1. Acceptance and Parties
Business service. Tejma Dispatcher is a business tool offered to companies and professionals (each organization, a "Customer"). The person who founds an organization, or accepts these Terms on its behalf, represents that they have authority to bind that organization. Each member of an organization (a "seat" or "user") is also bound by these Terms when they sign in.
Recorded acceptance. Every user must accept the current version of these Terms and the Privacy Policy in the service before using it; the dashboard, the API, and the MCP endpoint are closed to a user who has not. Acceptance is recorded per version, with the time it was given, under your user identifier, and is kept as a consent record even after you leave an organization or your account is deleted. No seat is provisioned or billed for a user until they accept.
Organization roles. Authority within an organization is carried by three roles the service defines. The billing role manages seat plans, the subscription, the prepaid credit balance, the overage option, and any Anthropic API keys the organization or its members supply. The usage role allows a member to start and shape AI work — submitting messages, approving and retrying tasks, creating and editing agents, skills, approval workflows, and channels; the service does not process work belonging to a member without it, so revoking it halts that member's AI usage at once. The users role manages members: invitations, deactivation, session revocation, password resets, grants of Tejma connectors, and grants of these roles themselves. The member who founds an organization holds all three; later members hold none until a users-role holder grants them. Every member, with or without a role, can read their own data, stop or delete their own work, and manage the connectors they register. Each organization must keep at least one users-role holder. In these Terms, "your organization's administrators" means the holders of these roles, and their actions — inviting members, selecting plans, buying credit, switching billing modes, granting connectors, cancelling a subscription — bind the Customer.
Changes to these Terms. We may update these Terms from time to time. A new version is published at this address with its effective date at the top, and the service asks every user to accept it before continuing. The date at the top of this document reflects the latest revision.
2. The Service
What the service does. Tejma Dispatcher receives messages from sources you connect or submit through — email mailboxes, Slack channels, Discord channels, MCP (Model Context Protocol) clients, the dashboard, and our API — and stores them, together with their attachments and files, in your account. AI workers then split each message into individual tasks, route each task to one of your configured agents, and, once a task is approved, execute it automatically using that agent's skill instructions, the connectors wired to it, and any outbound channels you have linked to the agent.
You direct the automation. Agents, their skill instructions, your extraction and routing skills, connectors, outbound channel links, and approval workflows are created and configured by you. You may start an agent from a template in the library Tejma publishes; a template is a starting point copied into your account for you to review and edit, and the resulting agent is yours and your responsibility like any other. Tasks execute only after approval — by one of your members, or by an approval workflow you defined — and tasks routed to the built-in User Input agent always wait for a human. You are responsible for what your agents are instructed to do and for the tools and channels you give them access to. An agent wired to a connector that is not yet connected, whose grant has been revoked, or that is parked by the seat's plan (Section 5) is itself parked: it is not routed tasks, does not execute, and cannot act as a judge, until the connector is usable again or unwired.
Connectors. A connector is a remote MCP server whose tools an agent may call. There are two kinds. Connectors you register are servers you choose and authenticate yourself, by URL, with a token or an OAuth sign-in; they are third-party services under Section 10, and nothing limits how many you register. Tejma connectors are Tejma's own connector services, offered in the catalog inside the service (for example Tejma Books). A Tejma connector is installed on a member's seat only after a users-role holder has granted that member the connector, with one or more of the roles the connector defines (such as a read-only or a full role); the service passes those roles to the connector, which decides what they allow. Revoking the grant leaves the install in place but unusable until it is granted again. Registering by URL a server on a Tejma connector's host is treated as installing that Tejma connector. Each Tejma connector is a Tejma service with its own terms of service and privacy policy, which apply to your use of it alongside these Terms, and Tejma connectors count toward the seat plan's connector allowance as described in Section 5.
