Navigate AI risk

with confidence

papermoon keeps your ai on track

We help you deploy AI more confidently, with fewer safety and quality surprises.

Our tools and services help you set the guardrails and monitor your AI, so you can focus on what your customers need.

Anticipate. Analyze. Act.

We help you transform vague AI concerns into structured, actionable oversight.

Our AI-assisted policy & monitoring tools provide the groundwork to create clarity on expectations from both AI and your team.

  • We provide assessments of which AI risks your organization faces, and prioritize which ones to address first.

    Product roadmaps: Whether you are evaluating an external system or building a new prototype, we help you not only to identify risks, but also identify the tools and capabilities that will help you to scale your product vision safely.

  • We build robust governance frameworks to ensure your AI deployments align with organizational goals and regulatory requirements. This includes investigating risks, and developing policies and rollouts that fit your organization’s specific context.

    Policy & process development: we help you prioritize critical risks and create a practical mitigation roadmap. This includes implementation strategies and operational frameworks, ensuring policies are enforceable and aligned with real-world workflows.

    AI-assisted policy tracking: we help you centralize policy tracking to enable comparison, consistency, and surfacing relevant updates to the right teams.

  • Making an AI prototype is easy, but improving real-world quality requires clear KPIs, evaluations and metrics.

    Define success. Together we’ll define clear KPIs and quality metrics that link technical performance to your broader organizational objectives.

    Deliver evaluation & simulation libraries. We help you with metric design, validation and workable, operational implementation. We adjust benchmarks and red teaming approaches to your context, generate test sets that your system shouldn’t fail on, and implement evaluation metrics that fit your context.

Not sure what’s actually happening on your platform?

Get a handle on surprise behavior

When your AI system (or your users) behave in strange ways, we give you clear insights and actionable options.

We investigate interactions and anomalies using AI-assisted techniques, and review whether existing policies and safeguards are still fit for purpose.

We flag priority behavior and content, and reduce the time spent on interventions that won’t have an impact.

prepare for the unexpected

Your product will be used in unintended ways, and bad actors might find your platform.

We deliver insight in vulnerabilities through benchmarking, tailored evaluations, and simulated red team attacks.

Additional tabletop exercises prepare your human team for real scenarios and validate that your procedures hold up under pressure.

the team

We are technology and policy veterans with prior leadership roles at organizations and companies like Spotify, Yahoo, ML Commons and Accenture. We have deep expertise in the intersection of AI and UX, Trust and Safety and policy development.

We have built systems from 0 to production, wrangled data at scale, managed worldwide rollouts, investigated crisis escalations, and set up algorithmic responsibility programs.
‍
Our international network ensures both in-depth domain expertise and scale to fit your needs while your business grows.

We're based in San Francisco, California.