Uno Data Hub

📡 Studio Live Telemetry

Live telemetry trends for Studio Live sessions, agent interactions, compilation, downloads, and errors using events prefixed with uno/studio-live.

Loading Studio Live telemetry...

Studio User Action Coverage

Are the key Studio user actions from issue #845 tracked as events? Each card shows the tracking verdict and the current count where data exists. CI traffic excluded.

Reached a Running Preview

-

-

ProxyNo dedicated preview event. Proxy = sessions with a successful compilation-succeeded (the rendered preview).

Made an Edit / Iterated

-

-

ProxyNo dedicated edit event. Proxy = sessions with ≥2 compilation-started (iterative compile cycles).

Returned ≥2 Times

-

-

ProxyUsers with ≥2 session-start events in the period. Identity is the anonymous, browser-scoped user_Id that resets per visit (no account identity yet), so repeat visits are undercounted.

Saved a Project

-

-

CanonicalNo in-place save in web Studio Live; the save is a workspace download — project-download-completed with Trigger=Workspace.

Exported to IDE / CLI

-

-

CanonicalBlockedWeb export tracked via project-download-completed (Trigger=Export). IDE-side open correlation (uno/ide-extension/studio-live-project-opened) not yet emitted.

KPI values are coloured against an assumed target shown on each tile: on target near off target. Volume counts (sessions, reach, prompts, Weekly Active Creators, Critical Errors) have no fixed target and stay neutral. Targets are placeholders pending official product/investor goals — edit KPI_TARGETS in studio-live-telemetry-dashboard.js.

Reach DAU

-

-

ProxyDirect session-start events plus Wasm-inferred users. Reach is over-counted until inferred coverage drops; the split below shows the confirmed (direct) and Wasm-inferred counts.

Reach WAU

-

-

ProxyWeekly reach combining direct and Wasm-inferred activity. Fixed complete calendar weeks; the split below shows the confirmed (direct) and Wasm-inferred counts.

Inferred Coverage

-

Proxy% of reach estimated from Wasm events vs direct studio-live events. High value = most reach is inferred, not confirmed.

Activation Rate

-

Canonical% of direct Studio Live users who reached a first-value action (compilation-succeeded). CI traffic excluded; the Wasm-inferred reach baseline was removed so this counts confirmed users only.

Total Sessions

-

CanonicalCount of session-start events (CI traffic excluded).

Engaged Time / Active User

-

CanonicalLatest-day foreground minutes per active user from heartbeat ActiveMs. Replaces the session-end duration metric, which under-sampled (~2% of sessions) on WASM flush loss.

Failure-Free Sessions

-

CanonicalDirect Studio Live sessions with no failure signal — a Critical uno/studio-live/error or a project-download-failed. CI traffic excluded; Wasm-inferred sessions removed from the denominator.

Time to First Value P50

-

CanonicalMedian time from a user's first session-start to their first successful compilation-succeeded (per-user, in-window).

D7 Retention

-

Proxy% of users seen again within 7 days, using first-seen vs last-seen recurrence. Baseline includes Wasm-inferred reach, and identity is the anonymous, browser-scoped user_Id that resets per visit — returns are undercounted, with no cross-device or account tracking until AccountIdHash ships.

Download Conversion

-

Canonical% of project-download-started events that reached project-download-completed. CI traffic excluded.

Agent Prompts

-

CanonicalTotal agent-prompt-sent events in the selected period. CI traffic excluded.

Agent Success Rate

-

Canonical% of agent-suggestion-generated events with OutcomeCategory = Success.

Agent Duration P50

-

CanonicalMedian agent generation duration across all outcome categories from agent-suggestion-generated.

Agent Cancellation Rate

-

Canonical% of agent prompt cycles cancelled before completion from agent-generation-cancelled.

Compilation Success Rate

-

Canonical% of compilation-started events that reached compilation-succeeded. CI traffic excluded.

Critical Errors

-

CanonicalCount of uno/studio-live/error events with Severity = Critical in the selected period. View all →

Investor KPIs (Tier 2)

Growth and monetization KPIs for investor reporting, built on the North Star: Weekly Active Creators — users with at least one successful compilation per complete calendar week. CI traffic excluded.

Weekly Active Creators

-

CanonicalDistinct users with ≥1 compilation-succeeded in the latest complete calendar week. North Star.

New User Activation (7d)

-

Canonical% of new users reaching their first successful compile within 7 days of first session.

Stickiness (DAU/MAU)

-

ProxyAverage DAU over MAU for the latest complete calendar month (requires a 90d timeframe). Leans on the anonymous, browser-scoped user_Id that churns per visit, so this is a low-biased floor until a durable account identity ships.

Gate → Entitled Conversion

-

Proxy% of users denied at the license gate who later show a granted check for the same feature. Inferred from license-feature-* event ordering.

Entitled → Active AI

-

Proxy% of entitled users with at least one agent-prompt-sent in the period.

Weekly Active Creators

CanonicalNorth Star: distinct users with at least one uno/studio-live/compilation-succeeded per complete calendar week (startofweek buckets, partial leading and current weeks excluded).

Signup to Value Funnel

ProxyDistinct users at each stage from license-gate touch (uno/studio-live/license-feature-*) through session, agent prompt, and compile to project export (project-download-completed, Trigger=Export). Proxy: gate events stand in for signup, identity is machine-scoped, and stages are window-scoped, not sequence-enforced.

