Camunda declares ProcessOS, an agentic working system for AI-first enterprise transformation
Agentic orchestration platform supplier has introduced ProcessOS, a brand new intelligence layer that the corporate says leads into “the good re-engineering” of processes and concurrent workflows.
In line with the corporate, to maneuver previous mere AI process help, corporations might want to rethink about which duties needs to be completed, and whether or not brokers or people ought to carry out them. “Simply bolting AI onto legacy processes compounds technical and organizational debt, including complexity, fragility, and value. AI-first course of transformation should work backwards from outcomes, not ahead from present actuality,” Camunda wrote in its announcement.
Constructed for enterprise deployment, ProcessOS ensures full governance and management:
- Verification by design: visible course of fashions clearly present which steps are carried out by AI, beneath which situations, and which steps contain people
- Human within the loop for any change: each course of modification is reviewed and authorized by people earlier than it reaches manufacturing
- Re-use and studying inbuilt: ProcessOS prioritizes authorized course of patterns and connectors, enhancing with every human suggestions loop
ProcessOS runs natively on AWS, with deep integration into Amazon Bedrock and Amazon Bedrock AgentCore for basis fashions, agent reminiscence, identification, and gateway providers. Camunda is obtainable on AWS Market with production-ready reference architectures for Amazon EKS, ECS, and EC2.
ProcessOS is obtainable in closed beta for chosen enterprises starting immediately. Organizations can register their curiosity to take part at camunda.com/process-os.
Hadrian Releases OpenHack, Democratizing AI Vulnerability Discovery
Hadrian has launched OpenHack, an open-source device for AI-powered supply code assessment that works inside Claude Code, Codex, and Cursor.
The temptation once you give a robust LLM a codebase is to let it improvise. “Learn this repo and inform me what’s susceptible.” It should produce one thing. The output might be a mix of believable bugs, hallucinated bugs, actual bugs defined wrongly, and the occasional sharp perception. Triage takes longer than simply studying the code your self.
We’ve discovered two failure modes drive most of that noise:
Unscoped prompts. The agent doesn’t know what query it’s answering, so it solutions all of them at low confidence.
Self-graded findings. The identical agent that proposed the bug decides whether or not the bug is actual. There’s no unbiased verify.
The workflow of OpenHack is designed round fixing these two issues. Critiques are scenario-first: each unit of agent work is strictly one routing unit, one professional, and one proof query. And the agent that proposes a discovering isn’t the agent that admits it.
How OpenHack works:
Situation-based scoping. Each unit of labor is one routing unit, one professional, and one particular proof query. No unscored prompts asking the mannequin to search out something fallacious.
Impartial triage. The agent that proposes a discovering isn’t the agent that admits it. A separate triage agent opinions every candidate in opposition to the unique proof earlier than it turns into a recorded discovering.
Inspectable artifact path. Recon output, situation backlogs, professional outcomes, triage selections, and findings all dwell as plain recordsdata on disk. The complete assessment is auditable finish to finish.
Harness-agnostic, model-agnostic. Runs inside Claude Code, Codex, or Cursor, with any mannequin the harness helps.
OpenHack is obtainable instantly at github.com/hadriansecurity/
Devart releases dbForge 2026.1
The AI-powered capabilities of dbForge
Conversion of pure language to SQL
Context-aware SQL question era
Optimization of pre-written SQL queries
Detection of redundant or lacking indexes
Detailed SQL explanations
Error evaluation and troubleshooting
Built-in AI chat for questions associated to databases and SQL.
dbForge 2026.1 is obtainable immediately for dbForge Edge, dbForge