Enterprise moves to always‑on AI agents
- Adesso presented adSCAILE, a delivery model that treats AI agents as continuous workers supervised by humans. - Their pilots reportedly raised release cadence from one biweekly release to two releases per week. - OpenAI is also launching workspace agents that automate team workflows, shifting competition toward workflow control. (theverge.com)
Companies are starting to buy AI agents as always-on workers, not just chat tools. Adesso says its adSCAILE model puts agents on continuous software delivery while humans keep control of requirements, architecture, and final approval. (adesso.de) Adesso describes adSCAILE as an end-to-end software development process in which humans define the “why” and “what,” then agents generate code, run tests, fix bugs, and draft documentation in long-running loops. The company says those implementation cycles run in two to four days and can work on new systems, older “brownfield” software, and legacy modernization projects. (adesso.de) The practical change is release speed. Adesso’s older global delivery model centered on functional software increments every three weeks, while adSCAILE pitches much shorter cycles measured in days rather than weeks. (adesso.de, adesso.de) An AI agent in this setup works more like a cloud contractor than a chatbot: it keeps executing tasks after a person stops typing. OpenAI used the same framing on April 22, saying its new workspace agents in ChatGPT can handle “complex tasks and long-running workflows” and keep working in the cloud when users are away. (openai.com) OpenAI’s product moves the contest from model quality alone to workflow control inside the workplace. The company said teams can share one agent across ChatGPT or Slack, connect it to internal systems, and let it gather context, request approvals, and move work across tools under company permissions. (openai.com) That puts service firms like Adesso and platform vendors like OpenAI on overlapping ground. Adesso is selling a delivery method for software projects, while OpenAI is packaging agents as reusable team infrastructure for sales, reporting, information technology requests, product feedback routing, and risk reviews. (adesso.de, openai.com) Adesso has been building toward that position for months. In November 2025, it announced an “Agentic Enterprise Fabrics” initiative with Salesforce focused on helping customers deploy AI agents inside business processes, especially in regulated environments that need governance and compliance controls. (adesso.de) The unresolved question is whether companies trust agents with enough access to make the speed gains real. Both Adesso and OpenAI are answering that with the same structure: agents do the repetitive execution, and humans keep the policy, architecture, review, and acceptance gates. (adesso.de, openai.com)
