for every successful API call regardless of whether the final agent output satisfied your intent, we charge you on the same basis: you pay for compute consumed, not for outcome.

Compute charges are never outcome-refunded; agent fees are completion-contingent. An agent run that produces empty, incomplete, or verifier-rejected output is not refunded its compute charges, because the underlying provider charges we incur are the same as for a successful run. The per-task agent fee (and the per-run fee for workflow runs) is different: if a run terminally fails or is cancelled, that fee is automatically returned to your balance — you pay agent fees only for runs that complete. Runs that complete with warnings (for example, a capped run delivering partial output) count as completed and keep their fee. Discretionary Platform refunds under §5.5 remain available for exceptional cases.

Per-task safety cap (best-effort). The Platform enforces a per-task compute cap (visible on the Billing page under "Per-task safety cap") to limit runaway cost from a single agent run. The cap is checked between LLM calls; when the running cost of a task exceeds the cap, the Platform halts the run before the next LLM call. Because the check runs between calls rather than during a call, a task may exceed the cap by up to one additional LLM call's worth of compute before the halt takes effect. You are charged for the compute actually consumed, up to and including the call that triggered the cap. Any partial output produced up to the halt is delivered to you.

Daily budget cap (best-effort). The Platform enforces a per-day budget cap (visible on the Billing page under "Daily budget cap"). Charges of every kind from all of your activity — compute charges and agent, subscription, and workflow fees; direct, trigger-fired, scheduled, and background — count toward the daily total. Subscription-period fees count spread evenly across the billing period they cover, rather than as a single charge on the renewal date; all other charges count on the day they are incurred, and charges that are refunded no longer count. Once the total exceeds the cap, the Platform pauses the new-work lanes described in Section 5.3.1 — direct submissions (including workflow runs), trigger fires, and assistant-initiated runs (chat delegations, task retries, and runs invoked through the MCP integration); scheduled and background activity is paused by its own controls (Section 5.3.1) rather than by this cap, subscription renewals are never blocked by this cap, and concurrent in-flight tasks may push the total slightly above the cap before a pause takes effect.

Cost estimates are estimates. Any cost estimates shown before or during a run are best-effort predictions based on the models and tokens the agent is expected to use. Actual charges reflect what the providers actually billed us and may differ from estimates.

6. User Content and Data

6.1 Your Content

You retain ownership of all content you submit to the Service ("User Content"), including task instructions, service configurations, user context data, and any outputs generated by agents on your behalf.

6.2 License Grant

By submitting User Content, you grant Versatl a limited, non-exclusive, worldwide license to process, store, and transmit your content solely for the purpose of providing the Service. This includes sending your content to third-party AI model providers (currently Anthropic, OpenAI, Google, and OpenRouter, per the self-healing model catalog described in §8.1) for processing.

6.3 Agent Memory

When agents complete tasks on your behalf, the Platform stores episodic records (task summaries, steps taken, outcomes) and may generate semantic memories (distilled knowledge patterns) to improve future agent performance for you. You can control cross-agent memory sharing through your settings. Memory data is deleted when you delete your account.

6.4 Prohibited Content

You may not submit content that:

  • Violates any applicable law or regulation
  • Infringes the intellectual property rights of others
  • Contains malware, viruses, or harmful code
  • Attempts to extract, reverse-engineer, or abuse AI model capabilities
  • Seeks to generate content that is illegal, harassing, defamatory, or harmful
  • Attempts to circumvent Platform guardrails, safety filters, or usage limits

7. Third-Party Integrations

7.1 Connected Services

The Platform enables you to connect external services so that agents can act on your behalf. The current inventory of supported services and their connection categories (Direct OAuth, Composio-brokered, MCP-brokered) is maintained in the Connected Services Appendix at versatl.ai/legal/connected-services-appendix, which is incorporated by reference into these Terms. By connecting a service, you authorize the Platform to store the corresponding credentials (encrypted) and to make API calls to that service as directed by your agents.

