Terms of Service

Last published September 14, 2026

Business Terms of Service · version 7

Business Terms of Service

Effective date: September 11, 2026 · Version: 2

These Terms govern your business's use of the Locally AI platform. They are between Locally AI ("Locally AI", "we") and the business identified in your account ("you", "your business"). By signing, you confirm you are authorized to bind your business.

Your Partner is the independent contractor who introduced you to Locally AI, configures your agent, and is your primary point of contact. Your agreement for the service is with Locally AI, not with your Partner — we provide the service, we invoice you, and we are responsible to you under these Terms. Your Partner is not our employee and cannot make commitments, agree prices, or vary these Terms on our behalf. If your Partner also provides you with separate services they bill you for directly, that arrangement is between you and them.


1. What the service does

1.1 Locally AI provides an AI agent dedicated to your business. The agent is configured with a written playbook about your business and can be connected — only with your authorization — to third-party applications you already use, such as your email, calendar, spreadsheets, CRM, or accounting software.

1.2 The agent does two kinds of billable work:

(a) Saved workflows — defined sequences of steps that produce a business outcome, such as replying to enquiries, logging leads, or preparing reports. These run on a schedule, when you ask in chat, or when your Partner triggers them.

(b) Tasks you ask for in conversation — one-off work you describe to the agent in chat. These are priced and recorded the same way as a workflow.

Once a conversational task has completed successfully, you may save it as a reusable workflow of your own. Saving it shows you the price it will run at, and from then on it runs at that price whenever you ask for it by name, without quoting you again each time. You can ask us or your Partner to remove a saved workflow at any time.

1.3 Talking to the agent is free. Work inside your connected applications is not. Questions, explanations and planning cost nothing. Once the agent reads from or writes to one of your connected applications on your behalf, that is billable work — reading included, because gathering and organising information from your applications is the work you are paying for.

1.4 The agent is software, not a person. It is not an employee, agent-at-law, professional advisor, or licensed practitioner.

2. Your authorization

2.1 By connecting an application, you authorize the agent to access and act within that application on your behalf, within the scopes you grant, until you revoke access.

2.2 You may revoke any connection at any time from the Connected Apps page in your account. Revocation takes effect immediately for future actions. It cannot undo actions already taken.

2.3 You can see, at any time, which applications are connected, what access was granted, who connected them, and when they were last used.

2.4 We never ask for your passwords. All access is granted through each provider's own sign-in and consent flow.

3. Human oversight — your responsibility

3.1 You are responsible for reviewing what the agent does on your behalf. Run transcripts and outcomes are available in your account.

3.2 You should not rely on the agent for activity requiring professional judgement, legal or regulatory compliance decisions, medical or financial advice, or anything where an error would cause serious harm.

3.3 If a workflow can send communications to your customers or publish publicly, you are authorizing those communications. Tell your Partner to disable any workflow you do not want running.

3.4 You are responsible for ensuring communications sent on your behalf comply with law, including CAN-SPAM, the TCPA, and California equivalents. You confirm you have the necessary consents for any customer you ask the agent to contact.

4. Fees and billing

4.1 You pay only for completed outcomes. Every billable task, whether a saved workflow or one you asked for in conversation, is assessed against success criteria recorded before the work begins. If the outcome is not achieved, you are not charged.

4.2 Prices depend on the complexity of the work. For a saved workflow, the price is shown in your account before the workflow is enabled. For a task you ask for in conversation, the agent tells you what it will do and what it costs before it starts, and that quoted price is what you pay — it is never increased afterwards, even if the work turns out to be larger than expected. Current prices are visible at any time under Workflows.

4.2.1 How conversational tasks are authorized is your choice. Your account is set, by default, so that the agent quotes each task and waits for one of your contacts to approve it before doing anything. You may instead choose to let the agent price and carry out tasks without asking first. That choice is shown on the agent page, can be changed at any time by any of your contacts, and we record who changed it and when. Choosing not to be asked is a standing authorization to charge you for conversational tasks the agent completes, subject to everything else in this section — including that failed work is never charged and that your spend cap still applies.

