Back to Case Studies

Innerworks + HumblebeeAI

From Collaboration to Innovation: Inside the Innerworks and HumblebeeAI Partnership

About Innerworks

Innerworks is a cybersecurity / behavioral intelligence company focused on helping exchanges, finance platforms, and digital ecosystems identify and block bad actors. Their solution stack includes Account Takeover Protection, Geo Compliance, Bot detection, and deeper behavioral analysis, leveraging 200+ data points across device, network, and behavioral signals. (Innerworks)

They also employ a “RedTeam ethical hacking platform” (built on BitTensor) to stay ahead of AI-driven fraud techniques. (Innerworks)

Starting Point / Challenge

When our collaboration began in January 2024, Innerworks had a single SDK (a foundational detection and fingerprinting component) but lacked a fully integrated solution. Internally, they lacked a dedicated R&D team, senior engineering leadership in fraud & AI, and the infrastructure to grow modules into a cohesive, scalable platform.

They faced a high-stakes landscape: regulations for exchanges were tightening, forcing platforms to proactively prevent fraud and sanction breaches, rather than just reacting. Meanwhile, increasingly sophisticated fraudsters, especially using generative AI and identity spoofing, were outpacing legacy detection systems.

Our Partnership Model & Work Approach

We stepped in not as an outsourcing shop, but as their R&D & product arm. Innerworks defined the vision and product goals; our role was to design, research, prototype, iterate, and build those capabilities in collaboration with the other teams of Innerworks.

Key principles of engagement

  • Deep domain research: industry best practices, academic papers, red-team methods, advanced heuristics.
  • Transparent knowledge transfer: all research, findings, designs, code, and docs are shared in their repositories, fully accessible. This ensures trust and enables seamless eventual transition or scaling by their team.
  • Successive iterations: we launched MVPs, collected feedback, adapted to adversarial behavior, and hardened models.

Innerworks + HumblebeeAI partnership cover

What We Delivered / Built

We structured the architecture of Device Fingerprinting, VPN / True Location Detection, Bot Detection / Behavioral Profiling, and RedTeam & Adversarial Testing Framework to be modular for research and development (R&D) purposes, while also supporting collaboration for production integration. Each component is designed to operate independently or feed into composite risk scoring flows. We have built pipelines for internal data ingestion, feature extraction, model serving, API endpoints, and monitoring, ensuring flexibility for both R&D experimentation and seamless production deployment through collaboration.

Throughout, we prioritized observability, scalability, reliability, and defensibility so that Innerworks could offer credible enterprise-grade services.

Results & Impact

  • In August 2025, Innerworks raised $4M in a Seed round led by AlbionVC, demonstrating investor confidence in their technological foundation and trajectory.
  • They now maintain a stable client base across exchanges and digital platforms.
  • They use their full suite: device fingerprinting, VPN detection, bot detection, and threat intelligence – capabilities we co-built and matured.
  • Importantly, we ensured that all knowledge, research, code, documentation, and design artifacts remain in their repos (not ours). This fosters mutual trust and enables seamless handover or future scaling by their internal team.

Why This Partnership Works & Why Others Should Care

  • We act as a technical cofounder, not just an external contractor.
  • We anchor on research-driven approaches (not just copying tools), enabling more innovativeness and robustness.
  • We commit to transparency and handover – the client retains full ownership, no black box.
  • Our model is flexible: from ground-up builds to module extensions, for startups, regulated platforms, or government programs.

Ideal Client / Engagement Profile

We are best suited for:

  • AI-centric startups that need deep technical infrastructure but lack the full internal team.
  • Platforms with high stakes that require defensible, reliable anti-fraud / trust systems.
  • Ethical, visionary ventures who value transparency, long-term partnerships, and technical depth (like Innerworks).

Engagement Models

  • 6–18 month R&D engagements where we build and hand over.
  • Ongoing co-development / extended engineering partnership.
  • Module augmentation (you have some infrastructure; we plug in capabilities).