Sycamore: Enterprise AI Agent Operating System
Sri Vishanath, former Atlassian CTO, discusses building Sycamore, an operating system for enterprise AI agents. The analysis covers the shift from SaaS to agentic software, the critical role of governance in scaling autonomy, and the strategic imperative for enterprises to adopt systems thinking to navigate the AI transformation.
The Rise of the Agentic Operating System
The enterprise software landscape is undergoing a fundamental shift from static SaaS applications to dynamic, agentic operating systems. Sri Vishanath, former CTO of Atlassian and current CEO of Sycamore, argues that traditional software models are decaying and must be replaced by platforms that build, deploy, and maintain autonomous agents. Sycamore positions itself as an "operating system for enterprise AI agents," offering a modular framework that allows large enterprises to build custom SaaS solutions with full compliance, security, and auditability. This approach addresses the critical gap where existing SaaS products bolt on AI features, resulting in bloated and inefficient systems that fail to meet the rigorous demands of enterprise governance.
Governance as a Strategic Enabler
A core strategic insight from the discussion is that governance is not a barrier to AI adoption but a prerequisite for speed. Vishanath emphasizes that without robust verification, validation, and guarding mechanisms, enterprises cannot trust agents to operate autonomously. Sycamore’s "progressive trust" model allows agents to gain permissions over time based on verified performance, mirroring human onboarding processes. This framework enables companies to scale autonomy safely, ensuring that agents adhere to corporate values, security protocols, and compliance standards. The integration of governance directly into the development pipeline allows for faster iteration and deployment, turning regulatory constraints into competitive advantages.
The Imperative of Systems Thinking
The transition to agentic software requires a shift in executive mindset from isolated workflow automation to comprehensive systems thinking. Vishanath advises that companies must step back to design the entire operational system, identifying common knowledge bases and high-risk areas before deploying agents. This holistic approach ensures that automation solves multiple problems simultaneously, creating a durable moat. Furthermore, the role of the engineer is evolving from code writer to agent manager, requiring strong systems design skills to orchestrate complex, multi-agent workflows. Enterprises that fail to adopt this systems-level perspective will struggle to realize the full potential of AI, remaining stuck in inefficient, fragmented automation efforts.
Strategic Implications for Leadership
For CTOs and executives, the immediate action is to move beyond experimental AI pilots and invest in foundational agentic infrastructure. The "last mile" gap between AI capabilities and human business needs remains a critical bottleneck, requiring forward-deployed engineers to bridge the translation layer. Companies that proactively adopt agentic operating systems and flatten their organizational structures will be better positioned to navigate the rapid changes in the next five years. The era of bolt-on AI is ending; the future belongs to platforms that are agentic by design, offering living software that continuously improves and adapts to business needs.
Key insights
-
Traditional SaaS models are becoming obsolete as they cannot natively support the dynamic, autonomous nature of AI agents. Enterprises require new operating systems that build, deploy, and maintain agentic applications with enterprise-grade security and compliance.
Impact: Companies that stick to legacy SaaS with bolt-on AI will face higher costs and lower efficiency compared to those adopting native agentic platforms.
-
Governance is a critical enabler of AI speed, not a hindrance. By implementing robust verification and progressive trust models, enterprises can safely scale agent autonomy, reducing risk while accelerating deployment.
Impact: Organizations with strong governance frameworks will be able to deploy AI agents faster and more reliably than competitors relying on ad-hoc security measures.
-
The bottleneck in AI adoption has shifted from technical capability to systems thinking. Executives must design holistic operational systems rather than automating isolated workflows to achieve scalable, long-term value.
Impact: Companies that adopt systems thinking will create durable competitive moats, while those focusing on point solutions will struggle to scale their AI investments.
-
The human-to-agent ratio is rapidly increasing, with engineers now managing multiple agents simultaneously. This shift requires a new skill set focused on orchestration, verification, and systems design rather than manual coding.
Impact: Engineering teams will become smaller but more powerful, with a focus on managing agent workflows and ensuring output quality through automated verification.
-
Enterprise go-to-market strategies must evolve to combine enterprise sales for initial deployment with self-service expansion for internal adoption. This hybrid model accelerates value realization and reduces friction in scaling AI solutions.
Impact: Vendors that offer self-service expansion will see faster customer growth and higher retention rates compared to those relying solely on traditional enterprise sales cycles.
Action items
-
Implement a progressive trust framework for AI agents, starting with restricted permissions and expanding autonomy based on verified performance metrics and compliance checks.
Impact: This approach reduces security risks while allowing agents to become more effective over time, enabling safer and faster AI adoption.
-
Conduct a systems design audit to identify common knowledge bases and high-risk areas across the organization before deploying AI agents for workflow automation.
Impact: This holistic approach ensures that automation efforts are scalable and create long-term value, avoiding the pitfalls of fragmented, point-solution automation.
-
Shift engineering hiring and training to focus on systems thinking, agent orchestration, and verification skills, rather than traditional coding proficiency.
Impact: This prepares the workforce for the evolving role of engineers as agent managers, ensuring the team can effectively leverage AI to increase productivity.
-
Adopt a hybrid go-to-market strategy for AI products, using enterprise sales for initial deployment and self-service mechanisms for internal expansion and feature adoption.
Impact: This model accelerates value realization and reduces friction in scaling AI solutions, leading to faster customer growth and higher retention.
-
Flatten organizational structures by using agents for communication and coordination, reducing the need for middle management layers and improving decision-making speed.
Impact: This leads to a more agile and responsive organization, capable of adapting quickly to market changes and leveraging AI for operational efficiency.
Quotes
“So governance gives you speed.”
“We are building a builder for custom SaaS, exactly.”
“The bottleneck becomes, what do you build?”