4.2.2 Any of your contacts may approve a task or change the setting in 4.2.1. It is your responsibility to decide who you invite as a contact.

4.3 Charges accrue during the month and are invoiced monthly to the payment method on file. If the total is below the minimum invoice amount, it carries to the next month.

4.4 Spend cap. You may set a maximum monthly amount. Once reached, billable work pauses until the next month or until you raise the cap, and the agent will say so rather than starting a task it cannot bill. If the cap is reached while a task is already underway, we finish it and do not charge you for it. Using a cap is your responsibility; we do not set one by default.

4.5 Disputes. You may dispute any charge within 30 days from that item's page in your account. We will review the transcript, the outcome assessment, and — for a conversational task — the quote and how it was authorized. If we agree the outcome was not achieved, the charge is credited to your next invoice. Our determination is final for the purposes of these Terms, without prejudice to your rights under law.

4.6 Non-payment. If payment fails repeatedly, scheduled workflows are paused and the agent will not start new billable tasks. You can still talk to it — questions and explanations remain available. We may suspend the account for continued non-payment.

4.7 AI compute. If we enable it for your account, the cost of the AI processing used by your work may appear as a separate "AI compute" line on your invoice, at cost plus any stated margin. Where this applies it is shown with current prices under Workflows before it is charged.

4.8 Taxes are added where required. Prices are exclusive of tax unless stated.

5. Your data

5.1 You own your data. You retain all rights to the data in your connected applications and anything the agent reads or produces for you.

5.2 We process your data to operate the service: running workflows, producing outcomes, supporting you, billing, and securing the Platform. Exhibit A sets out the data processing terms.

5.3 Your Partner can see your account, including agent configuration, workflow definitions, run transcripts, and analytics. That is how they manage your service. If you do not want a Partner to have this visibility, do not use the service through a Partner.

5.4 AI processing. Your data is sent to our AI model provider to generate agent output. We do not permit your data to be used to train third-party models. See Exhibit A.

5.5 Retention. Run transcripts and chat history are retained for 90 days by default and then deleted. Billing records are retained as required by law.

5.6 On termination you may request export of your data within 30 days, after which we may delete it.

6. Acceptable use

You will not use the service to: send unsolicited bulk messages; publish false, deceptive, defamatory, or misleading content; impersonate a person; violate the terms of a connected application; infringe intellectual property; or conduct activity that is unlawful or that we reasonably determine creates risk. You will not present agent output as reviewed by a licensed professional when it was not.

7. Availability

We aim for continuous availability but do not guarantee it. The service depends on third parties, including AI model providers and the applications you connect. Maintenance, outages, and provider changes can interrupt service. No uptime commitment or service level agreement is offered.

8. Disclaimers

8.1 THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE". To the maximum extent permitted by law, we disclaim all warranties, express or implied, including merchantability, fitness for a particular purpose, and non-infringement.

8.2 AI OUTPUT MAY BE INACCURATE, INCOMPLETE, OR INAPPROPRIATE. You should not treat it as professional advice or as a substitute for your own judgement.

8.3 We do not guarantee business results — not more customers, not more revenue, not improved ratings. Any time-savings or outcome figures shown in your analytics are estimates.

8.4 That a task is assessed as "achieved" means it satisfied the criteria recorded before the work began. It is not a warranty that the work was ideal or suited to your circumstances. For a conversational task those criteria are written by the agent when it quotes the work; you can see them on the quote and in the task record, and dispute the charge under 4.5 if the result does not match what you asked for.

9. Limitation of liability

9.1 TO THE MAXIMUM EXTENT PERMITTED BY LAW, WE ARE NOT LIABLE FOR INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, EXEMPLARY, OR PUNITIVE DAMAGES, OR FOR LOST PROFITS, REVENUE, DATA, GOODWILL, OR REPUTATIONAL HARM.

9.2 OUR TOTAL AGGREGATE LIABILITY arising out of or relating to these Terms will not exceed the greater of (a) the amounts you paid us in the 12 months before the event giving rise to the claim, or (b) $100.

