Memory-Native Apps Additional Terms

These Additional Terms apply when your organization uses Memory-Native Apps, the App store in your Achiral AI workspace. Achiral AI is operated by Swizzle Inc. We, us, and our have the meaning given in the Terms of Service.

These Additional Terms are part of the Terms of Service. Terms defined there, such as Service and Customer Content, mean the same here. The Data Processing Policy, Privacy Policy, Billing & Refunds Policy, and Usage Policy also apply to the add-on.

Memory-Native Apps is a paid add-on that lets members of your organization build apps on your organization's memory and connected sources. Each app is defined by a specification: the sources it reads, the entities and signals it tracks, the formulas, stages, and scores it computes, and the views it shows.

The add-on fee covers the Memory-Native Apps workspace, app specifications, projections, the app runtime, version history, and governed app actions. Projections are the scores, stages, and other results an app computes. Apps run on infrastructure hosted by Achiral AI.

Model inference is not included in the fee. Narratives run on the model provider your organization connects, and your provider bills you for that use. Section 9 explains this and the fallback to models Achiral AI hosts.

The add-on costs US$19 per published app per month, plus applicable taxes. Building apps, previewing them, and keeping drafts costs nothing. An app is billed while it is published, in whichever environment it is published, and an archived app is not billed. The price for an app is the same however many members open it or how often it runs, subject to the limits in Section 9.

The add-on renews every month until an Owner or Admin cancels it. We add it as one item on your organization's existing subscription, with a quantity equal to the number of published apps, and charge it on the same invoices. We do not prorate changes to that number. An app published partway through a billing period is first charged at the next renewal. An app archived partway through a period stops being charged at the next renewal, with no refund or credit for the rest of that period. If you activate the add-on while apps are already published, the first charge may be prorated for the rest of that period. If you turn a canceled add-on back on before its paid period ends, there is no extra charge; billing resumes at the next renewal.

If your organization is on a Custom plan, your order form sets what you pay for the add-on, and we do not add a separate monthly charge.

Only an Owner or Admin of your organization can activate the add-on, cancel it, or accept changes to these Additional Terms. The person who does so confirms that they have authority to bind the organization.

Activation requires a paid Achiral AI plan or a Custom plan. Other members cannot activate the add-on. Until it is active, they are asked to contact an Owner or Admin.

Activation uses one checkbox. It starts unchecked, and the activation button stays disabled until it is checked. The checkbox reads: "I agree to add Memory-Native Apps for $19 per published app per month plus applicable taxes, renewing monthly until canceled by an Owner/Admin, and I accept the Memory-Native Apps Additional Terms, Data Processing Policy, Privacy Policy, Billing & Refunds Policy, and any organization agreement that applies."

Checking the box and activating does two things. Your organization agrees to pay for the add-on as described in Section 3. Your organization also accepts these Additional Terms, the Data Processing Policy, the Privacy Policy, the Billing & Refunds Policy, and any written agreement it has with us.

We accept only the current text. If the checkbox or price disclosure you saw is out of date, activation fails, and you need to review and accept the current text.

Each acceptance creates a record. It contains the organization, the accepting user and their role, the date and time, the price, currency, and billing cadence, the versions of these Additional Terms and the three policies, a fingerprint of the exact checkbox and price disclosure text shown, billing identifiers, a keyed hash of the IP address, and the browser's user agent. We do not store the IP address itself.

We never edit an acceptance record. A later acceptance creates a new record. We use these records to show who agreed to what and when, and to answer billing questions. The Privacy Policy explains how we handle the personal information in them.

Your organization owns its apps, specifications, versions, projections, generated narratives, and the data its apps read and produce. All of it is Customer Content under the Terms of Service. Apps belong to the organization, never to the member who built them.

You grant us a limited, non-exclusive license to host, copy, process, run, and display your apps and their data, only as needed to provide the add-on to your organization, keep it secure, support you, and comply with law.

We do not use your apps or app data to train or fine-tune AI models, as the Terms of Service state.

We keep all rights in the Service, including the app runtime, the specification format, the app components, and the builder.

