AI Disclosure

When you chat with Ted, you are chatting with software backed by a large language model. Ted is not a person. Even when the conversation feels human, you are interacting with a probabilistic text generator coordinated by code.

We make this disclosure so there is no ambiguity. If you're uncertain in any conversation whether you're talking to a human or to AI, the answer is AI.

Claude models developed by Anthropic, PBC power Ted's conversational reasoning and primary build agents. The specific model varies by feature. Foundable also uses OpenAI for selected embeddings, structured extraction, and transcription; Cohere for reranking; and fal.ai for enabled media generation. Agnost receives feature-scoped conversation and tool data for quality analytics. The current provider scopes and data categories are maintained on our provider registry. We may change models over time and do not guarantee a specific model indefinitely.

Within an authenticated foundable, Ted can:

• Read messages you send it, files you upload, and data from third-party Integrations you have connected.
• Use explicit public-web research tools when a feature calls for them and identify the source for personal context it uses.
• Plan multi-step projects and execute them by spawning workers that call tools.
• Take actions in connected services on your behalf: create a Linear issue, send a Slack message, draft a Google Doc, etc., using credentials you provided.
• Generate text, code, summaries, and structured outputs.
• Maintain context across messages within a chat thread and across threads within a foundable.

• Access the contents of your device, browser history, email inbox, contacts, or private accounts unless you explicitly connect an Integration that provides the specific data. Ted can use explicit web-search tools to access public pages.
• Verify facts on its own. It can and will produce confident, fluent text that is wrong.
• Replace professional judgement (legal, medical, financial, regulatory).
• Guarantee that every provider-side action can be stopped or reversed. Stop controls prevent or interrupt supported future work; they do not undo a message, publication, purchase, or other action already completed in a third-party service.
• Persist memory beyond what we store in the database for the foundable. There is no shared memory across different foundables.

During onboarding, Foundable may use the name on your account and, when your email domain is not recognized as a common personal-email provider, a broad organization affiliation derived from that domain to search public professional pages. This lets Foundable explain a possible match in Ted's welcome, but a search result is only a possible identity match.

Foundable may also compare a normalized version of your email username locally with handles on the returned public profiles. The email username and full email address are not sent to the public-search provider for this comparison.

When Foundable finds a profile that is likely to be yours, its trusted server adds a source-linked note to Ted's welcome, presents the match as tentative, and invites you to correct it. The unconfirmed profile is not provided to Ted's AI model or used to influence its questions. If results are ambiguous or conflicting, or do not support a likely match, Foundable suppresses the profile details and continues using information you provide directly.

This lookup does not inspect your device, browser history, email inbox, contacts, or private-account content. See our Privacy Policy for the data sent to our search provider and how long Foundable keeps the temporary results.

Foundable does not train its own foundation models on Customer Content.

Anthropic's commercial terms prohibit training on Customer Content submitted through its services. OpenAI states that business and API data is not used for training by default, and fal.ai's API terms include a no-training commitment subject to stated exceptions. See Anthropic's Commercial Terms for the source.

Other service providers are limited to the feature scope and data categories described in our provider registry. Contact us with account-specific contractual or data-handling requirements.

Anthropic publishes model-specific information through its system and trust reporting materials. We do not have visibility into provider training data beyond what each provider publishes or contractually discloses.

Output produced by Ted is AI-generated. It may be:

• Factually wrong ("hallucinated"): Ted can invent citations, names, statistics, or facts and present them confidently.
• Biased in ways inherited from the training data.
• Inappropriate for your context, audience, or jurisdiction.
• Similar or identical to output produced for other users given similar prompts.
• Out of date: the model has a knowledge cutoff and doesn't know about events after it.

Always review Output before publishing, sending, or relying on it for any consequential decision. If your jurisdiction requires you to disclose AI-generated content to recipients (e.g. AI-generated political advertising in some U.S. states), that obligation is yours to meet.

The full restrictions are in our Acceptable Use Policy. Some uses are prohibited even with human review, including criminal-justice determinations such as sentencing, parole, or predictive policing. High-impact decisions in legal, health, finance, employment, housing, education, and similar domains require qualified human review and disclosure that AI is being used, and may still be prohibited by a provider policy or applicable law.

Foundable is not approved as the sole basis for a legally or similarly significant decision about an identifiable person.

Operator controls depend on the workflow:

• Ordinary work starts from an operator instruction. When you enable a scheduled or autonomous motion, Foundable may continue within the scope, autonomy mode, budget, schedule, and connected accounts you selected until the motion is paused or disabled.
• Stop, cancel, and disconnect controls prevent or interrupt work where supported. Completed actions follow the connected provider's available edit, recall, refund, or reversal controls.
• Foundable surfaces supported project, run, output, and tool lifecycle records, while connected providers maintain their own service-side records.

Do not submit content to Foundable that you are not authorized to share with our subprocessors (see /subprocessors). In particular:

• Protected Health Information (PHI) — Foundable's standard service is designed for general business data. Don't submit PHI.
• Full payment method numbers and cardholder data — handle through your payment processor instead.
• Government-classified information — we are not authorized to process classified material.
• Other regulated data (CJI, FERPA student records, etc.) — only with our prior written agreement that the specific use is supported.

We take feedback seriously and use it to improve guardrails.

• Harmful or unsafe output — email abuse@foundable.com.
• Factually wrong output — email feedback@foundable.com.
• Bias or fairness concerns — email privacy@foundable.com.
• Security or jailbreak findings — email security@foundable.com and follow our Responsible Disclosure Policy.

This disclosure explains Foundable's product behavior; it is not a representation that a particular customer use complies with the EU AI Act, California's AI laws, Colorado's AI laws, or any sector-specific rule. Customers remain responsible for notices, qualified human review, content labeling or machine-readable marking, bias or impact assessments, and other obligations that apply to their use.