Outbound messages. Where you link outbound channels to an agent, the service can send messages on your behalf through those channels (for example replying in a Slack thread or sending an email through a connected mailbox), either when the agent decides to or as a completion report at the end of every task, as you configure. A completion report carries a short model-written description of the task, the agent's message, and a link to the task in the dashboard. Depending on how you configure recipients, these messages — completion reports included — may reach people outside your organization, such as the other participants of an email thread. Messages sent this way are sent at your direction and are your responsibility, including compliance with anti-spam laws and the rules of the workspace or mailbox they are sent from. Every successful send is recorded in an audit trail available in your account.
Agent-created tasks. An agent you enable for parent tasks may pause a task it cannot finish and create prerequisite tasks on the same message. Those tasks are routed and reviewed like any other and each counts against your seat's quota on its own; the paused task resumes once they complete, and each resumed execution counts as a new execution.
Service evolution. We may improve, modify, or discontinue features of the service. If we discontinue the service entirely, we will inform your organization's administrators and honor the data rights described in the Privacy Policy.
3. Accounts and Organizations
Sign-in. Authentication is provided through WorkOS AuthKit, under the single WorkOS application every Tejma service shares, so a person signed in to Tejma Dispatcher is signed in to Tejma connectors with the same account. You are responsible for safeguarding your credentials and sessions; users-role holders can revoke member sessions, reset passwords, and deactivate and reactivate members at any time, and revoking a member's usage role halts their AI work immediately.
Founding an organization. A user who signs up without an invitation founds their own organization and receives all three roles. Founding requires no Anthropic API key: a new organization starts on managed AI billing (Section 4) and may receive a one-time founding credit (Section 6). An organization that prefers to pay Anthropic directly may supply its own key when it is founded or at any later time. Invitations are sent by a users-role holder and may carry the seat plan, roles, and Tejma connector grants the new member starts with; they take effect when the invitee first signs in.
Data isolation. Every message, task, agent, connector, channel, workflow, and document belongs to exactly one user. Organizations exist for billing and administration: members of the same organization do not see each other's content through the service, and identifiers belonging to another user behave as if they do not exist. Tejma's own access to stored content is described in the Privacy Policy.
Accurate information. You agree to provide accurate account and billing information and to keep it current. We may suspend accounts that provide false information or that we reasonably believe are being used fraudulently.
4. AI Processing and Billing Modes
Two ways to pay for AI. Every AI call the service makes on your behalf — task extraction, task routing, approval-workflow judgements, and task execution — runs on Claude models through Anthropic's API and is paid for in one of two modes. Your organization's billing-role holders choose the mode and may change it at any time; the change applies from the next call.
Managed AI billing (the default): your calls run on Anthropic API keys that Tejma holds, and each call is metered against your organization's prepaid credit balance at the per-model prices in Tejma's model catalog. Own-key billing: your calls run on an Anthropic API key your organization supplies, Anthropic bills your organization directly, and Tejma charges nothing for AI usage.
Your billing standing in the dashboard always shows which mode applies to your seat.
Managed AI billing
How a call is priced. Each managed call is priced from the token usage Anthropic's API reports for that call — input tokens, output tokens (which include the model's reasoning), prompt-cache writes, and prompt-cache reads — multiplied by the per-model prices for each kind of token in Tejma's model catalog. There is no separate surcharge or markup: the catalog price is the price. The result is debited from your organization's credit balance and recorded against the call. A call on a model that has no price in the catalog is not charged.
Tejma's catalog prices are Tejma's prices, not Anthropic's. What you pay for a managed call is not a pass-through of Anthropic's pricing: Tejma sets each model's catalog price, which may be above Anthropic's published price for the same model and may change at any time.
Anthropic has no role in setting Tejma's prices and receives no payment from you for managed calls; your payment relationship for managed usage is with Tejma alone.
Published prices. The current per-model prices are shown to your organization's billing-role holders in the dashboard's pricing page and through the billing API, before and after any credit purchase. We may change catalog prices at any time, for example when Anthropic changes its prices, releases or retires models, or when we change our margin. A price change applies only to calls made after it, never to calls already settled; the price in effect when a call runs is the price you pay.
