AIOA Aegis / Business-impact-first AIOps decision platform
After an equipment alert, operators check the related business probes and confirmed dependencies, then use monitoring issues, assets and topology to narrow the investigation. Inspections and the assistant provide reviewable evidence and suggestions. People decide how to act. Dependencies need human confirmation; insufficient evidence or no identified impact does not prove there is no fault.
Content reviewed · v1.4.3
Business owners and operations staff first define which business services are to be protected, then configure probes and confirm dependencies. Device discovery and dependency candidates provide leads for review; only relationships confirmed by a person take part in business-level judgments. When relationships are missing or a probe is unavailable, restore the business mapping or data collection first, then assess impact.
After a device alert, review Zabbix problems, metrics, and history in the context of the related assets and business. Check whether the anomaly matches changes in the business probes, and use the topology to judge the scope of what is affected. A device alert and a business outage are different facts; and not identifying business impact is no proof that nothing is wrong on site.
Obtain reviewable evidence through inspection runs or the restricted assistant tools, then interpret it and follow up. Collection and interpretation are stored separately; if a batch partially fails, the results already obtained and the undetermined items are both kept. As long as the raw collected evidence is still available and the job state allows it, parsing can be re-run without collecting again. Items lacking evidence or with failed parsing still need further investigation.
Operations staff decide how to remediate based on the evidence, recording ownership, progress, and the re-check result. What the assistant produces is an advisory change draft; even once approved, the platform will not execute that draft automatically. Existing management and device configuration features have their own permissions and operational limits. After remediation, business probes and related problems still need to be verified, producing a traceable record.
IT operations teams, information centers and service providers connecting infrastructure conditions with business service health.
Private deployment uses PostgreSQL and pgvector. Configure access to monitoring, assets and site accounts with appropriate permissions. Discovery methods, sites and collection protocols depend on the environment.
Approving an assistant’s proposed change does not automatically execute it. Some management and equipment configuration features perform writes; the whole platform is not read-only. Demonstrations are not production acceptance evidence.
HONGCHUANGDA TECHNOLOGY
Three independent products, deployable separately. Cross-product integration is scoped per project.
