Sonar releases tools to verify code in agentic development
Following the recent launch of Sonar’s framework for software development in the age of AI, the Agent Centric Development Cycle, the company has announced the open beta of three new products to autonomously verify code in agent-driven development.
AI agent coding are reinventing the way software is built, generating massive amounts of code at a breakneck speed. But testing and verification have begun to become bottlenecks to software delivery. To meet that gap, Sonar has taken a new approach to AI code trust and verification, based on guidance, verification and resolution.
The first new release for guidance is Sonar Context automation, which the company said “injects project-specific standards and architecture into the AI agent’s workflow.”
Verification is realized with SonarQube Agentic Analysis, that integrates Sonar’s analysis engine into the agent’s actions so vulnerabilities and logic errors can be caught as the code is written.
Finally, those issues are remediated in SonarQube Remendiation Agent, which the company said generates verified fixes for issues in pull requests, as well as removing technical debt from backlogs. Further, the analysis engine is used to confirm the issue was resolved without introducting new ones.
To learn more, read the Sonar blog announcement.
Nutrient expands AI Assistant, automating multi-step document workflows
Document intelligence company Nutrient has added agentic document editing to its AI Assistant that can access the company’s document tools to complete end-to-end document workflows inside any application.
The company described the capabilities and features of the document editing agent:
- Agentic execution inside the viewer: A reasoning layer plans work, selects tools, and carries multi-step document tasks through to completion using the viewer’s high-fidelity rendering and document structure.
- Observe, verify, adapt: The agent executes actions in sequence, observes results, verifies outcomes, and adjusts as needed across extraction, form filling, annotation, and redaction workflows.
- Staged changes with approval: Edits are staged for review and applied permanently only after approval, keeping humans in control of the final output.
- Policy-governed autonomy: Organizations define what the agent knows (skills), how it operates (context), and what it can do (approval policies) – from fully autonomous actions to hard prohibitions.
- Not a single-shot feature: A tool-using system designed to execute document work reliably inside the application where users already operate.
To learn more, watch a demo or visit https://www.nutrient.io/sdk/
OutSystems Introduces Agentic Systems Engineering
OutSystems, creators of an AI development platform with a low-code foundation, today announced Agentic Systems Engineering, designed to help organizations build, manage, and evolve governed agentic systems for the enterprise.
According to the company, Agentic Systems Engineering “addresses the technical complexity and fragmented architectures of modern enterprises that are often difficult to reason over. By developing the rich context and guardrails agents require, it ensures the delivery of reliable, secure, and compliant systems.” A new Enterprise Context Graph provides an understanding of the enterprise architecture — applications, agents, workflows, data and their dependencies — to enable agents to work effectively across systems.
The updated Outsystems Mentor now delivers conversational application generation for teams to build these complex systems with architectural coherence and token efficiency, the company said, noting that these capabilities transform developers into highly productive architects.
OutSystems expects to open an early access program to customers in the second quarter of 2026.
To learn more about OutSystems Agentic Systems Engineering, click here.