Ava — The Intelligence Layer

Ava is AVO’s intelligence engine — the layer between measurement data and practitioner action.

The division of labour

AVO’s architecture separates measurement from intelligence deliberately. Karsa measures: it runs AS scans, fires VS probes, computes scores, and stores results. Ava interprets: it reads Karsa’s output and produces recommendations. The two engines share data but never share responsibilities. Karsa does not tell practitioners what to do. Ava does not run scans.

This separation matters because the two problems require different architectures. Measurement is deterministic — the same domain under the same conditions produces the same score. Intelligence is contextual — the right action for a domain depends on industry, maturity stage, competitive position, and what has already been tried.

What Ava produces

Cards are Ava’s atomic output — one specific, actionable recommendation for one domain. Cards come in three tracks:

Fix track — a Datapoint scored below threshold. Ava identifies the specific structural problem, provides remediation guidance (code, content, or configuration), and projects the score improvement based on IKL network data from comparable domains.

Content track — a content gap identified from VS probe analysis. AI platforms consistently fail to mention the brand on a topic that matters to the Focus. Ava surfaces the gap, generates a Content Brief, and tracks whether publishing the resulting content moves VS.

Strategic track — cross-network intelligence that requires IKL context. A competitor improved their Knowledge Graph depth. Domains in the same industry that fixed entity-schema saw +12 VS within 60 days. These insights require network data; a single domain’s history cannot produce them.

Account Insights are Ava’s portfolio-level view — patterns across all domains in an account, benchmarked against IKL industry data.

The IKL dependency

Ava’s projections are only as credible as the IKL data behind them. A Fix card that says “fixing entity-schema typically improves VS by 8–15 points within 60 days” is only meaningful if that estimate comes from real outcome data across comparable domains. The IKL is what converts Ava from a heuristic advisor into an evidence-grounded one.

What Ava does not do

Ava does not modify scores. It does not run scans. It does not write to Karsa’s measurement tables. Scores in, intelligence out — the boundary is absolute.