Sente Labs Open-Sources Open Executive, an AI Agent for the C-Suite

On October 19, US research-and-development company Sente Labs released Open Executive, an open-source project that lets organisations hand parts of the C-suite to configurable AI agents. According to Raconteur, it is less a single model than a persona-driven framework that can be adapted to different leadership roles and levels of authority.

Co-founder Alex Cohen says the system can be configured to replicate roles such as head of marketing, director of software engineering or project manager. It offers three authority levels: fully autonomous, where an agent effectively replaces an executive; a “duplication” mode that creates a digital double synced with an executive's personal data and apps such as email, Microsoft Teams and Jira; and an executive-assistant mode for scheduling, coordination and routine correspondence.

Open Executive connects natively to Salesforce, Microsoft Teams, Jira and Confluence, according to Cohen, giving agents access to real-time operational data. Sente Labs also says the project was trained on accounting practices, business ethics, regulations and case studies in an effort to embed explicit ethical guardrails in executive decision-making.

The company highlights a “dream” capability, in which agents periodically review their plans, identify gaps in their knowledge and research competitors online, then write the findings into long-term memory. Cohen concedes that this self-reflection is still bounded by existing data and falls well short of human executive judgement.

Outside experts are considerably more cautious. Nicos Savva, a professor of management science at London Business School, called full CEO replacement “very unlikely with current (or foreseeable) technology”, arguing that executive work depends on tacit knowledge that cannot be fully codified; the realistic path is augmentation, not automation. Dr Ying-Ying Hsieh of Imperial Business School similarly expects AI to support executives rather than replace them.

Trust and accountability remain unresolved. Pedro Varela, head of AI at consultancy Slalom, notes that trust in executive leadership involves a specific responsible party and real stakes, which is different from trusting a system in other contexts. The article also asks who a regulator or court would hold accountable when an agent's decision causes harm, and warns that compliance alone does not guarantee safety.

None of that has stopped interest in moving agents up the org chart. Hatim A. Rahman of the Kellogg School of Management argues that a CEO with sound judgement can direct agentic AI to process data, stress-test ideas and prototype solutions at speed and scale. Open Executive's significance is that it turns the debate from whether AI can do executive work into which authorities companies are willing to delegate.

What to watch next is whether real deployments emerge, how open-source reviewers assess the project's permission controls and data access, and how regulators assign liability for actions taken by AI agents. For Open Executive, the binding constraint may not be model capability but whether an organisation is willing to pay for its mistakes.