Holds and settlement. Because a call's cost is known only after it runs, the service reserves a small fixed hold from your balance immediately before each managed call and settles it to the call's actual cost immediately afterwards, returning any difference. A call that fails releases its hold in full, and a hold that is never settled is released after a short time. The hold amount is shown in your billing standing as the minimum balance a managed organization needs to run anything. If a call costs more than its hold, your balance may briefly fall below zero by that excess; the shortfall is deducted from the next credit added, and you remain responsible for it.
Running out of credit. When your balance can no longer cover one hold, nothing is lost: incoming messages are still received and stored, and extraction, routing, and executions simply wait until credit is added, then resume automatically. The service also tells you at review time, when you approve or retry a task, that the organization's credit cannot cover it. A per-execution cost ceiling may apply: when one execution attempt reaches it, the run stops, what ran is charged, and the task is released for a later attempt. The service does not send balance alerts; your billing standing in the dashboard shows the balance at any time.
Shared keys and capacity. Managed calls run on a pool of Anthropic API keys that Tejma holds and operates. You never see or hold these keys, which key served a call is never disclosed to you, and the pool's capacity and Anthropic's rate limits are shared across Tejma's customers. We may throttle managed usage to protect the pool, and when no managed key is available your work waits rather than failing.
Anthropic's rules flow down. For managed calls, Tejma is Anthropic's customer and your content is processed under Tejma's agreement with Anthropic. You must therefore comply with Anthropic's Usage Policy and Commercial Terms of Service as if you were Anthropic's customer, and we may suspend managed AI billing for an organization whose use puts Tejma's standing with Anthropic at risk.
Usage records. Every call is recorded with the user, agent, task or message, job, model, token counts, and the amount billed (zero on your own key). Billing-role holders can review usage by job, model, user, agent, or day, each task's AI cost, and the credit ledger entry each call produced. Usage records support billing and are retained as described in the Privacy Policy.
Own-key billing
Bring your own Anthropic account. An organization may store its own Anthropic API key when it is founded or later, and billing-role holders may set a per-user key override for individual members. Every call on such a key is billed by Anthropic directly to your Anthropic account, at Anthropic's rates, outside these Terms, and is never debited from your credit balance; the service still records the call's token usage, at a billed amount of zero, so your usage reporting stays complete. Removing the organization's key returns it to managed AI billing from the next call; per-user overrides continue to apply. Seat plans, task quotas, and the overage option (Sections 5 and 6) apply to an organization on own-key billing exactly as to any other.
Anthropic's terms apply to you. Because AI processing then runs under your own Anthropic account, your use of the service must comply with Anthropic's Commercial Terms of Service and Usage Policy. A violation of Anthropic's terms through the service is a violation of these Terms.
Key handling. API keys are write-only secrets: they are encrypted before storage (AES-256-GCM), decrypted only at the moment of an AI call, and never returned by any endpoint. Billing-role holders may rotate, replace, or remove keys at any time; the change applies from the next call.
Models
Model catalog. Each agent, and each user's extraction and routing skill, pins a model from the catalog of Claude models the service offers. The catalog may change as Anthropic releases and retires models; the dashboard lists what is currently available, and its pricing page shows each model's catalog price.
5. Subscriptions, Seats, and Payment
Per-seat plans. Each member of your organization occupies a seat with a billing plan. A plan sets the seat's monthly price, its monthly task-execution quota, caps on the seat's active channels, enabled agents, and installed Tejma connectors (a cap may be unlimited), the per-execution price of overage past the quota (a plan may sell no overage), and the credit bonus rate that Section 6 describes. Plan prices and inclusions are shown in the plan catalog in the dashboard; billing-role holders pick a plan when a member is invited and may change it later. The plan prices the platform only. AI usage is paid separately as described in Section 4 — through the credit balance under managed AI billing, or through your own Anthropic account under own-key billing.
