What uStudio Labs is

uStudio Labs builds products, prototypes, and solutions for uStudio customers at speed. Operating outside the core uStudio product organization, the team draws on the uStudio Platform and its podcasting capabilities to turn customer requests into working solutions quickly. Because these are prototypes and beta releases, customer feedback shapes them directly - it is what lets the team iterate and deliver business value sooner. That same pace means Labs products change more often than the core uStudio products and may introduce new features, or the occasional rough edge, without prior notice. To request access or share feedback, email labs@ustudio.com.

What Stream Reporter is

Stream Reporter is an AI-powered analytics assistant for podcast producers on uStudio. You ask questions about your audience in plain English - “How are we doing this quarter?”, “Top 5 episodes by plays last month”, “Which topics keep listeners to the end?” - and get instant, streaming answers backed by interactive charts, tables, and KPI cards on a visual canvas.

It replaces the slow loop of exporting analytics to spreadsheets or clicking through fixed dashboards. Instead of building a report, you have a conversation - and the answers assemble themselves into a shareable workspace you can revisit anytime.

Why teams use it

Answers in seconds. Ask a question and watch the answer build - no exports, no dashboard-hunting, no waiting on an analyst.
Numbers you can trust. Every figure on the board is computed by code, not written by the AI. (More on this below.)
Content meets the audience. Stream Reporter connects what was said in an episode to how people actually listened - something generic analytics tools can’t do.
A living dashboard. Answers accumulate as draggable cards on a canvas, so a conversation becomes a workspace you can arrange, save, and share.
Share without logins. Turn any board into a clean, read-only report link for stakeholders in one click.

What you can do with Stream Reporter

Every capability below lives in the same conversational workspace.

Capability	What it does for you
Ask in plain English	Ask about plays, listeners, engagement, and episodes the way you'd ask a colleague - from “total plays this month” to “which topics drive the best completion across our catalog?” Stream Reporter plans the data it needs, fetches it, and answers.
Watch answers build live	Responses stream in as they’re assembled. You see each data step as it runs and can expand a reasoning panel to check how an answer was reached - no black box.
Work on a visual canvas	Every answer drops KPI, chart, table, or text cards onto a full-screen board, so a session becomes a living dashboard instead of a scroll of chat. Drag, resize, and arrange cards; the layout persists.
See it the right way	Six card types and 17 chart types cover every shape of answer - trends, rankings, proportions, correlations, funnels, cohorts, heatmaps, and more - alongside rich tables with inline bars, sparklines, thumbnails, and badges.
Cover the whole picture	Plays and impressions, listening time, completion and drop-off, audience size and growth, publishing cadence, per-show engagement, and your top listeners - all in one place.
Break it down any way	Slice by episode, show, platform, listener, or time - including hour of day and day of week - and use custom metadata fields. A flexible aggregation engine groups raw events by any dimension you name.
Connect content to attention	Stream Reporter ties what was said to how people listened: it segments each episode into labeled topics and maps them against retention, so you can see which topics hold attention and which lose listeners - within an episode and across your whole catalog.
Drill into anything	Every card has a one-click prompt to dig deeper (“What’s driving this trend?”, “What’s behind this number?”). The follow-up updates that card in place rather than cluttering the board.
Save and revisit	Sessions auto-save and auto-title from your first question. Reopen, search, and manage past workspaces anytime, each showing its card and message counts and when you last touched it.
Share in one click	Turn any board into a public, read-only report link - cleanly rendered, light/dark aware, and responsive - for stakeholders who don’t log in. A shared report is a snapshot of the board at share time.
Switch between studios	If you manage multiple brands, pick which studio you’re exploring from a searchable dropdown. Your choice is remembered and every query is scoped to that studio’s data.
Built for trust and comfort	uStudio sign-in, with access tokens kept server-side and never exposed to the browser. Your data is never used to train AI models. Full light and dark theming and a genuine mobile layout.

Why the numbers are trustworthy

The Zero-LLM-Math approach: the AI decides what to show and how to visualize it, but every number it displays is produced by deterministic code - never typed by the model.

This is by design, not by hope. When the assistant builds a card, it can only point to real results returned by the data layer and name the columns to display; there is no place in the system for the model to enter a figure of its own. Totals, trends, and formats such as 1.2k, 1h 23m, and 85% all run through dedicated functions.

If a card’s underlying data can’t be resolved, the card is skipped rather than shown - so you never see an invented or miscalculated figure. The one deliberate exception is the narrative text card, which is clearly presented as written commentary and summary, not as a source of numbers.

A closer look: content-to-retention

Most analytics tools can tell you that listeners dropped off at a certain point. Stream Reporter can tell you what they were hearing when they did.

Topic segmentation - each episode is parsed and grouped into labeled topics, with a summary and key terms for each.
Retention curves - deterministic drop-off curves show how many listeners remain at each point in an episode.
Topic-to-engagement mapping - see which topics hold attention and which lose listeners, within one episode or across your whole catalog.
Point drill-down and search - ask what was discussed at the 35% mark, or search episode content by keyword to find where a subject came up.

How it works

Stream Reporter pairs a reasoning model that plans and visualizes with a deterministic engine that does the math. It reads your account’s first-party analytics directly, so the figures are real platform data - not third-party estimates.

Component	Role in Stream Reporter
Deterministic compute engine	Runs every calculation, aggregation, and format you see - the source of every number on the board.
Gemini 3.1 Pro	Plans multi-step questions and decides what to show and how to visualize it - it never writes the numbers.
Gemini 2.5 Flash	Extracts and labels episode topics from transcripts, cost-efficiently.
uStudio analytics	Native, first-party access to your plays, impressions, and engagement data - not a third-party estimate.

On privacy: you sign in with uStudio, access tokens stay server-side and are never exposed to your browser, and your data is never used to train AI models.

This initiative is part of uStudio Labs and is currently classified as a Concept Car.

Concept Cars are early-stage prototypes designed to help us explore new ideas, workflows, and opportunities in collaboration with customers. The goal is not to perfect a solution, but to learn what problems are worth solving and which approaches create the most value.

As with all uStudio Labs initiatives, customer feedback plays an important role in shaping what comes next.