Any active member of your organization can create an app, edit it, and preview it. The member who creates an app is its steward. Owners and Admins can edit any app in the organization.

If a steward leaves or should stop managing an app, an Owner or Admin can make another active member the steward. The app, its versions, and its history stay with the organization.

Each app has one of two audiences. A by-invitation app is visible to its steward, to Owners and Admins, and to the members invited to it. An app for everyone is visible to every member of the organization. In both cases, other members see an app only once it is published.

Publishing an app for the first time starts its monthly charge, so it needs an Owner or Admin. They can publish it themselves, or the steward can submit one revision for approval, and approval publishes exactly the revision they reviewed. Later releases of a published app do not change the charge and need no approval. The steward, Owners, and Admins can change an app's audience and its invitations at any time.

Apps are never public. No one outside your organization can view your apps, and no setting makes an app public. A public app would need a separate product and separate terms.

An app shows its whole audience the data from the sources it uses. Before opening an app to everyone or inviting a member to it, make sure they may see that data. Your organization decides who holds the Owner and Admin roles and is responsible for the apps its members build.

Apps show two kinds of results. Some come from data recorded in your source systems and the formulas in the app's specification. Others are produced by AI: narratives written by Chiro, and conclusions inferred from your data.

To write a narrative, Chiro reads a limited set of evidence records and cites them. Chiro is instructed to use only that evidence, but a narrative can still be wrong, incomplete, or out of date. Check the cited evidence before you rely on a narrative. An app reuses a stored narrative only while its evidence is unchanged. If Chiro cannot write a new narrative, the app says so and does not show an older narrative as current.

An inference is a conclusion drawn by AI rather than recorded in a source system. Each inference has a confidence score. An inference counts toward a published score or stage only if its confidence meets your organization's inference threshold.

Your organization's Owner chooses the threshold from three presets: Strict (0.9), Balanced (0.8, the default), and Permissive (0.7). No setting or app can count inferences below 0.7. An app can raise the threshold for a factor but cannot lower it.

Inferences below the threshold do not count, and apps do not present them as fact. They wait for review and count only after an Owner or Admin approves them.

Confidence scores are not probabilities. A confidence of 0.8 does not mean a conclusion is right 80% of the time.

Scores and stages follow the formulas, weights, and rules your organization writes into its apps. We run that logic as written and do not check whether it suits your purpose.

Your organization makes its own decisions. Nothing an app shows is legal, financial, medical, tax, employment, or other professional advice. Do not use an app as the only basis for a decision that significantly affects a person.

Chiro writes narratives through the model provider your organization connects, using your organization's own account with that provider. Your agreement with that provider governs how it processes the evidence we send it. If your organization has not connected a provider, or its provider fails, Chiro uses models Achiral AI hosts, unless your organization has turned that fallback off.

Use of models Achiral AI hosts is not part of the add-on fee. It has an allowance and rate limits, and any charge for it is disclosed and billed separately under your plan. When the allowance or a limit is reached, apps stop writing new narratives and say so.

Each app run has server-side limits on queries, records returned, retrieval depth, run time, and narrative generations. An app specification cannot raise them. We may change these limits to protect the Service. If a change materially reduces what the add-on does, we will treat it as a material change under Section 14.

In this phase, an app can offer two actions: correcting an organization memory, and recording a decision, approval, or rejection. Only Owners and Admins can take them.

Each action is recorded with the user who took it, the app, the app version, and the app run it came from. The memory it writes carries the same attribution. Actions follow your organization's approval settings, so some may wait for approval before they take effect.

When an Owner or Admin approves an inference below the threshold, it counts as an approved fact. Your organization is responsible for the actions its users take through apps.

Apps cannot run arbitrary code or take actions in outside systems. Every app runs only inside its own organization and environment.

The Usage Policy applies to apps, including its rules for Memory-Native Apps. Your organization is responsible for the specifications its members write, the sources they connect to apps, and the decisions it makes using apps.

