Bevelites GmbH
Bevel provides a vendor-agnostic, Git-backed control plane for enterprise AI agents. It keeps agent context, skills, tools, permissions, and identity in company-owned infrastructure, and builds and operates AI agents inside customers' systems.
At a Glance
- Large enterprises
- Insurance and financial services
- Manufacturing and electronics
- Tech-heavy B2B companies
- +1 more
AI Tools by Bevelites GmbH
(2)
UTCP Code-Mode
AI Agent Tool Workflow Library
Bevel
Enterprise AI Agent Control Plane
Discussions
No discussions yet
Be the first to start a discussion about Bevelites GmbH
Latest News
Ali Raza and Razvan Radulescu scheduled to present the vendor-neutral control-plane thesis at AI Context Hamburg.
Code Mode repository received a major release/update, including the TypeScript library's memory-limit error handling and current 1.x package line.
Hexis, the open-source core of Bevel, launched on Product Hunt.
Bevel/UTCP presented at the MIT Next Generation Communications Forum on integrating LLMs and agents securely in organizations.
Products & Services
Vendor-agnostic, Git-backed control plane for enterprise AI agents. It manages versioned knowledge/context, Markdown skills, tool manifests, access rules, per-agent identity, and governed changes across agent runtimes.
Apache-2.0 open-source core of Bevel: a Git-backed home for AI skills, tools, and knowledge, with review, access control, versioning, MCP connectivity, a self-hosted deployment path, and a live demo.
Open protocol and tool-description standard enabling agents to discover and call tools directly through native protocols without wrapper servers; supports HTTP, CLI, MCP, text, and other protocol plugins.
Open-source tool-chaining libraries that let agents execute TypeScript or Python code to call and combine multiple tools in one workflow rather than making sequential tool calls. Distributed through TypeScript, Python, CLI, and MCP packages.
Market Position
Bevel positions itself against proprietary AI-agent consoles and vendor-specific harnesses by keeping context, skills, tools, permissions, and identities in customer-owned infrastructure. Its UTCP layer is presented as an open alternative to MCP with direct native tool calling, while Hexis adds a Git-backed governance layer. In legacy-code understanding, Bevel differentiates through deterministic static analysis and business-logic context rather than relying only on non-deterministic LLM-generated diagrams; it is complementary to GitHub Copilot, Cursor, and code-quality tools such as Sonar, and competes with consultancies and specialized code-understanding products.
Leadership
Founders
Ali Raza
Co-founder and CEO. His background includes Management and Technology at TUM/CDTM, operations and strategy, and experience at PepsiCo, KSB, Fernride, Alasco, and UVC Partners.
Juan Ignacio Viera Garcia
Co-founder and CPO. He directs product strategy and has research experience at MIT, finance experience at CIBC and BNP Paribas, and startup/venture roles at Arive, ClearGlass, and Hi Inov.
Razvan Ion Radulescu
Co-founder and CTO. His background combines TUM/Cambridge academic work, full-stack software development, product management, and authorship/maintenance of the Universal Tool Calling Protocol (UTCP).
Executive Team
Ali Raza
Co-founder and CEO
TUM/CDTM Management and Technology alumnus; operations and strategy lead with prior experience at PepsiCo, KSB, Fernride, Alasco, and UVC Partners.
Juan Ignacio Viera Garcia
Co-founder and CPO
Product-strategy lead with MIT research experience, prior finance roles at CIBC and BNP Paribas, and startup/venture experience at Arive, ClearGlass, and Hi Inov.
Founding Story
Bevel began as a CDTM Spring 2023 team project focused on making legacy code understandable and modernizable. The founders initially developed AI-powered code visualization and business-logic understanding tools, then expanded the vision toward a vendor-neutral infrastructure layer because enterprise agents otherwise leave context, procedures, tools, and permissions fragmented inside vendor products.
Business Model
Revenue Model
Enterprise build-and-operate partnership: Bevel finds AI opportunities, builds agents in the customer's infrastructure, and operates and maintains them long term. The company also offers its enterprise control-plane platform and related software consulting services; Hexis is open source and self-hostable.
Pricing Tiers
Pricing is organization-specific; the company directs prospective customers to contact Bevel for an engagement.
Target Markets
- Large enterprises
- Insurance and financial services
- Manufacturing and electronics
- Tech-heavy B2B companies
- Organizations with legacy software and technical debt
- Enterprise AI transformation and operating agent fleets
- RFI, RFP, and tender management agents
- Go-to-market campaign design and execution
- Market intelligence agents
- Electronics procurement over large parts datasets
- Legacy-code documentation, maintenance, migration, and modernization
- Unite Procurement Deutschland AG
- EGYM SE
- Workpath GmbH
- Luminovo GmbH