For Google services (Gmail, Google Calendar, Google Drive, YouTube), the default connection path is brokered through Composio Inc. ("Composio"), a CASA Tier 2-verified application. When you use the default path, you authenticate directly with Google through a consent screen that identifies Composio as the application receiving access; Composio holds the OAuth credentials on its infrastructure and dispatches tool calls on the Platform's behalf. A direct-OAuth path (Versatl-held credentials) is retained as a development and fallback option.

7.2 Your Responsibility

You are solely responsible for ensuring that your use of connected services complies with those services' terms of use. The Platform acts as your agent when accessing these services — we do not independently decide what actions to take. Actions are performed based on your task instructions and agent configurations.

7.3 Credential Security

Your credentials for connected services are encrypted using Fernet symmetric encryption and stored in our database. Credentials are never exposed to AI models, never logged, and are decrypted only at the moment they are needed to make an API call on your behalf.

7.4 Revocation

You may disconnect any service at any time through the Connections page (or, for Bring-Your-Own-Model LLM provider keys, through Settings → AI Models). The default disconnect behavior is a hard delete of the corresponding credential record on the Platform, and the Platform additionally calls the vendor's published revoke endpoint (where one exists) to invalidate the OAuth grant at the upstream provider. For Composio-brokered Google connections, disconnection also instructs Composio to delete the corresponding connected account on its infrastructure, severing the underlying OAuth grant.

Exception for billing-relevant credentials. When you disconnect a Bring-Your-Own-Model (BYOM) LLM provider key, the credential row is marked revoked (revoked_at timestamp set) rather than hard-deleted, and the encrypted key blob is retained for up to eighteen (18) months. The retained row is invisible to every read path and cannot be used to make a call. It is preserved solely so that the Platform's daily billing reconciliation can determine — at the close of each accounting period — whether usage events that were tagged as BYOM in fact correspond to a credential that was active at the time the call was made. On account deletion, retained BYOM credential rows are deleted via the same cascade as the rest of your data, without waiting for the 18-month window to elapse.

8. AI Processing

8.1 AI Model Providers

The Service uses third-party AI model providers to power agent execution, currently including Anthropic (Claude family), OpenAI (GPT family), Google (Gemini family), and OpenRouter (an aggregator that routes to upstream model hosts). The Platform maintains a self-healing model catalog that is refreshed from each provider's official model feed; deprecated models are retired on a published schedule and replaced with successor models. You may select a preferred model per tier from Settings → AI Models, and you may supply your own API key for any supported provider ("Bring Your Own Model" / BYOM), in which case that tier's traffic is sent directly to that provider using your key. By using the Service, you acknowledge and consent to your content being processed by these providers subject to their respective terms and privacy policies.

8.2 No Guarantee of Accuracy

AI-generated outputs may contain errors, inaccuracies, or hallucinations. You are responsible for reviewing all agent outputs before acting on them or publishing them externally. The Platform provides guardrails (PII detection, content safety checks, budget limits) as a safety net, but these are not guaranteed to catch all issues.

8.3 Human-in-the-Loop

Certain agent actions (e.g., publishing social media posts, sending emails) may require your explicit approval before execution. Whether a given action requires approval is determined by the agent's definition: for agents you build yourself, you set these approval gates; for marketplace agents, they are set by the agent's publisher. When an agent requests approval, execution pauses until you approve or deny the action.

