SolarWinds launched Avior on Oct. 14, 2026, introducing an agentic AI system designed to pursue operational goals rather than respond to individual prompts. The company calls its approach deterministic agentic AI, combining specialized agents with defined controls for enterprise IT environments.
The initial rollout is narrower than its broader ambitions. Early access focuses on network engineers using SolarWinds Observability Self-Hosted, with guided onboarding and a curated set of pre-validated goals.
According to SolarWinds' launch announcement, Avior can coordinate agents, investigate problems, take permitted actions, and measure whether operational conditions improve.
What SolarWinds Avior AI agents can do
Avior works toward ongoing objectives rather than completing one request at a time. Teams define the desired outcome, accessible data sources, permitted actions, and approval requirements.
Role-based agents then gather evidence, investigate problems, test hypotheses, and verify results. These capabilities build on agentic AI systems that coordinate multiple tasks using connected tools.
SolarWinds identifies five operational use cases:
- Alert management: Correlate related events into actionable incidents.
- Compliance monitoring: Audit devices against policy baselines.
- Configuration management: Detect drift and maintain backups.
- Capacity planning: Track utilization and costs.
- Asset management: Identify outdated or unknown infrastructure records.
According to the company's technical introduction, Avior operates through supported product connections rather than accessing arbitrary machines. Third-party connectivity can use Model Context Protocol, although supported integrations depend on the release.
SolarWinds describes three autonomy levels: Conservative requires approval for environment changes, Balanced permits routine reversible actions, and Autonomous allows configured actions within policy limits.
The most restrictive setting across the organization, agent, and goal takes precedence.
Actions inherit the connected user's permissions, while unrecognized tools are held for approval. These restrictions address risks associated with AI agent security, including unauthorized changes.
SolarWinds says customer information is masked, encrypted, and excluded from training underlying AI models. Recovery and rollback capabilities vary by workflow.
What eWeek found: four tests before trusting Avior
SolarWinds has documented Avior's intended controls, but its launch materials provide no independent performance benchmarks, quantified customer results, or general-availability date. Pricing and the complete list of supported integrations also remain undisclosed.
The initial release depends on Observability Self-Hosted and Platform Connect. That limits how thoroughly organizations can assess its promised cross-environment capabilities.
For eligible customers, four tests could establish whether Avior provides measurable operational benefits without introducing unacceptable risks.
1. Test alert reduction without losing critical signals.
Compare Avior-assisted triage with existing workflows using the same incident scenarios. Measure investigation time, correctly grouped alerts, and critical events missed or incorrectly deprioritized.
A smaller alert queue is not an improvement if important incidents disappear from view.
2. Verify the available integration coverage.
Identify which monitoring, configuration, and ticketing systems Avior can actually access during early access. Confirm whether each proposed goal has sufficient data and whether stale or missing information interrupts its investigation.
Broader hybrid-environment support should be evaluated only when the required connections become available.
3. Challenge the permission boundaries.
Use a controlled test account to attempt an action outside its permissions, then check whether Avior blocks execution, requests approval where appropriate, and records the outcome.
Test the autonomy settings against SolarWinds' published definitions. A policy setting is useful only when the system enforces it during real workflows.
4. Verify recovery before allowing changes.
Test a reversible configuration action in a nonproduction environment. Confirm whether Avior can restore the previous state or requires an operator to execute a documented recovery procedure.
SolarWinds acknowledges that rollback behavior depends on the workflow, making this especially important before enabling unattended remediation.
Avior's guided onboarding program provides an opportunity to establish these baselines before expanding automation. Organizations should retain read-only or approval-required operation wherever permissions, recovery, or reliable outcomes remain unverified.
Steven Khan, an observability engineer at Radiology Partners, described faster triage in SolarWinds' announcement, but no supporting measurements were disclosed. Broader customer results will be needed to assess performance outside the initial validated workflows.
OpenAI agents may also have contributed to a recent Wikidata outage, another case raising questions about the boundaries of automated activity.
Want to learn more AI tips, tricks, and prompting techniques? eWeek readers get free 7-day access to The Neuron Academy, our practical learning platform designed to help professionals use AI more confidently at work. Browse all lessons →
Let us teach you How to Talk to AI for free! Try our six-minute course at The Neuron Academy and learn a few simple ways to write better prompts and get more useful results from AI, or browse our other AI course for free for seven days. Check out all the lessons here →