Monthly cycles, day-based proration. All seats share the organization's monthly billing cycle, which begins when the organization is founded and begins again when a paid subscription starts. A seat's charge for a cycle is prorated strictly by whole UTC calendar days on each plan it held during the cycle, so mid-cycle joins and plan changes bill fractionally by day. The billing view itemizes every segment, and any amount collected through Stripe matches that view to the cent.
Upgrades apply — and bill, prorated — immediately. Downgrades and cancellations take effect at the end of the paid cycle.
The current cycle is always already paid for: a downgrade keeps the seat's plan until the cycle ends, and a cancellation keeps every seat's paid plan live until the cycle ends and stops charges from the next cycle on. Neither generates a partial-cycle refund.
Plan limits and parking. A plan change is refused while a seat's active channels, enabled agents, or counted Tejma connectors exceed the target plan's caps; pause or disconnect channels, disable or delete agents, or delete Tejma connectors until the seat fits. Installing a Tejma connector, or granting one, is refused once the seat's connector cap is reached; whether an install counts toward the cap, and whether only some of its roles do, is set per connector and shown in the catalog. If a seat nonetheless ends up over its plan's caps — for example after a scheduled downgrade takes effect — the excess is parked: the oldest channels, agents, and connectors within the caps keep running, the newest are skipped by routing, dispatch, and ingestion, a parked Tejma connector refuses the member's requests, and agents wired to it are parked with it, until the seat is back under the caps or its plan grows. The dashboard shows what is parked and what a pending change would park.
Payment through Stripe. Where card payment is enabled, subscriptions and credit purchases are collected through hosted Stripe Checkout pages and a recurring Stripe subscription. Tejma is the seller; Stripe processes payment on our behalf. Card details are entered on Stripe's pages and never touch Tejma's systems. The checkout collects your billing address and, optionally, a tax ID, which are saved to your Stripe customer record for invoices. Stripe renews subscriptions monthly in advance; mid-cycle upgrades and new seats are invoiced their day-prorated amount immediately. Where Stripe is not enabled for your organization, charges are invoiced and collected out of band on the same terms. No free trial is offered; the catalog's default plan is the only free tier.
Taxes. Prices are stated exclusive of taxes. You are responsible for sales, use, VAT, GST, and similar taxes on your purchases; where we are required to collect them, they are added at checkout or on the invoice.
What is metered against the quota. One quota unit is one task execution, billed to the task owner's seat when the task is first picked up for execution. Automatic crash-recovery redeliveries and transient-error retries of an already-billed task are not billed again; a manual retry you request, a paused task resuming after its prerequisites complete, and each prerequisite task an agent creates each start a new billed attempt. Message ingestion, task extraction, task routing, and approval-workflow evaluation — judge runs included — are not metered against the quota. (Under managed AI billing, every AI call — extraction, routing, and judge runs included — is separately metered against the credit balance as described in Section 4.)
Soft limits. When a seat's quota is spent and the overage option (Section 6) does not cover the execution, nothing is lost: ingestion, extraction, and routing keep running, and the seat's queued fresh tasks simply wait until the cycle rolls over, the plan changes, or overage applies. Executions already paid for are never held back.
Failed payments. If a renewal payment fails, Stripe retries on its schedule and your seats keep their paid plans during the retry window; billing-role holders can replace the card through the billing portal in the dashboard. If payment is not collected by the end of the retry window, every seat on a priced plan drops to the default plan immediately and the subscription ends; a new subscription checkout starts the next one. The credit balance is not affected by a failed renewal.
Seats, deactivation, and deletion. Deactivating a member, or revoking their roles, does not by itself end that seat's billing — a billing-role holder should also move the seat to the default plan to stop future charges. Deleting a member removes their seat and its content (see the Privacy Policy) and stops its billing from the next renewal; the current cycle is not refunded. Fees already paid are non-refundable except where required by law or expressly stated otherwise.
