An honest record of GeoScript builds: what is working, what is in development, and what is planned. No hype, no futures as facts.
GeoScript is currently in private beta. Access is by invitation. Contact us to inquire about access.
Private Beta
Google AI Overviews — live monitoring
Shipped
All six monitored engines now return real production observations: ChatGPT, Claude, Gemini, Grok, Perplexity and Google AI Overviews.
Google AI Overviews verified on real client prompts
The adapter reads the AI Overview block from live Google results and stores the overview text and every cited source. A prompt for which Google shows no overview is stored as a completed observation (shown: no), never as a failed run. The Engines page shows how often an overview appeared; the answer comparison labels no-overview runs.
Private Beta
Private beta — initial platform build
Shipped
GeoScript's core platform reaches private beta. The agency console, opportunity pipeline, monitoring, optimization, deployment, and reporting verticals are functional end-to-end for invited agencies. This entry documents what shipped, what remains in development, and what is planned.
Agency console
Multi-tenant agency workspace with member roles, tenant isolation enforced at the service layer, and a client seat entitlement model.
Ranked list of businesses with observable AI-visibility gaps. Save, unlock, and work opportunities through the pipeline. Exclusivity enforcement (OPP_EXCLUSIVE_HELD) prevents cross-agency conflicts.
Deterministic, versioned scoring formula with 8 named components, per-component explainability, missing-evidence penalty, and tenant-isolated score persistence. Status: in development — several input signals (CPC, enrichment) await real data sources.
Built-in relationship tracker covering SAVED → CONTACTED → PITCHED → NEGOTIATING → WON → ACTIVE. WON stage does not trigger billing — seat activation is a separate explicit action.
End-to-end prompt execution against real AI engines. ChatGPT (OpenAI), Claude (Anthropic), and Perplexity are live and returning real observations. Gemini, Google AI Overview, and Copilot are in development — coverage will expand as those integrations complete.
Real static-HTTP crawler with SSRF hardening, politeness rules, and deterministic content-hash snapshots. Builds the data foundation for all optimization work.
Deterministic, typed extraction of content blocks, structured data, FAQ items, service area signals, and entity definitions from every crawled page. Versioned representation that never silently overwrites historical snapshots.
Turns Content Genome findings into typed, deterministic recommendations: missing LocalBusiness schema, missing FAQ block, thin content, missing meta description, multiple H1, missing service area block. DB-enforced deduplication.
Standard Script
Lightweight fail-open browser tag applying typed, constrained DSL optimizations from a signed deployment manifest. No arbitrary JavaScript execution, no inline event handlers.
Automated Deployment pipeline
Risk-aware full deployment lifecycle: recommendation → intervention → signed package → risk-gated approval → preflight → activate → independent verification → rollback. State machine verified end-to-end.
Outcome Measurement
Compares AI observations inside a bounded half-open window before and after deployment. Pipeline is working; evidence depth is limited by the 3-of-6 live engines.
Reports — white-label, immutable snapshots
Period-scoped reports with 4 locked narrative sections. Shareable stable-token public URL, agency-branded client-facing page (GeoScript not mentioned). PDF download. DB-enforced immutability prevents post-publish edits.
Natural-language interface for agencies to query client data. Strictly grounded in stored workspace data — no live web queries, no speculation. Deterministic zero-cost refusal when workspace has no evidence. Tenant isolation enforced before data retrieval.
Authoritative definitions for core AI visibility industry terms (AI visibility, AEO, GEO, AI citation, AI share of voice, AI crawler, agent traffic, recommendation rate) and GeoScript proprietary concepts (Content Genome, Opportunity Score, Pitch Room, Business Truth).
These capabilities are on the roadmap but have not started or are not yet available in private beta. They are listed here for transparency, not as commitments.
Competitor Intelligence
Extraction of which competing entities AI engines recommend in response to the prompt universe, per engine and over time. Not yet started.
Citation Intelligence
Preservation and ranking of ordered citations each engine surfaces, for measuring which sources drive AI recommendations. Not yet started.
Agent Traffic
Server- and CDN-level visibility into AI crawler and agent requests, with identity verification and purpose classification. Depends on Advanced Edge adapter, which is also not yet started.
Advanced Edge (Cloudflare adapter)
Origin-side or CDN-worker integration. Required for Agent Traffic and Advanced Agent Delivery. Not yet started.
Content Studio
Human-review surface for editing, approving, and versioning recommendations before deployment. Not yet started.
White-Label Distribution
Agency-branded client-facing surfaces backed by GeoScript infrastructure. Not yet started.
Private beta notice
GeoScript is in private beta. APIs, data models, and capabilities may change without notice. Published documentation reflects the current state of the build, not future promises. When capabilities change from planned to available, this changelog will be updated with an honest assessment of what shipped and how it works.