One audit for modern visibility
Seryvon brings technical search checks, content signals and AI discovery measurements into one website audit. It helps you investigate what a crawler can access, what an answer engine can extract and what an autonomous agent can understand. Seryvon is the product; Powehi is its publisher. The public v1.0.0rc4 release is available for self-hosted evaluation; the managed product is coming soon.
Product capabilities
The audit follows a concrete sequence: discover pages, crawl the site, extract signals, evaluate rules and assemble a report. The audit implementation connects those steps with optional external measurements. You can inspect the logic behind a finding rather than relying on a score without an explanation.
- SEO: investigate crawlability, indexability, metadata, page structure and technical signals.
- GEO: examine content and entity signals relevant to generative discovery, with citation measurements when configured.
- GSO: assess signals associated with Google AI search and provider-dependent observations.
- AEO: examine the structure and supporting context of direct answers.
- ASO: inspect agent readiness, including action descriptions, forms and machine-readable interfaces.
From a finding to a practical decision
For example, a team preparing a service page can first inspect whether its canonical, headings and internal links describe the intended page clearly. It can then review whether the page answers a visitor’s main question and exposes useful machine-readable information. These are suggested review steps, not a claim that a particular website has passed an audit.
Seryvon turns warning and critical criteria into a prioritized action list. The issue prioritization code uses criterion impact, severity and estimated effort. Treat that ordering as a starting point: your team still decides which changes fit the site, available resources and business priorities.
Read scores alongside their coverage
The scoring calculation is deterministic for the same signals and configuration. A fresh crawl or provider response can change those inputs. Unavailable measurements are marked not_measured; they are not invented. The current engine also multiplies each measured pillar average by its coverage, so incomplete measurement can lower the displayed score. A score is therefore most useful together with the measurement profile and coverage label.
The scoring engine documents that calculation. Audit comparison checks profile compatibility before interpreting changes. Adding a connector between runs may change coverage, so an increased score alone is not evidence that a content edit improved visibility.
Product status
The repository contains CLI and Docker workflows, a web dashboard, report generation and comparison code. JSON, HTML and Markdown exports support review and further processing; a PDF report endpoint is also implemented. These source capabilities do not establish that a managed service is available or that every provider configuration has been validated in your environment.
Use the quickstart to evaluate the development version and read the limitations before interpreting results. Seryvon does not guarantee rankings, citations, AI Overview inclusion or completed agent transactions.