Plan changes by Tejma. Plan prices may change; a price change applies from your next billing cycle and never retroactively. A plan's other attributes — its task quota, caps, overage price, and credit bonus rate — may change at any time, with effect on the current cycle, and the dashboard always shows the plan as it currently stands. We may also place a seat on a plan outside the public catalog by agreement with you.
6. Prepaid Credit Balance
One balance per organization. Your organization holds a single prepaid credit balance, denominated in US dollars and pooled across its seats. It is increased by credit purchases (at least one dollar of credit for every dollar paid — see the credit bonus below), by a one-time founding credit we may gift a newly founded organization, and by any grants or gifts Tejma chooses to make. It is drawn down by two things: managed AI usage (Section 4), and — when a billing-role holder has switched on the organization's overage option — each task execution past a seat's monthly quota, at the seat plan's overage price. The overage option is a separate switch: buying credit does not turn it on. An organization on own-key billing uses the balance for overage only.
Credit bonus. Each plan carries a credit bonus rate, shown in the plan catalog and in your billing standing. When your organization buys credit, the amount paid is added to the balance and, on top of it, promotional credit equal to the amount paid multiplied by your organization's bonus rate. The rate that applies is the rate of the most expensive seat plan your organization holds at the time; where two plans share a price, the higher bonus applies. The rate is fixed when a checkout begins, so a plan change before the payment settles does not change what the checkout promised. The bonus applies to purchases only — not to the founding credit, to grants or gifts, or to subscription renewals — and a plan's rate may be zero. Bonus credit does not expire and is spent like any other credit, but it is promotional credit: it is recorded in the ledger as a separate entry from the amount paid, it has no cash value, and we may reduce a balance by bonus credit obtained through fraud or abuse, never below zero.
Buying credit. Only billing-role holders can buy credit, from one dollar to ten thousand dollars per purchase. Where Stripe is enabled, credit is bought through hosted Stripe Checkout and added to the balance only once the payment settles — an abandoned checkout adds nothing. Where Stripe is not enabled for your organization, purchases are invoiced and collected out of band and recorded in your balance on payment.
Ledger. Every change to the balance — purchases and their bonus, gifts, grants, each metered AI call, each overage execution, and any correction — is journaled in the same transaction that changes the balance, and billing-role holders can review the full ledger at any time. Each AI-usage entry names the call it paid for.
No cash value. Credit — purchased, bonus, gifted, or granted — has no cash value, is not transferable between organizations, and is not refundable except where required by law. Gifted and promotional credit, including the founding credit and the credit bonus, is a courtesy: it may carry conditions we state when we grant it, and we may withdraw it in cases of fraud or abuse. Switching off the overage option stops overage consumption but does not forfeit the balance; switching to own-key billing stops AI consumption, and the balance remains available for overage executions.
On closure. When an organization terminates, its remaining credit is forfeited, except where the law requires a refund of unused purchased credit.
7. Connected Channels and Customer Content
Your authority to connect. Connecting a channel — an email mailbox, a Slack channel, a Discord channel — or submitting messages through an MCP client or our API causes the service to receive, store, and process the messages and files that flow through it, including messages written by people other than you. For an email that continues a conversation, the service also reads the earlier messages of that thread from your mailbox, which may include your own replies and messages that predate the connection. You represent that you are authorized to connect each channel and to have its content processed as described in these Terms and the Privacy Policy, and that you have provided any notices and obtained any consents required by the laws that apply to you and by the rules of the connected workspace or provider. A mailbox connected again by another member of your organization moves to that member from then on; what was already stored stays with the member who connected it first.