9.3 These limits do not apply to liability that cannot be limited under California law.

10. Indemnification

You will defend and indemnify us against claims arising from: your data or content; communications sent on your behalf; your violation of law or of a connected application's terms; and your breach of these Terms — except to the extent caused by our breach.

11. Term and termination

11.1 These Terms continue until terminated. You may terminate at any time; you remain liable for outcomes already completed and not yet invoiced.

11.2 We may suspend or terminate for breach, non-payment, or legal or security risk, with notice where practicable.

11.3 If your Partner's relationship with Locally AI ends, your service continues; account management transfers to Locally AI or another Partner. We will notify you.

11.4 Sections 5.1, 5.6, 8, 9, 10 and 13 survive termination.

12. Changes

We may update these Terms on notice. Material changes require your acceptance before continued use. Price changes apply to runs occurring after the effective date.

13. Governing law and disputes

13.1 Governed by the laws of the State of California, without regard to conflict-of-laws rules.

13.2 Any dispute arising out of or relating to these Terms will be brought exclusively in the state or federal courts located in California, and each party consents to the personal jurisdiction of those courts and waives any objection to venue. Nothing in this section prevents either party from bringing an individual claim in small claims court.

14. General

Entire agreement (together with the Business Liability Protection Agreement); no assignment by you without our consent; severability; no waiver; notices to the email on your account; force majeure.


Exhibit A — Data Processing Terms

A.1 Roles. For personal information in your connected applications, you are the business/controller and Locally AI is the service provider/processor. Your Partner is a sub-processor with access limited to your account.

A.2 Instructions. We process personal information only to provide the service, as instructed through your configuration and these Terms, and as required by law. We do not sell or share personal information, and do not retain, use, or disclose it for any purpose other than providing the service, as those terms are defined under the CCPA/CPRA.

A.3 Categories. Depending on what you connect, processing may include: customer names, email addresses, phone numbers, review content, message content, appointment details, and CRM records.

A.4 Sub-processors. Current sub-processors are available on request and include our cloud host, database provider, AI model provider, integration provider, payment processor, and email provider. We will give notice before adding a sub-processor that materially changes processing.

A.5 AI model processing. Agent inputs and outputs are sent to our AI model provider to generate responses. We contract that your data is not used to train their models and is retained only transiently for abuse monitoring per that provider's terms.

A.6 Security. We maintain reasonable and appropriate technical and organizational measures, including encryption in transit, access controls, tenant isolation enforced at the database layer, and scoped OAuth tokens rather than stored credentials.

A.7 Breach notification. We will notify you without undue delay after becoming aware of a personal data breach affecting your data, with the information reasonably available to us.

A.8 Assistance. We will provide reasonable assistance with data-subject requests, impact assessments, and regulator enquiries.

A.9 Deletion. On termination we delete or return personal information within 30 days, except as required by law.

A.10 Audit. On reasonable notice and no more than once a year, we will provide information reasonably necessary to demonstrate compliance with this Exhibit.

Business Liability Protection Agreement · version 4

Business Liability Protection Agreement

Effective date: September 11, 2026 · Version: 2

This Agreement supplements the Business Terms of Service between Locally AI and your business. It explains, in plain terms, what an autonomous AI agent will do on your behalf and who is responsible for what. Where this Agreement and the Business Terms of Service conflict on risk allocation, this Agreement controls.


1. What you are authorizing

By completing setup, you authorize an AI agent to act on your business's behalf inside the applications you connect. Depending on your configuration, this may include:

  • Reading email, reviews, calendar entries, spreadsheets, and CRM records
  • Writing records, updating spreadsheets, creating calendar entries
  • Publishing replies to public reviews and posts on your business listing
  • Sending email or text messages to your customers
  • Notifying your team in chat tools

The agent acts automatically, on a schedule or on request, without a person approving each action — unless a workflow is specifically configured to ask first.

2. What you should understand about AI

Please read this section carefully. By signing, you confirm you understand it.

2.1 AI can be wrong. The agent may produce content that is inaccurate, made up, off-tone, or inappropriate — sometimes while appearing confident and reasonable.

