The Insight System uses AI to provide personalized experiences. We believe in full transparency about how AI is integrated into our platform.
Personality reports are written by DeepSeek V4 Pro, an open-weight model; the couples syntheses, values reports, journal coaching and type-discovery features use Meta Llama 3.3 70B. Both run on Cloudflare Workers AI — a private inference platform within Cloudflare’s global infrastructure. When you interact with AI features, your data is processed within this infrastructure and the response is returned to you.
For the long version — exactly what goes into a report, layer by layer — read Truth in AI.
What “open-weight” means here: the model files are published by their makers and run on Cloudflare’s servers. Nothing you type is sent to DeepSeek, Meta, or any other model maker, and none of them can see it.
Your data never leaves the Cloudflare infrastructure the platform runs on. It is processed by Cloudflare Workers AI in the same environment that serves these sites. No third-party AI companies (such as OpenAI, Anthropic, or Google) receive your content. Cloudflare does not use your content to train AI models, does not share it with other customers, and does not use it to improve any services.
Cloudflare’s infrastructure is trusted by major financial institutions, healthcare organizations, and governments worldwide. The same security and privacy standards that protect those industries protect your data on The Insight System.
For full details, see Cloudflare’s official data usage policy: developers.cloudflare.com/workers-ai/platform/data-usage
Most AI-powered apps send your personal data to a third-party AI company for processing. We don’t. Here’s the difference:
What is Cloudflare? Cloudflare is one of the world’s largest internet infrastructure companies. They power over 20% of all websites and are trusted by major banks, hospitals, and governments to keep data secure.
What does “private AI” mean? When you use features like report generation or journal coaching, the AI runs directly on Cloudflare’s servers — the same servers that already store your account. Your data doesn’t take a detour to a different company. It stays where it already lives.
Cloudflare’s guarantees (from their official data usage policy): They don’t use your content to train AI. They don’t share it with other customers. They don’t use it to improve their own products. You own your data, period.
What AI does: Generates personalized personality profile reports based on your type and interview answers.
What data is sent: Your 4-letter type, temperament category, and interview responses.
What AI does NOT do: AI does not determine your personality type. Type is determined algorithmically based on your card selections.
What AI does: Generates in-depth reports on career, relationships, stress, and growth, tailored to your personality type.
What data is sent: Your personality type, temperament description, report topic, and personal context you provide.
What AI does: Synthesizes both partners’ responses into a unified analysis, and provides conversational coaching.
What data is sent: Both partners’ non-private responses, personality types, relationship context, and chat messages.
Important: Private question responses are NOT sent to AI or shared with your partner.
What AI does: Writes your values report from your card sort, and writes a comparison when two people enter each other’s codes.
What data is sent: Your sorted values, your top seven and their status, and your first name if you gave one. Runs on the same private Cloudflare Workers AI.
What AI does: Generates personalized journal prompts, provides coaching, and analyzes mood and themes.
What data is sent: Your journal entry, personality type, goals, life stage, and recent context.
We continuously improve our AI prompts to maximize accuracy and helpfulness, but AI is inherently imperfect. Use AI-generated insights as one input among many, not as definitive answers.
The personality frameworks, assessment questions, and therapeutic principles embedded in our AI prompts are grounded in over 30 years of clinical work by Dr. Martine J. RoBards. The AI models (DeepSeek V4 Pro for reports, Meta Llama 3.3 70B elsewhere) operate within structured prompts that include the full RoBards type descriptions, temperament profiles, clinical observations, stress patterns, and growth frameworks. This ensures responses align with established psychological research rather than generic AI output.
However, no human reviews individual AI-generated responses before they are delivered to you. If you receive a response that concerns you, please contact us.