8.4 Model Selection and Substitution

The Service routes each request to an underlying large language model selected automatically or based on the preferences you configure. When a model you have selected becomes unavailable, is deprecated by its provider, or is otherwise unable to serve the request, the Service may substitute a functionally similar model in the same tier and (where applicable) the same provider family. Substitutions are recorded in the run's metadata, accessible on the task detail page, so you can identify when one occurred. Substitution can be disabled per agent by that agent's author: enabling "Require pinned model" in the agent's definition causes requests that cannot be served by the pinned model to fail rather than substitute. For agents you build yourself, you are the author and may set this option; for marketplace agents, it is set by the publisher where reproducibility is essential (see the Marketplace Publisher Agreement § 2.8). Two account-level controls in Settings → AI Models further restrict which models may serve your runs: "Strict BYOM mode" limits runs to models for which you have provided your own API key, and "No model substitution" limits runs to the exact models you have configured per tier — when it is enabled, any task that would require a different model, including a substitute for a retired or unavailable model, is refused before it starts (and before any charge) rather than run on a model you did not choose. In the rare case that a model is retired while a task is already running, the outcome depends on your settings: for agents with "Require pinned model" and for accounts with "No model substitution" enabled, the task fails at that point rather than continuing on a substitute; otherwise the run continues on a substitute selected under this Section, and the substitution is recorded in the run's metadata as described above. In either case, compute already consumed is charged on the basis described in Section 5.6.

9. Workspaces

9.1 Workspace Creation

You may create workspaces to organize team collaboration. The workspace creator is the owner. Owners can invite members and assign roles (admin or member).

9.2 Roles and Permissions

  • Owner: Full control over the workspace, including member management and deletion.
  • Admin: Can manage workspace settings and members but cannot delete the workspace.
  • Member: Can view and interact with workspace resources.

9.3 Workspace Data

Tasks, services, and agent subscriptions can be scoped to a workspace for organizational purposes. Workspace-scoped data is accessible to all workspace members according to their role.

10. Intellectual Property

10.1 Platform IP

The Service, including its software, design, architecture, and documentation, is owned by Mfini Inc. and protected by intellectual property laws. These Terms do not grant you any rights to our intellectual property except the limited right to use the Service as described herein.

10.2 Your Outputs

You own the outputs generated by agents on your behalf. We claim no ownership over agent-generated content. However, we cannot guarantee that AI-generated outputs are original or non-infringing — you are responsible for ensuring outputs comply with applicable laws before use.

11. Building Agents

Any user with an approved Versatl account may create AI agents, skills, and workflows (collectively, "Agents" for purposes of this Section 11) for personal use, share them with other users, or — subject to the additional terms in the Marketplace Publisher Agreement — publish them to the Versatl marketplace.

11.1 Ownership of Your Agent Definitions

You retain ownership of the agent definitions, prompts, configurations, and other content you author when building an Agent on the Platform ("Agent Materials"). Mfini does not claim ownership over your Agent Materials.

11.2 License Grant to Mfini

To operate the Platform on your behalf, you grant Mfini a worldwide, non-exclusive, royalty-free license to:

  • Host, store, and display your Agent Materials;
  • Execute your Agents on your behalf and on behalf of users to whom you grant access (via direct sharing or marketplace distribution);
  • Create automated copies of Agent Materials as required for redundancy, backups, version snapshots, and operational continuity;
  • Use anonymized telemetry from Agent execution (token counts, error rates, performance metrics) to operate, improve, and secure the Platform.

This license terminates when you delete the Agent, except that backups, versioned snapshots, and aggregated telemetry may be retained for the periods described in our Privacy Policy.

11.3 Your Representations and Warranties

By creating, sharing, or publishing an Agent, you represent and warrant that:

  • You have all rights, licenses, consents, and permissions necessary to upload, share, and execute the Agent Materials, including all prompts, instructions, embedded content, and any external resources the Agent references;
  • The Agent Materials do not infringe any third party's intellectual property, privacy, publicity, or other rights;
  • The Agent Materials do not contain content that is unlawful, defamatory, fraudulent, or designed to deceive, harm, harass, or violate the rights of others;
  • The Agent does not circumvent platform safety controls, the Acceptable Use Policy, or applicable law;
  • The Agent does not collect, transmit, or otherwise process personal data of recipients in a manner that violates applicable privacy laws.

You are solely responsible for the Agent Materials and the behavior of your Agents during execution.