2.2 Some actions cannot be undone. A published review reply, a sent email, or a text message to a customer may not be retractable. Public content can be seen and captured before removal.

2.3 The agent reads content other people wrote. Reviews, emails, and documents can contain text designed to manipulate an AI system into behaving unexpectedly. We take steps to prevent this, but no protection is complete.

2.4 Outcome assessment is automated. Whether work "achieved its outcome" is decided by an automated evaluation, which can be wrong. For a task you ask for in conversation, the agent also writes the success criteria itself when it quotes the work — you see them before approving, and you can dispute any charge.

2.5 The agent is not a professional. It is not a lawyer, accountant, doctor, or licensed advisor, and its output is not professional advice.

3. Your responsibilities

3.1 Review what goes out. Check run transcripts regularly, especially for workflows that contact customers or publish publicly.

3.2 Set boundaries. Tell your Partner what the agent must never do — topics to avoid, claims it must not make, customers it must not contact, discounts it may not offer. These become guardrails in your playbook. Boundaries you do not communicate cannot be enforced.

3.3 Consent for outreach. You confirm you have the consents required by law to contact the customers you ask the agent to contact.

3.4 Accurate information. Your playbook drives the agent's behaviour. If it contains wrong prices, hours, or policies, the agent will repeat them.

3.5 Act on problems. If you see incorrect or inappropriate output, tell your Partner or us promptly at james@realisinvestments.com and disable the workflow. Continuing to run a workflow you know is producing bad output is your responsibility.

3.6 Use a spend cap if you want a hard ceiling on monthly charges.

4. What we are responsible for

4.1 Operating the Platform with reasonable care and skill, and maintaining commercially reasonable security.

4.2 Enforcing the guardrails configured in your playbook and workflows as documented.

4.3 Not charging you for work that does not achieve its defined outcome, whether a saved workflow or a task you asked for in conversation.

4.4 Making run transcripts and outcome assessments available so you can see what the agent did.

4.5 Giving you one-click revocation of any connected application.

5. What we are not responsible for

Subject to Section 7, we are not responsible for:

5.1 The content of agent output, where the agent operated within your configured playbook and guardrails.

5.2 Actions taken by workflows you or your Partner authorized to act without confirmation.

5.3 Business consequences of agent activity — lost customers, reputational effects, or missed opportunities — absent our breach of Section 4.

5.4 Errors originating in information you supplied.

5.5 Failures, changes, or outages of the third-party applications you connect.

6. Your protections

6.1 You are never charged for failed work, and never more than you were quoted. You are also never charged for a conversational task you did not authorize — either by approving it, or by choosing to let the agent charge without asking. That choice is yours to change at any time, and choosing it does not waive anything else in this section.

6.2 You can dispute any charge within 30 days; we review the transcript and credit it if the outcome was not achieved.

6.3 You can revoke access instantly to any connected application.

6.4 You can cap your monthly spend.

6.5 You can cancel at any time without penalty.

6.6 If we breach Section 4 and you suffer loss as a result, the Business Terms of Service govern our liability.

7. Limits that do not apply

Nothing in this Agreement limits our liability for gross negligence, wilful misconduct, fraud, or any liability that cannot be limited under California law, including, where applicable, liability for personal injury.

8. Indemnification

8.1 You indemnify us against third-party claims arising from communications sent or content published on your behalf where the agent operated within your configuration, your breach of these terms, or your violation of law — except to the extent caused by our breach of Section 4.

8.2 We indemnify you against third-party claims that the Platform, used as permitted, infringes a US intellectual property right.

9. Your acknowledgment

By signing, you confirm that:

  • You have read and understood Section 2 and accept that AI output can be wrong and some actions cannot be undone;
  • You authorize the agent to act in the applications you connect, including automatically;
  • You accept the responsibilities in Section 3;
  • You are authorized to bind your business.

If you are not comfortable with an AI system acting automatically on your behalf, do not sign this Agreement. Ask your Partner to configure workflows that require confirmation before acting, or not to enable outbound communication workflows at all.