Weekly Cohort Retention (W1/W4/W12)

ProxyWeekly cohorts by first uno/studio-live/session-start, showing the share returning in cohort week + 1, 4, and 12. Proxy: anonymous web identity churns per visit (1 of 3,592 users seen in 2+ weeks over 90d), so retention reads ~0% until a durable account-backed identity ships.

Engaged Minutes per Active User

CanonicalDaily foreground minutes per active user: sum of ActiveMs from uno/studio-live/session-heartbeat per user per day divided by 60000, averaged over users with heartbeats that day (replaces the WASM-flush-lossy session-end duration metric).

Errors per 1k Heartbeats (Weekly)

CanonicalWeekly uno/studio-live/error events (and Severity=Critical ones) per 1,000 uno/studio-live/session-heartbeat, so error rates track quality rather than traffic growth. Complete calendar weeks only.

Studio Live x Hot Design

Two cross-product views. Design-surface engagement: of real Studio Live sessions, the share that reached the (embedded, WASM) Hot Design designer — engagement, not adoption (web-to-desktop cross-use is not measurable via Machine ID). Cross-product AI accounts: licensed accounts making AI calls on both Studio Live (/v2/chat) and Hot Design/IDE (/v1/chat), post-2026-05-29. CI traffic excluded.

Reached Hot Design Surface

-

-

-

-

ProxyOf machines with a real Studio Live session-start, the share that also emitted uno/hot-design/* — reached the Hot Design designer embedded in the Studio Live browser tab (~99.9% WASM, shared Machine ID). Excludes license-probe-only tabs (which inflated an earlier 83% to the honest ~69%). Genuine web-to-desktop cross-use is structurally 0 via Machine ID (disjoint id namespaces).

Cross-Product AI Accounts

-

-

-

-

ProxyLicensed accounts (AI-endpoint userId) making AI generation calls on both /v2/chat (Studio Live) and /v1/chat (Hot Design/IDE), post-2026-05-29 (after Studio Live migrated off /v1). Covers only AI-active accounts; /v1/chat may include a few legacy consumers, so it is an upper bound.

Hot Design Surface Engagement Trend (Weekly)

ProxyStudio Live machines that reached the (embedded) Hot Design designer per complete calendar week.

Daily Active Users

CanonicalDistinct daily active users, counting unique CI-excluded users who fired uno/studio-live/session-start each calendar day via dcount on the canonical user id.

Weekly Active Users

CanonicalDistinct weekly active users, counting unique users who fired uno/studio-live/session-start per calendar week (startofweek) via dcount on the canonical user id.

Direct vs Inferred Mix

ProxyWeekly share of events that are direct (name starting uno/studio-live/) versus Wasm-inferred, as a percent of their total; inferred activity is non-studio-live event names counted only because OS Architecture is wasm, so it over-attributes.

Download Funnel

CanonicalDownload funnel over the window: counts of uno/studio-live/project-download-started, -completed, and -failed, with completion and failure rates as a % of started (CI-excluded).

Download Conversion Trend (Weekly)

CanonicalWeekly download conversion rate: uno/studio-live/project-download-completed as a % of -started, both bucketed by startofweek(timestamp) and plotted over time.

Agent Prompt Volume

CanonicalDaily count of AI prompts submitted to the Studio Live agent plus distinct users, from uno/studio-live/agent-prompt-sent events counted per day and deduplicated by canonical user id.

Agent Generation Outcomes

CanonicalShare of agent generations landing in each OutcomeCategory (Success, Warning, Critical, IterationLimitReached, ContextOverflowExhausted), computed as each category's count over all uno/studio-live/agent-suggestion-generated events.

Agent Duration by Outcome (ms)

CanonicalHow long agent generations take: median (P50) and P90 of DurationMs from uno/studio-live/agent-suggestion-generated, shown overall ("All") and broken out by OutcomeCategory.

Agent Cancellation by Reason

CanonicalShare of agent runs cancelled before finishing, broken down by CancelReason (UserStop, ConversationCleared, NewGeneration); each uno/studio-live/agent-generation-cancelled count is divided by cancelled plus completed (uno/studio-live/agent-suggestion-generated) runs.

Agent Empty Responses (Daily)

CanonicalDaily count of empty agent completions (uno/studio-live/agent-empty-response, the alertable model-outage signal #2578) and empty responses per 1,000 agent-prompt-sent. CI-excluded.

Compilation Funnel

CanonicalCounts of uno/studio-live/compilation-started, uno/studio-live/compilation-succeeded and uno/studio-live/compilation-failed over the window, with success and failure rates as a percent of started.

Compilation Failure Outcomes

CanonicalFailed compilations from uno/studio-live/compilation-failed, counted and grouped by OutcomeCategory (including worker-infrastructure failures) to show what kind of failure dominates.

Compilation Error Categories

CanonicalShare of build errors by type: uno/studio-live/compilation-failed events counted and grouped by sanitized ErrorCategory (Syntax, Type, Dependency, Other).

Compilation Duration P50/P90 (ms)

CanonicalDaily median (P50) and 90th-percentile (P90) build time for successful compilations, from DurationMs on uno/studio-live/compilation-succeeded events, per calendar day.

Compilation Retry Depth

CanonicalHow often users get stuck in compile-retry loops: uno/studio-live/compilation-retried events counted by ConsecutiveAttempts depth to show the distribution of back-to-back retries.

Errors by Category

CanonicalRuntime errors by area: uno/studio-live/error events counted and grouped by Category, with a parallel count of Critical-severity errors (Severity).