The RIA console
Ask RIA about the screen you're on, hand it a data file, and follow what it does from a pane you can open, move and resize anywhere.
Where: RIA diamond (lower right)
The console is where you talk to RIA: a small diamond over the page that opens into a pane with your conversation and a message box. RIA knows your screen, model and scenario, so "this" needs no name.
Open and close the console
Click the diamond, or focus it and press Enter or Space (Open RIA chat); the cursor lands in the message box. To close, click X in the pane's header (Collapse RIA chat) or press Esc anywhere in the pane; focus goes back to the diamond. A new browser starts with the console closed.
- The diamond is on every screen except a workspace in fullscreen, until you leave it.
- On a window 640 pixels wide or less, the open console fills the whole window, with no drag or resize.
- Opening and closing animate, unless Motion in Page settings or your system's reduced-motion setting says to reduce.
Move and resize it
- Move the diamond: drag it anywhere; it stays there in this browser. A press that moves less than 4 pixels counts as a click.
- Move the open pane: drag its header (not up under the app's top bar). Close a dragged pane and the diamond moves under where it was.
- Resize: drag the pane's left edge (Resize RIA console); double-click it for the default width. The width is kept in this browser.
The pane opens centred on the diamond, 7/8 of the window tall, and always stays whole on screen.
Ask RIA something
Type in the message box and press Enter, or click the send arrow (Send); Shift+Enter starts a new line. The box grows to ten lines, then scrolls. While RIA answers, the box reads Working… and you can't type ahead.
On an empty conversation, a Suggested row sits above the box. A chip puts its text in the box for editing; nothing is sent until you send it. The opening line, placeholder and chips depend on the screen:
| Screen | Placeholder | Suggested |
|---|---|---|
| Home | Ask about your plan, scenarios, or what needs you… | What changed this week, Pick a model to dive into, Plan a model |
| Model Library | Ask which model to open, or find one by tag… | Which models are ready to promote?, Find by tag, Audit recent activity |
| Datasets | Ask about your imported data… | What have I imported?, Find a dataset |
| One dataset | Ask about this dataset… | Describe this data, Spot data problems |
| A model's Overview | Ask about this model… | Compare the scenarios, What's the current plan, Open the data |
| Scenarios | Type a what-if: e.g. 'What if oven time doubles?' | Compare top 3 scenarios, Probe what's binding, Try a +10% demand swing |
| New Model, before a model exists | Describe your business: products, prices, capacities, demand. | Plan a 10-product bakery, Describe my own decision, Import a workbook |
| New Model, with a model loaded | What would you like to change about <model>? | Add a constraint, Refresh data, Tune a KPI (Modelers get a set of their own) |
| A workspace | Compare, diagnose, or summarize these scenarios… | Compare to Base, What's holding the plan back, Summarize for the Go team |
| Workflows | Ask about workflows and runs… | What's running now, What needs me |
| A workflow run | Ask about this run: diagnose, compare, or summarize for sign-off… | What's left on this stage, Diagnose the at-risk scenario, Summarize for sign-off |
| Editing a workflow | Ask about stages, roles, or gates… | — |
| Credentials | Ask what a verdict means, or what to check next… | Explain the verdicts |
| Review imported data | Ask about a mismatch, or what to accept… | Explain the differences |
| Change history | Ask what changed, or why a number moved… | — |
| Review Queue | Ask about the review queue… | — |
| Organization | Ask about members or roles… | — |
| Beta feedback | Ask about the product while you write feedback… | — |
| Anywhere else | Ask about this model…, or Ask RIA… with no model open | — |
On Home, the opening line also changes with your persona. There's no Stop, Retry, Regenerate, Copy or Edit button. To retry, send the message again; sending the same turn twice never runs it twice.
Make a what-if from the Scenarios screen
On Scenarios with a model open, a band above the box reads Each message becomes a new what-if scenario and the send button becomes Create scenario (Create a scenario from this prompt (Enter)). What you send is filed as a scenario named after your text, minus a leading "What if" and the trailing "?". Writing and running what-ifs is on What-ifs. Two kinds of message go to RIA instead:
- A question. Text that reads as a question and changes nothing goes to RIA as a chat message, and the console opens to answer it.