If a payment on your organization's subscription fails, apps lock until the payment succeeds. While apps are locked, no one can create, revise, publish, or run apps, regenerate narratives, or take actions through apps. An Owner or Admin cannot reactivate the add-on while payment is past due. Your organization still owes unpaid amounts.

An Owner or Admin can cancel the add-on at any time in your workspace's App store settings. We remove the add-on from your subscription straight away, so it does not renew. Apps keep working until the end of the billing period you have already paid for, then stop. We do not give a refund or credit for the rest of that period. If the add-on is past due when you cancel, or your organization is on a Custom plan, apps stop when you cancel.

The add-on also ends if your organization's subscription ends or the add-on is removed from the subscription.

We do not delete your apps when the add-on ends. Specifications, versions, release history, and action and audit records stay with your organization under the Terms of Service and Data Processing Policy. Owners, Admins, and stewards can still view and archive them. App run records and stored narratives are deleted after 90 days, as usual. To ask us to delete app data sooner, an Owner or Admin can write to [email protected].

To use apps again, an Owner or Admin must activate the add-on again and accept the price and terms current at that time.

Section 13 of the Terms of Service applies to the add-on. Under it, we may suspend the add-on or a single app if we reasonably believe it breaks these Additional Terms or the Usage Policy, or creates a security, legal, or abuse risk. Where we can, we will tell an Owner or Admin first, limit the suspension to what is needed, and lift it once the problem is fixed.

We may change these Additional Terms. The date at the top of this page shows the current version.

A material change to these Additional Terms, the add-on price, or the billing cadence applies to your organization only after an Owner or Admin accepts it, and only from the next renewal after that. Until then, apps keep working on the terms and price you last accepted, and Owners and Admins are asked to review the new terms.

If no Owner or Admin accepts the change before the next renewal, the add-on does not renew. We remove it from your subscription so nothing more is charged, and apps stop when the period you have already paid for ends. To use apps again, an Owner or Admin activates the add-on and accepts the current terms.

Changes that are not material, such as clarifications, take effect when we post them. Changes to the Data Processing Policy, Privacy Policy, and Billing & Refunds Policy follow the rules in those documents.

We plan to offer a marketplace for app templates. It is not available yet. We will publish separate terms before it launches.

We will not place any of your apps or data in a marketplace unless an Owner or Admin chooses to publish a template under those future terms.

Memory-Native Apps is a new product, and its features will change. Features we label as preview or beta in the product follow Section 11 of the Terms of Service.

The Disclaimers and Limitation of Liability sections of the Terms of Service apply to the add-on. In addition, we do not promise that narratives, inferences, scores, or stages are accurate, complete, or fit for any purpose, or that an app will find every relevant record. Nothing here limits a warranty or liability that the law does not allow us to limit.

Add-on fees count toward the amounts paid under the Limitation of Liability section. These Additional Terms do not create a separate liability cap or expand either party's liability.

We are not responsible for decisions your organization makes using apps, or for losses caused by app logic your organization defines, except to the extent our breach of these Additional Terms or the Terms of Service caused them.

Section 21 of the Terms of Service governs indemnity for the add-on. These Additional Terms add no separate indemnity.

Under that Section, your organization defends us against third-party claims arising from data your organization or its users connect to apps, and from use of apps in breach of the Usage Policy. Its exclusions, its procedure, and its rules on our defense of your organization apply to apps as they apply to the rest of the Service.

If documents conflict about the add-on, they apply in this order: first, a signed order form or other written agreement between your organization and Swizzle Inc.; second, these Additional Terms; third, the Terms of Service; and fourth, the Data Processing Policy, Privacy Policy, Billing & Refunds Policy, and Usage Policy.

Anything these Additional Terms do not cover follows the Terms of Service. That includes confidentiality, feedback, suspension, California governing law, dispute resolution and arbitration, and legal notices.

Questions about these Additional Terms may be sent to [email protected] or to Swizzle Inc., Attn: Legal, 1908 Thomes Ave STE 71455, Cheyenne, WY 82001. Billing questions may be sent to [email protected]. Privacy and deletion requests may be sent to [email protected].