Listen to this articleMoonbounce, an Oakland-based AI control engine, raised $12 million in its first major funding round, co-led by Amplify Partners and StepStone Group. The capital will accelerate development of its real time policy enforcement technology, enabling scalable, auditable guardrails for generative AI applications across high volume platforms.
Moonbounce, an Oakland-based AI control engine founded in 2024, emerged from stealth as Clavata, with a $12 million funding round that coincides with its rebranding and public launch. The round was co-led by Amplify Partners and StepStone Group, with participation from angel investors PrimeSet and Josh Leslie, former CEO of Cumulus Networks and Gremlin. This capital infusion marks the company’s first major disclosed raise and fuels its positioning as a critical infrastructure layer for predictable AI behavior at scale.
What is Moonbounce’s main focus?
The company addresses a core failure mode in the generative AI era: the inability of large language models and AI systems to consistently enforce complex, human written policies in real time. Traditional content moderation (reliant on manual review, rigid rules, or post hoc filtering) cannot match the speed of AI systems that make thousands of decisions per second across user generated content, chatbots, image generators, and enterprise workflows. Moonbounce’s patented realtime AI control engine translates static policy documents into executable, testable logic. It evaluates inputs or outputs in under 430 milliseconds (often 300ms), applies nuanced decisions (block, flag for review, throttle distribution, or iteratively steer prompts), and delivers auditable, traceable actions. A lightweight API enables integration in under an hour, while a pre built policy library and Playground sandbox accelerate testing and deployment. The platform supports compliance with frameworks like the EU Digital Services Act and California’s SB 243, maintains SOC 2 certification, and prioritizes privacy with minimal data retention.
Key metrics underscore early product market fit. Moonbounce processes over 50 million pieces of content daily, has evaluated more than 1 trillion tokens, and powers experiences reaching 250 million monthly active users (with some reports citing 80–100 million daily active users). Customers span dating platforms, AI companion and character apps (including Channel AI, Dippy, and Moescape), and generative media tools (Civitai for image and video generation). These deployments already deliver measurable gains, such as 10x improvements in detection accuracy for certain harms compared to legacy approaches.
The founding team brings unmatched domain expertise. CEO Brett Levenson previously led business integrity efforts at Meta (Facebook) during the Cambridge Analytica era and held roles at Apple. He witnessed firsthand the limitations of human driven moderation, reviewers memorizing lengthy policy documents and achieving roughly 50% accuracy under extreme time pressure. Co-founder and CTO Ash Bhardwaj, also an Apple veteran, built large scale cloud and AI infrastructure. The broader team draws from privacy and security leadership at Meta, Apple, and Evernote, embedding enterprise grade reliability and compliance into the product from day one.
Investors highlight the thesis explicitly. Lenny Pruss, general partner at Amplify Partners, noted that content moderation has long plagued platforms but becomes exponentially harder when LLMs sit at the core of every application; Moonbounce’s real time, objective guardrails can become the “enabling backbone” of AI mediated experiences. StepStone Group’s participation signals institutional conviction in the category’s scalability. The angels add operational depth from infrastructure and reliability backgrounds.
Strategically, the timing is impeccable. As enterprises and startups race to embed generative AI into products, they face mounting regulatory, legal, and reputational risks, evident in high profile incidents involving self harm guidance, non consensual imagery, and policy violating chatbots. Moonbounce flips safety from a reactive cost center into a proactive product differentiator. Customers already use it not just to mitigate harm but to create branded, trustworthy experiences (for example, steering conversations toward empathy or helpfulness rather than simply blocking). The pay per use pricing model with volume discounts lowers barriers to adoption while aligning incentives for high scale usage.
Recommended: Hamilton AI Raises $7.5 Million In Seed Funding Round
Looking ahead, the $12 million will support engineering expansion for a currently 12 person team, broader vertical reach into healthcare, financial services, and consumer social, and evolution of the action toolkit. Future capabilities include advanced “iterative steering”, intercepting and modifying prompts or responses mid conversation to redirect harmful trajectories without breaking user flow. This positions Moonbounce as more than moderation: a full decision layer that narrows the gap between organizational intent and actual system behavior.
In a market flooded with AI point solutions, Moonbounce stands out through its policy as code architecture, sub second latency at massive scale, and focus on auditability and privacy. By making safety enforceable, testable, and product native, the company reduces the friction that currently forces teams to choose between velocity and control. The funding validates that real time AI governance is no longer optional infrastructure, it is becoming table stakes for any organization deploying generative systems responsibly. Moonbounce is poised to capture significant share of the emerging AI safety layer, enabling faster, safer innovation across the ecosystem.
Please email us your feedback and news tips at hello(at)superbcrew.com
Activate Social Media:
