Changelog
What’s new in Autonoma
See the latest features, improvements, and updates across Autonoma, and discover what’s changing to make AI workflows smarter and easier.
Smarter confidence-based approvals
Agents now have more flexible confidence controls, so teams can decide exactly when an action should run automatically and when human approval is required.
This update lets you set confidence thresholds for individual actions, workflows, or risk levels. High-confidence tasks can move forward instantly, while uncertain or sensitive actions are paused and routed to the right reviewer with full context.
More control over autonomous actions
Different actions carry different levels of risk. A support agent answering a routine question should not follow the same approval rules as an agent issuing a refund or changing an account.
You can now define separate confidence thresholds for specific actions and decide whether the agent should proceed, request approval, or escalate.
Better context for reviewers
When an action requires review, Autonoma now includes the context behind the decision.
Reviewers can see the agent’s confidence level, relevant source information, proposed action, and why approval was requested without repeating the agent’s work.
Getting started
Open your agent configuration and go to Guardrails & Review. From there, choose an action and define the confidence level required for automatic execution.
You can also set a lower threshold for escalation and specify which person or team should receive review requests.
Existing agents will continue using their current approval settings until you add confidence-based rules.
Replay any agent run step by step
ou can now replay completed agent runs to understand exactly how an agent moved through a workflow, which tools it used, and where each decision came from.
Run Replay gives teams a clearer way to investigate unexpected outcomes, review escalations, and improve agent behavior without digging through separate logs.
See every step in context
Each replay shows the agent’s inputs, tool calls, decisions, confidence, and outputs in the order they happened.
You can move through the run step by step and quickly identify where a workflow changed direction.
Compare successful and failed runs
Use replay to compare similar runs and understand why one completed successfully while another required escalation or failed.
This makes it easier to identify missing context, weak instructions, unclear decisions, or unreliable workflow steps before they affect the final outcome.
Improve without guessing
Once you find the problem, jump directly to the relevant agent configuration and update its instructions, permissions, or workflow logic.
Changes can then be tested before publishing a new version.
Getting started
Open any completed agent run and select Replay run from the activity view.
Scheduled research briefings
Research agents can now monitor selected sources and automatically deliver structured briefings on a schedule you define.
Choose the topics, competitors, websites, or internal sources you want tracked, then run briefings daily, weekly, or at a custom interval.
Focus on what matters
Autonoma compares findings across sources and prioritizes the updates most relevant to your research criteria.
Ready when your team starts
Briefings are generated automatically and organized into key findings, source references, and notable changes.
Getting started
Open your Research Agent, select Schedule, choose your sources and frequency, then publish the briefing workflow.
