Goodfire Raises $7M for Its AI Observation Platform

Goodfire, a startup developing tools to increase visibility into the inner workings of generative AI models, today announced it has raised $7 million in seed funding led by Lightspeed Venture Partners, with participation from Menlo Ventures, South Park Commons, Work-Bench, Juniper Ventures, Mythos Ventures, Bluebirds Capital, and several leading investors.

Tackling the 'black box' problem

As generative AI models such as large language models (LLMs) become increasingly complex, hundreds of billions of parametersor internal institutions that determine their behavior — they have also become more opaque.

This “black box” nature poses significant challenges for developers and companies looking to deploy AI safely and reliably.

A 2024 McKinsey study showed how urgent this problem is and that 44% of business leaders have experienced at least one negative consequence as a result of unintended model behavior.

Goodfire aims to address these challenges by using a new approach called “mechanistic interpretability.”

This field focuses on understanding how AI models reason and make decisions at a detailed level.

Edit model behavior?

Goodfire’s product is a pioneer in the use of interpretability-based tools for understanding and editing AI model behavior. Eric Ho, CEO and co-founder of Goodfire, explains their approach:

“Our tools break down the black box of generative AI models and provide a human-interpretable interface that explains the internal decision-making process behind a model’s output,” Ho told VentureBeat in an email response. “Developers can directly access the model’s internal mechanisms and change how important different concepts are to alter the model’s decision-making process.”

The process, as Ho describes it, is similar to performing brain surgery on AI models. He outlines three key steps:

1. Mapping the brain: “Just as a neuroscientist would use imaging techniques to look inside a human brain, we use interpretability techniques to understand which neurons correspond to different tasks, concepts, and decisions.”
2. Visualize behavior: “After we map the brain, we provide tools to understand which parts of the brain are responsible for problematic behavior. We do this by creating an interface that makes it easy for developers to find problems with their model.”
3. Performing an operation: “With this understanding, users can make very precise changes to the model. They can remove or enhance a specific feature to correct the model's behavior, much like a neurosurgeon would carefully manipulate a specific brain region. By doing this, users can improve the model's capabilities, remove problems, and fix bugs.”

This level of insight and control could reduce the need for expensive retraining or rapid, trial-and-error engineering, making AI development more efficient and predictable.

Building a world-class team

The Goodfire team brings together experts in AI interpretability and startup scaling:

• Eric Ho, CEO, previously founded RippleMatch, a Series B AI recruiting startup backed by Goldman Sachs.
• Tom McGrath, Chief Scientist, was previously a senior research scientist at DeepMind, where he founded the company's mechanistic interpretability team.
• Dan Balsam, CTO, was the founder and engineer at RippleMatch, where he led the core platform and machine learning teams.

Nick Cammarata, a leading interpretability researcher formerly at OpenAI, emphasized the importance of Goodfire's work: “There is currently a critical gap between cutting-edge research and practical use of interpretability methods. The Goodfire team is the best team to bridge that gap.”

Nnamdi Iregbulem, Partner at Lightspeed Venture Partners, expressed his confidence in Goodfire’s potential: “Interpretability is emerging as a critical building block in AI. Goodfire’s tools will serve as a foundational primitive in LLM development, allowing developers to interact with models in entirely new ways. We’re backing Goodfire to lead this critical layer of the AI ​​stack.”

Looking ahead

Goodfire plans to use the funding to expand its engineering and research team and enhance its core technology.

The company aims to support the largest, most advanced open-weight models available, refine model editing functionality, and develop new user interfaces for interacting with the inner workings of models.

As a public organization, Goodfire is committed to advancing humanity’s understanding of advanced AI systems. The company believes that by making AI models more interpretable and actionable, they can pave the way for safer, more reliable, and more useful AI technologies.

Goodfire is actively seeking “agentic, mission-driven, kind, and thoughtful people” to join their team and help build the future of AI interpretability.