Customer Content is yours. All messages, attachments, documents, thread transcripts, tasks, agent skills, approval workflows, outbound messages, and task outputs in your account ("Customer Content") belong to you. You grant Tejma a limited, non-exclusive license to host, copy, transmit, and process Customer Content solely to provide, secure, bill, and support the service, including sending it to Anthropic and to the connectors and channels you configure. We do not sell Customer Content, use it for advertising, or use it to train AI models.
Retention by design. Attachments and files are stored as documents on their message precisely so tasks keep working after the originating email, chat message, or channel connection is gone. Disconnecting a channel stops new ingestion but does not delete already-stored content or the audit trail of messages sent through it. Documents can be deleted individually; messages, their thread transcripts, and the send history are removed only when the seat that holds them is deleted, or by Tejma on your written request, as the Privacy Policy describes.
The service is not an archive. The service stores content to run your automation, not as a system of record or a backup service. Keep originals in their source systems.
8. Acceptable Use
You agree not to use the service, including its agents, approval workflows, connectors, and outbound channels, to:
- send spam, bulk unsolicited messages, or messages that violate the CAN-SPAM Act, the TCPA, or equivalent laws where recipients are located;
- violate Anthropic's Usage Policy, whichever billing mode you use, or the terms, policies, or rate limits of a connected provider (Slack, Discord, your email provider) or of a Tejma connector;
- connect mailboxes, workspaces, or channels you are not authorized to connect, or ingest content you have no right to process;
- circumvent metering, quotas, holds, plan limits, connector grants, or approval gates, obtain credit bonus by manipulating seat plans, or attempt to extract, discover, or misuse Tejma's managed Anthropic API keys;
- transmit malware, attempt to breach or probe the service's security, or access another tenant's data;
- process content that is unlawful, infringing, or that you lack the rights to process;
- resell, sublicense, or white-label the service, or resell managed AI capacity, without a separate written agreement with Tejma;
- use the service to build a competing product by systematic extraction of its functionality.
We may suspend or throttle accounts engaged in the above, and will notify the organization's administrators where the law and the circumstances allow.
9. AI Outputs, Approval, and Human Review
AI can be wrong. Task extraction, task routing, approval judgements, and task execution are performed by AI models. Outputs may be inaccurate, incomplete, or unsuitable for your purpose, and identical inputs may produce different outputs. The approval step exists so that what will be executed is reviewed: you are responsible for reviewing tasks before approval and for reviewing agent output before relying on it, especially where it triggers outbound messages or actions through your connectors.
Approval workflows are your choice. Approval workflows let you approve tasks automatically by rules you write — conditions on the task and its message, optionally confirmed by a judge agent of yours. A judge run is a full agent execution, with the judge's connectors and channels, and under managed AI billing it is metered against your credit like one. Tasks approved by your workflows execute without a human reviewing them, with the same consequences a human approval would have, and you are responsible for the rules you set. A failing judge never rejects a task; it leaves the task waiting for a human with a suggested reason.
At-least-once execution. Executions, outbound messages, and completion reports are delivered at least once: a crash between an agent's side effects and its bookkeeping may re-run the task and repeat a message or action. Design connector tools to tolerate retries and expect occasional duplicates; the send audit trail records every delivery.
Costs follow your configuration. Each execution, manual retry, resumed task, and agent-created prerequisite task consumes quota and, under managed AI billing, credit; judge runs, extraction, and routing consume credit under managed AI billing but not quota. Chains of agent-created tasks, approval workflows, and tool-heavy agents can multiply calls. The seat quota, the credit balance, and any per-execution ceiling bound that spend, and the ledger and usage reports explain it; you are responsible for what your configuration consumes.
No professional advice. Outputs are not legal, financial, medical, or other professional advice. You retain full responsibility for decisions made and actions taken based on the service's output.
