BLUN Agents are coming soon. They are meant to complete clearly defined tasks with permitted tools and deliver a verifiable result. You set the permissions and approvals.
Announced, not yet available.A general chat becomes an accountable workflow.
Each agent gets a role, goal, area of responsibility, model, tools, and limits.
This structure keeps an agent from wandering outside its remit. Tasks, permissions and expected results all stay traceable.
An agent works on a task over multiple steps and checks the result against the goal and review criteria.
An agent isn’t considered finished just because it has produced a convincing-sounding answer.
Autonomy doesn’t mean unlimited access. Critical actions can be blocked or made subject to confirmation.
Typical limits are read-only access, changes in one project only, or no publishing without approval.
Planned, not yet available.
Support agent
Recurring customer enquiries.
Development agent
Clearly defined software tasks.
QA agent
Checks and deviations.
Research agent
Structured information gathering.
Editorial agent
Content and publications.
Project agent
Tasks, risks and progress.
Operations agent
Internal workflows.
Data agent
Review, extraction, and analysis of data.
Which actions are actually allowed is set by the configuration and the permissions of the respective tool.
That lets you automate work without quietly handing responsibility and control over to a model.
Collapsed until you need them.
BLUN Agents are autonomous AI roles for recurring and multi-step work. An agent receives a concrete goal, a clear area of responsibility, the right BLUN model, permitted tools, and binding limits. This page describes how Agents are meant to work. Agents are coming soon.
They can be started for individual tasks or set up permanently for a clearly defined area.
King for complex and highly interdependent tasks; Voice for spoken input, telephony, and voice-based agents; Imagine for visual tasks and image production.
An agent can work with a fixed model or, within a controlled workflow, pass tasks on to a suitable performance tier. That way, maximum performance is used only where it is needed.
Through defined tools and the BLUN MCP Server, agents can, for example, access files, databases, APIs, and internal business functions.
It should be possible to split complex workflows across several specialized roles: one plans, one handles the specialist implementation, one checks the quality. A clear division of roles prevents duplicate work and conflicting changes.
Possible teams: product management, development, and quality assurance; research, editing, and approval; support, technical analysis, and escalation; campaign planning, copy, and image production; data validation, reporting, and management summary.
Four plans: Starter, Pro, Pro Max, and Enterprise. All prices excl. VAT. All four plans are on the central BLUN waitlist. Agents have been announced and are not yet available. The full overview with all features is on the pricing page.
One account for everything in chat.blun.ai: Chat and Work, Sites, Code, images, and voice. Sign up for the right plan or start right away with BLUN Agents.
Try out Agents