11.4 Sharing Agents with Other Users

You may share Agents with specific users via email invitation, share link, or workspace membership. When you share an Agent:

  • Recipients receive access to a specific, immutable version of the Agent at the time the share was created or accepted ("Pinned Version"). Recipients are not automatically subject to subsequent modifications you make.
  • Each recipient must explicitly accept the Agent before it becomes part of their available agents. Sharing alone does not impose use on recipients.
  • You may revoke a share at any time. Revocation prevents new recipients from accepting the share but does not retroactively remove access from recipients who have already accepted, since they hold a valid license to the Pinned Version they accepted.
  • New versions of the Agent are not automatically applied to recipients' Pinned Versions. Recipients are notified that an update is available and may opt in.

11.5 Restrictions

You agree NOT to create, share, or publish an Agent that:

  • Reproduces, adapts, or substantially derives from third-party content, code, prompts, or characters that you do not have license to use, including copyrighted creative works, registered trademarks, or proprietary corporate IP;
  • Impersonates a real person or entity without consent, or is designed to deceive recipients about its origin, authorship, or capabilities;
  • Bypasses content filters, safety controls, rate limits, or billing on the Platform or any integrated third-party service;
  • Contains malware, exploits, prompt-injection payloads aimed at the Platform itself, or other malicious content;
  • Targets credentials, payment data, or other sensitive information from recipients;
  • Violates the Acceptable Use Policy or any applicable law in jurisdictions where it is shared or executed.

We may suspend, remove, or restrict execution of any Agent that we reasonably believe violates these Terms, with or without notice. We may cooperate with law enforcement and rights holders responding to credible claims.

11.6 Indemnification of Mfini for User-Authored Agents

In addition to the general indemnification provisions in Section 13, you agree to defend, indemnify, and hold harmless Mfini Inc., its officers, employees, and agents from any third-party claims, demands, damages, liabilities, costs, or expenses (including reasonable attorneys' fees) arising out of or related to:

  • The Agent Materials you create, share, or publish;
  • Any infringement, misappropriation, or violation of third-party rights by your Agents;
  • Any breach of your representations and warranties in Section 11.3;
  • Any harm to recipients, third parties, or external services caused by execution of your Agent.

11.7 Marketplace Publishing — Additional Terms

Publishing an Agent to the Versatl marketplace makes it available to all users of the Platform and may involve revenue collection on your behalf. Marketplace publishing is subject to additional terms in the Marketplace Publisher Agreement, including but not limited to: pricing, payout obligations, content review, refund handling, and tax responsibilities. You must accept the Marketplace Publisher Agreement before publishing. Publishing a paid listing additionally requires completing payout-account onboarding, including third-party identity verification through our payment processor; free listings do not.

The Marketplace Publisher Agreement applies in addition to these Terms. Where a conflict exists between this Section 11 and the Marketplace Publisher Agreement, the Marketplace Publisher Agreement controls with respect to marketplace activity.

12. Disclaimers

12.1 "As Is" Service

THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, OR AVAILABILITY.

12.2 No Uptime Guarantee

We do not guarantee uninterrupted or error-free operation of the Service. We may perform maintenance, updates, or modifications that temporarily affect availability.

12.3 Third-Party Services

We are not responsible for the availability, accuracy, or performance of third-party services (Stripe, Clerk, Anthropic, OpenAI, Google, OpenRouter, Composio, Twitter/X, LinkedIn, Instagram, Gmail, Google Calendar, Google Drive, Brave Search, and other integrated providers) integrated with the Platform.

13. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY LAW, MFINI INC. SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES, INCLUDING BUT NOT LIMITED TO LOSS OF PROFITS, DATA, REVENUE, OR BUSINESS OPPORTUNITIES, ARISING FROM YOUR USE OF THE SERVICE.

