TarraCoata builds and operates AI-assisted features across its website, its client deliverables and the TarraOne platform. This page explains how those features work, what you can rely on, and what you must verify yourself.
1. Where we use AI
- Website assistant — answers questions about our services and routes enquiries to a human.
- Editorial pipeline — researches topics and drafts articles that are reviewed and approved by a human editor before publication. Nothing is published automatically.
- TarraOne — lead scoring, summarisation, drafting of replies, task suggestions, classification and workflow automation.
- Client engagements — AI components we design, build and deploy on your behalf.
Where an article or interface element is materially AI-assisted, we label it.
2. AI-generated content may contain inaccuracies
Large language models generate text probabilistically. Output can be incomplete, outdated, internally inconsistent, or confidently wrong ("hallucinated"), including invented citations, figures, names and legal references. AI output is provided for informational and productivity purposes only and is not a statement of fact by TarraCoata.
3. You are responsible for verifying outputs
Before relying on, publishing, sending or acting on any AI output, you must review it, verify facts against authoritative sources, and confirm it fits your circumstances. You remain fully responsible for anything you send or publish from your workspace, and for compliance with the laws that apply to you.
4. No professional advice
AI features do not provide legal, medical, financial, tax, accounting, regulatory, insurance or engineering-safety advice, and no professional relationship is created by using them. Consult a qualified professional before making decisions in those areas.
5. AI responses are suggestions only
Scores, rankings, recommendations, drafted messages and automation proposals are decision support, not decisions. A human must review any output before it produces a legal or similarly significant effect on a person. Automated actions can be disabled per workspace.
6. Inputs, training and confidentiality
- We do not use customer workspace content to train foundation models, and we contract with model providers on terms that prohibit training on our submitted data.
- Prompts and outputs may be retained briefly for abuse prevention, debugging and quality monitoring, then deleted according to the retention rules in the Privacy Policy.
- Do not submit special-category personal data, credentials, payment card numbers or third-party confidential information into AI features unless your contract with us expressly permits it.
7. Service availability
AI features depend on third-party model providers and may be degraded, rate-limited, temporarily unavailable, or changed at short notice. Model versions may be replaced, which can alter output style and quality. AI features are excluded from any service-level commitment unless stated otherwise in writing.
8. Rate limits and fair usage
To protect availability we apply per-account, per-IP and per-workspace limits on requests, tokens, generation length and concurrency. Limits vary by plan and may be adjusted. If you exceed them we may throttle, queue or temporarily suspend AI features. Sustained usage that is materially disproportionate to your plan may require an upgrade.
9. Abuse prevention
You may not use our AI features to generate or facilitate: unlawful content; malware, exploits or phishing; harassment, hate speech or content sexualising minors; disinformation or impersonation; spam or bulk unsolicited messaging; scraping or model extraction; circumvention of safety filters; or any attempt to reverse engineer the underlying systems. We monitor for abuse and may suspend access immediately where we detect it. See the Acceptable Use Policy.
10. Intellectual property in AI output
Subject to payment and to the model provider's terms, you own the outputs you generate from your own inputs. Because models may produce similar text for different users, we cannot warrant that any output is unique or free of third-party rights, and you should check before using it commercially.
11. Liability
To the maximum extent permitted by law, TarraCoata is not liable for loss arising from reliance on AI output, from unavailability of AI features, or from decisions made without human review. Nothing here limits liability that cannot lawfully be limited.
12. Contact
Questions about our use of AI: tarracoatainfo@gmail.com.