10. Third-Party Services
The service interoperates with third-party services under their own terms: WorkOS (authentication), Anthropic (AI processing — under Tejma's account for managed AI billing, under yours for own-key billing), Stripe (payment processing), Nylas (email connectivity), Slack, Discord, Railway (hosting), and Cloudflare (file storage and network ingress), plus any connector servers you register. Your use of each is governed by that provider's terms, and their availability is outside our control — an outage, API change, rate limit, or account action at a provider may interrupt the corresponding feature. Connectors you register by URL are third-party services chosen by you; Tejma does not review, endorse, or assume responsibility for them or for what your agents do through them.
Tejma connectors. Tejma connectors are Tejma's own services, not third parties, and Tejma is responsible for them under their own terms. To decide whether to serve a member, a Tejma connector asks the service whether that member's seat holds a current, unparked install and grant, and receives the member's user identifier, organization identifier, the install's identifier, and the granted roles — never the member's name or email address. Any charges, quotas, or rules of a Tejma connector are set out in its own terms.
11. Privacy and Data Protection
Our Privacy Policy describes what we collect, how we use it, what reaches Anthropic in each billing mode, how long we keep it, and the rights available to you; it is incorporated into these Terms. Where the Customer's use of the service involves personal data subject to the GDPR, UK GDPR, or similar laws, our Data Processing Agreement applies and is incorporated by reference; Tejma acts as the Customer's processor for Customer Content.
12. Suspension and Termination
By you. Billing-role holders may cancel the organization's subscription at any time (effective at the end of the paid cycle, as above); users-role holders may deactivate or delete members; every member may disconnect their channels and delete their documents and agents. Cancelling the subscription does not delete any content. To close the organization and have its content deleted, a billing- or users-role holder emails us at the contact below; we then delete every seat of the organization, as the Privacy Policy describes. You may stop using the service at any time.
By us. We may suspend or terminate access for material breach of these Terms, non-payment after notice, a violation of Anthropic's policies through managed AI billing, legal requirement, or a genuine security risk. Where practicable we will notify the organization's administrators and give an opportunity to cure before termination.
After termination. Upon termination, the Customer may exercise the data rights described in the Privacy Policy — including access to and deletion of its Customer Content — by written request, and we delete the organization's content as described there. Remaining credit is handled as described in Section 6. Sections that by their nature survive — ownership, disclaimers, liability limits, and payment obligations accrued before termination, including any negative balance — survive.
13. Disclaimers and Limitation of Liability
Disclaimer. The service is provided "as is" and "as available." To the maximum extent permitted by law, Tejma disclaims all warranties, express or implied, including merchantability, fitness for a particular purpose, non-infringement, and any warranty that the service, its AI outputs, or managed AI capacity will be uninterrupted, error-free, or accurate.
Liability cap. To the maximum extent permitted by law, neither party's aggregate liability arising out of or relating to the service will exceed the amounts the Customer paid to Tejma for the service in the twelve (12) months before the event giving rise to liability, and neither party is liable for indirect, incidental, special, consequential, or punitive damages, or for lost profits, revenue, or data. These limits do not apply to a party's willful misconduct, to the Customer's payment obligations, or to either party's liability that cannot be limited by law.
Indemnity. The Customer will defend and indemnify Tejma against third-party claims arising from Customer Content, from channels connected without authority, from messages its agents send, or from use of the service in violation of these Terms, Anthropic's policies, or applicable law.
14. General Terms
Governing law and venue. These Terms are governed by the laws of the State of Nevada, excluding its conflict-of-laws rules, and disputes will be resolved in the state or federal courts located in Washoe County, Nevada, which both parties consent to.
Miscellaneous. These Terms, together with the Privacy Policy, the Data Processing Agreement (where applicable), the terms of any Tejma connector you use, and any ordered plan, are the entire agreement regarding the service and supersede prior discussions. Neither party may assign them without consent, except to a successor in a merger or asset sale. If a provision is unenforceable, the rest remains in effect. A failure to enforce is not a waiver. Notices to Tejma go to Tejma LLC, Reno, Nevada, at the contact below; notices to the Customer go to its organization's administrators through the service or email on file.
Matej Meciar · Tejma LLC
