Our principles

Responsible AI

Last updated: November 3, 2026

Northrise builds AI products for specific markets. We believe focused, useful AI built with care delivers more value to real people than general-purpose tools. This page describes the principles that guide every product we build, and what those principles mean in practice.

1. People first

We design for the user we're serving — not for the AI. Our products work without setup, without configuration, without learning curves. If a feature only works because the user becomes an “AI expert,” we've designed it wrong.

2. Useful from day one

We measure success by whether someone gets real value in their first session. If they don't, we go back and fix it. We don't ship aspirational AI that “could” work — we ship AI that works now.

3. Transparent about what AI does

Each product clearly indicates when AI is generating content, making suggestions, or taking action on the user's behalf. Users always know what the AI did and what they did themselves.

4. Built on trusted foundations

We build on top of established AI platforms (currently including OpenAI and Anthropic). We don't train our own foundation models. Their safety practices are the floor; our product-specific guardrails sit on top.

5. We don't train on your data

User content from our products is never used to train foundation models without explicit consent. We're a customer of these platforms — we configure our accounts so that user data isn't shared back for model training.

6. Human oversight on consequential actions

Any action with real-world consequences — sending an email, scheduling, transacting, sharing data — requires user confirmation. The AI proposes; the human disposes.

7. Honest about limits

Our products will sometimes be wrong. We tell users that. We invest in mechanisms — confidence indicators, undo, escape hatches — to make wrongness recoverable.

8. Bias and fairness

Foundation models carry biases from their training data. We test our products against the markets they serve, surface where bias might affect outcomes, and document mitigations in product-specific documentation.

9. Privacy by default

We follow data minimization. We collect only what a feature requires. See our Privacy Policyfor specifics on data handling at the company level, and each product's own privacy policy for product-specific practices.

10. We listen

We treat user reports of incorrect, harmful, or biased outputs as urgent. Each product provides a clear path to flag issues, and we respond.

What this means for builders, partners, and journalists

If you're evaluating whether to build with us, partner with us, or write about us:

Changes to these principles

These principles will evolve as the field evolves and as our products grow. The “Last updated” date above reflects the most recent change. Material changes will be highlighted.

Contact