AI agents now handle 51% of all internet traffic for the first time, surpassing human online activity as automation reshapes how we interact with digital systems. This flooding of the internet with artificial agents creates an urgent need for reliable infrastructure to manage them – and SmythOS believes it has the answer.
The company announced that it’s releasing its core platform as open source software, positioning the move as potentially as significant for AI agents as Linux was for servers. SmythOS is making its runtime environment, software development kit, command line interface and documentation freely available under an MIT licence, with plans to open source its visual agent builder in coming months.
Why Developers Wanted a Change
No ads. No tracking.
We don’t run ads or share your data. If you value independent content and real privacy, support us by sharing.
Read More
Current AI agent frameworks are failing developers when it comes to production deployment. The complaints are consistent across the board: brittle foundations, black box architectures that hinder debugging, constant breaking changes and dependency bloat that makes systems unnecessarily complex.
Developers report major pain points including complex debugging due to lack of visibility, frameworks like LangChain being overly bloated with unused features, and the challenge of maintaining stable dependencies as APIs rapidly evolve. Many struggle to decide on a tech stack amid frequent changes that introduce latency or bugs after updates.
‘Building autonomous AI agents today is far harder than it should be,’ said Alexander De Ridder, CTO and co-founder of SmythOS. ‘Popular agent frameworks are fragile and inefficient for production use. This brittle foundation threatens enterprise adoption, raises security concerns, and invites domination by closed, proprietary platforms.’
The result is that enterprise users demand modular, flexible, secure frameworks with strong governance and integration support, whilst many developers are trending toward simpler, single-task agents and self-hosted tools to improve maintainability.
SmythOS’ Technical Pitch
SmythOS built its Smyth Runtime Environment (SRE) from scratch as what it calls a lightweight, Linux-like kernel for agents. At just 50 megabytes, the SRE can run on anything from cloud servers to a Raspberry Pi or even IoT devices like thermostats. This approach to AI systems development creates what the company terms ‘Goldilocks Agents’ – enough autonomy to matter, with enough control to trust.
‘AI autonomy without a control layer doesn’t work in production. That’s where SmythOS shines,’ said CEO and co-founder, whose name is given as Umansky in the announcement. ‘We didn’t just wrap a visual shell around open source—we built from first principles. SmythOS is a better operating system where humans and agents collaborate securely.’
The platform separates workflow logic from execution logic, similar to how Java changed programming. This separation allows for built-in guardrails for security, observability and governance whilst maintaining the flexibility needed for complex multi-agent systems.
The Open Source Decision
The move to open source reflects a bet on community-driven development over proprietary control. ‘Open source leads to more trust, trust leads to more adoption, which means more contribution, more innovation, more transparency,’ Umansky explained. ‘This is a developer acquisition flywheel.’
SmythOS is embracing open standards including Anthropic’s Model Context Protocol (MCP) and Google’s Agent-to-Agent (A2A) protocol , which are complementary standards designed for AI agent interoperability. MCP standardises secure AI connections to external data and tools, acting as a ‘context bus’, whilst A2A standardises communication between diverse AI agents from different vendors.
‘SmythOS acts like a glue that brings all of these big player systems together,’ De Ridder noted. This approach prevents vendor lock-in and positions SmythOS as interoperable backbone infrastructure. The company is following a similar path to Meta’s open source AGI strategy, betting that open development will win over proprietary alternatives.
Early Adoption and Business Uptake
The platform has already built momentum with over 28,000 agents created and pilot projects secured with government agencies, OEM brands and large agencies. One customer testified to the platform’s stability and ease of use: ‘We rebuilt our agent in a day, nothing broke.’
This early traction validates SmythOS’s enterprise-ready approach in a market where case studies showcase applications in automating workflows, customer interactions and data processing whilst preserving human oversight. The platform supports over 300,000 SaaS integrations and offers flexible deployment options including cloud, on-premise and local setups.
What’s Next
SmythOS is preparing to open source its visual Agent Builder in the coming months, completing the transition of its core platform to community development. The company has ambitious plans to set standards for what it calls the ‘agent-to-agent’ economy.
‘We’re moving to an Internet of AI agents. The agent-to-agent economy will dwarf the human-to-human economy,’ De Ridder stated. ‘We believe that in the next few years, AI agents and AI assistants will do more browsing and online transactions than humans, and there will emerge an agent-to-agent economy.’
The timing appears deliberate. As global competition in AI intensifies, SmythOS is positioning itself as foundational infrastructure for the next wave of automation. Open source software revolutionises digital society and economy , serving as the basis for technologies including Alexa, Android, Azure and AWS.
The company believes this moment is decisive, with the architectures, infrastructure and standards for the next decade of autonomous agents being decided now. With open source acting as the backbone of digital innovation and enterprise growth, SmythOS is making its bet that open development will win over proprietary alternatives in building the foundation for what it envisions as a multi-trillion-dollar agent economy.
