How we use AI

Draftlytic is operated by Robert Boylan, trading as Draftlytic, based in Ireland. Draftlytic is an AI-powered project planning tool. AI is central to the Service — it generates specs, edits projects, creates logos, and powers the guided creation questionnaire. This policy explains how we use AI, what data is involved, and what control you have.

Draftlytic primarily uses Google Gemini via the Gemini API for text generation (project planning, editing, and PRD Workshop features). When Gemini is temporarily unavailable, requests may instead be served by OpenAI (gpt-5.4-mini) as a reliability fallback — Gemini is always tried first and the fallback is non-sticky (the next request goes back to Gemini). Image generation always uses Gemini regardless. Google AI may be served via either Gemini AI Studio or Google Cloud Vertex AI — both operated by Google LLC and covered by the same Google Cloud Data Processing Addendum. Google processes AI requests under the Gemini API Terms of Service. OpenAI processes requests under the OpenAI Terms of Use.

Only the specific input data you provide is submitted to our AI providers (primarily Google Gemini; OpenAI as a fallback — see Section 2) for processing. This includes:

• Your project prompt and any text you enter when generating or editing a project
• Existing project data (name, features, overview, etc.) when you use the "Edit with AI" feature, along with up to 10 prior messages from your conversation with the AI edit assistant as context for each edit request
• A style prompt when generating logos
• Your answers to guided creation questions
• Screenshots or images you upload to the sketchpad canvas during project creation — these are sent to Google Gemini as part of your generation request
• PRD Workshop chat — every turn re-sends your full pasted PRD content (and the prior messages in that workshop session) so Gemini can rewrite it with full context
• AI scan — your full project content (or pasted PRD) is sent so Gemini can rate it and produce a fix list
• Implementation plan generation — your project name, overview, features, tech stack, design choices, and business context are sent so Gemini can produce a step-by-step build plan tailored to your target tool

We do not send your Draftlytic account details (name, email address, payment information) to Google — only the project-related text and images you provide. As noted above, you are responsible for not including personal data in project inputs. If you upload screenshots, paste text from external systems, or describe real users by name in your project content, that content reaches Google's Gemini API along with the rest of your project.

Under the Gemini API Terms of Service, Google does not use data submitted through the API to train its AI models. Your prompts and the AI's responses are not used for model improvement or training purposes.

Under OpenAI's API usage policies, OpenAI likewise does not use API data to train its models by default.

Both providers may temporarily retain API data for safety and abuse monitoring purposes in accordance with their respective terms, but this data is not used for training.

For most AI features (project generation, logo creation, guided questions), each request is independent — data sent for processing is used only for that specific request.

The AI edit assistant maintains a short-term conversation history within your editing session. Up to 10 prior messages are included as context when you send a new edit request, so the AI can understand follow-up instructions. This history exists only for the duration of your session and is not stored persistently or sent to Google outside of active edit requests.

The PRD Workshop is an exception to the per-request independence above: it re-transmits your full pasted PRD on every message in the chat so the AI can keep rewriting it with complete context. See §3 for the full list of what each Workshop turn sends.

AI features are only activated through your deliberate actions. You choose when to generate a project, edit with AI, create logos, or use guided creation. AI processing never happens automatically or in the background without your action.

All AI-generated content (specs, features, logos) can be freely reviewed, edited, and deleted by you. You retain full control over the output.

AI-generated content (specs, features, logos, etc.) is stored within your Draftlytic account as part of your project data, subject to our standard Privacy Policy and data retention practices. When you delete a project or your account, all associated AI-generated content is deleted.

AI features depend on the availability of Google Gemini, a third-party service. Draftlytic does not guarantee that AI features will be available at all times. The AI provider may occasionally be unavailable or slow due to high demand, maintenance, or other circumstances outside our control.

If an AI request fails due to a provider outage or error, any credits deducted for that request are automatically refunded to your account. You will not be charged for failed AI generations.

Draftlytic makes no guarantees as to the accuracy, completeness, or suitability of any content produced by AI. All AI-generated content is provided "as is" and used at your own risk.

You are responsible for reviewing the accuracy and appropriateness of any AI-generated content before using it. Output may contain errors, inaccuracies, or material that requires further validation. We shall not be liable for any damages or losses resulting from reliance on AI-generated content.

Your use of AI features is subject to our general Terms of Service. Additionally, you must not use AI features to:

• Generate content that is illegal, harmful, abusive, or violates third-party rights
• Attempt to extract or reverse-engineer the AI models
• Violate Google's Gemini API usage policies

We may update this AI policy from time to time, including if we change AI providers or add new AI features. We will notify you of material changes by posting the updated policy on this page and updating the effective date.

If you have questions about how we use AI, contact us at draftlytic@gmail.com.

• June 12, 2026 — Added disclosure of OpenAI (gpt-5.4-mini) as a reliability fallback provider when Gemini is temporarily unavailable; updated no-training and data retention disclosures to cover both providers
• May 20, 2026 — Initial version