MFINI INC.'S TOTAL LIABILITY FOR ANY CLAIM ARISING FROM THESE TERMS OR YOUR USE OF THE SERVICE SHALL NOT EXCEED THE AMOUNT YOU PAID US IN THE 12 MONTHS PRECEDING THE CLAIM, OR $100, WHICHEVER IS GREATER.

14. Indemnification

You agree to indemnify, defend, and hold harmless Mfini Inc. and its officers, directors, employees, and agents from any claims, damages, losses, or expenses (including reasonable attorneys' fees) arising from: (a) your use of the Service; (b) your User Content; (c) your violation of these Terms; (d) your violation of any third-party rights; or (e) actions taken by agents on your behalf through connected services.

15. Account Termination

15.1 By You

You may delete your account at any time through the Settings page or by calling DELETE /auth/me. Account deletion is permanent and triggers cascade deletion of all your data, including tasks, services, episodes, memories, credentials, API keys, notifications, and webhook configurations. De-identified billing records are retained as described in the Privacy Policy.

If your account balance is negative at the time of deletion, small negative balances (currently up to $5) are written off and deletion proceeds; a negative balance beyond that amount must be settled by adding funds before the account can be deleted. Any unspent positive account balance at the time of deletion — including both paid funds and promotional credit — is forfeited under Section 5.5 (Refunds and Disputes), which makes account balances non-refundable. Before deletion is submitted, the Delete Account dialog displays your current balance (with the paid-funds and promotional-credit components broken out when a signup grant is active) and requires you to acknowledge the forfeiture in addition to typing the confirmation phrase. You may use up your balance before deleting; unspent balance remains available indefinitely until you use it or delete your account.

15.2 By Us

We may suspend or terminate your account if you violate these Terms, the Acceptable Use Policy, or any applicable law. We will make reasonable efforts to notify you before termination, except in cases of severe or repeated violations.

15.3 Effect of Termination

Upon termination, your right to use the Service ceases immediately. Data deletion follows the same cascade process as voluntary account deletion. Certain data may be retained as required by law (e.g., billing records, audit logs). Any unspent account balance is forfeited under Section 5.5, regardless of whether termination is initiated by you or by us.

16. Modifications to Terms

We may update these Terms from time to time. Material changes will be communicated via email or in-app notification at least 30 days before taking effect. Your continued use of the Service after changes take effect constitutes acceptance of the updated Terms.

The same process applies to changes in the pricing incorporated by reference into these Terms (including the Platform Fee rate on the Pricing Page, Section 5.1): any increase will be communicated via email or in-app notification at least 30 days before taking effect and applies prospectively only — usage before the effective date is charged at the prior rate. Rate decreases may take effect immediately. The Pricing Page always reflects the rate currently in effect. If you do not accept a fee increase, you may stop using paid features or spend down your remaining balance before the effective date; usage before that date is charged at the prior rate (account balances remain non-refundable as described in Section 5.5).

17. Governing Law and Disputes

These Terms are governed by the laws of the State of Delaware, United States, without regard to conflict of law principles. Any disputes arising from these Terms shall be resolved through binding arbitration administered by the American Arbitration Association (AAA), except that either party may seek injunctive relief in a court of competent jurisdiction.

18. Miscellaneous

  • Severability: If any provision of these Terms is found unenforceable, the remaining provisions remain in effect.
  • Waiver: Failure to enforce any provision does not constitute a waiver of that provision.
  • Assignment: You may not assign your rights under these Terms without our written consent. We may assign our rights freely.
  • Entire Agreement: These Terms, together with the Privacy Policy, Acceptable Use Policy, the Connected Services Appendix (incorporated by reference under Sections 2 and 7.1), and (if applicable) Marketplace Publisher Agreement and Data Processing Agreement, constitute the entire agreement between you and Mfini Inc.

19. Contact

For questions about these Terms, contact us at:

Mfini Inc.

Email: legal@versatl.ai

Address: 1401 Lavaca St, Ste 558, Austin, TX 78701