- A new business. A long description of a new model (a build verb and "model", or prices with units and limits, or a bulleted brief) starts a fresh build instead of a what-if.
Attach a data file
Drag a file onto the message box; there's no attach button. Only the first file you drop is taken. A
data file (csv, tsv, txt, md, json or tab, up to 1,000 KB) is read in your browser and sent
inside your message. Its chip shows the name, <n> KB · <n> lines, and an X (Remove), and the
placeholder becomes Add context about this file (optional)…. A pdf or docx is attached as a design
document instead: it stays with the conversation and can build a model straight from it. See
Describe your business.
| Message | Means |
|---|---|
File too large (<n> KB). Cap is 1000 KB. | Split the file, or paste the rows you need. |
| Excel parsing isn't wired yet: paste the rows as text, or save the sheet as CSV. | Workbooks don't attach. Save as CSV, or import it as a dataset. |
Unsupported file type .<ext>. Try CSV, TSV, JSON, MD, PDF, DOCX, or paste as text. | Use one of those types. |
Doc too large (<n> KB). Cap is 10 MB. | A PDF or DOCX over 10 MB. |
| Session not ready yet: try again in a moment. | The conversation was still starting. Drop the file again. |
Read RIA's reply
Your messages sit on the right, RIA's on the left, with bold, italics and lists rendered. The conversation scrolls to the newest line.
| You see | What it is |
|---|---|
| A spinner and Thinking… | RIA is working on the message you sent and hasn't started a step yet. Background work never shows one. |
| A list of steps, such as Reading your model or Working out the best plan | What RIA is doing, one line per step: a check on each finished step and a spinner on the one in hand. It replaces Thinking… and goes when the answer lands. |
| A spinner line saying your engine is spinning up | Your own optimization engine is starting. Clears when it's ready. |
| A spinner line saying how many are ahead of you | Your engine is busy with other work. There's no time limit on the wait. |
| Reconnecting… | The answer went quiet for a minute. The console reconnects once and picks up any answer that arrived meanwhile. |
| RIA is asking · reply below | RIA's last message is a short question waiting on you. |
| A dashed, quieter line | A side note from work RIA was already doing, filed above the question you're waiting on. |
| A name with a dotted underline | A link. Click it to open that scenario, model or workspace (Open this scenario); the conversation stays open. |
| A card with a check | A receipt: something RIA changed. open goes to it; Undo appears where one step reverses it, and the card then reads Undone — …. |
Resumed <title> | A conversation you reopened, or reloaded. RIA carries on from its record. |
Some answers land on the page rather than in the console: news factors and starter what-ifs on Scenarios, traces and comparisons on results, build progress in New Model. Reload the page and the same conversation comes back, with the scenario, workspace and pin it was on.
Let RIA take you there
Ask RIA to go somewhere and it does: Home, New Model, Model Library, Datasets, a model's Overview, Scenarios, Change history, Credentials or Review imported data, Workflows, a workspace, the Review Queue, or Beta feedback. A model screen with no model to go to is skipped.
See what RIA is looking at
Above the message box, Current names the model on screen and its selected scenario, as Model › Scenario. It's a label that follows the screen, not a control. During a workflow run it names the run's
model, and a Workflow / Stage card sits at the top of the pane. Every message also tells RIA the
screen you're on, the open workspace and your persona.
When you send a backlog item to RIA from Develop (on Change a model), a working on chip pins it: "this" now means that item, whichever screen you're on. Its X (Unpin change order) lets go.
Open the console from other screens
| Button | Where | What it does |
|---|---|---|
| Ask RIA to draft it | A backlog item on Change a model | Puts Help me address: "<item>" in the box and pins the item |
| Build it into my model | A runnable what-if on the backlog | Puts a request to build it into the model in the box and pins the item |
| Ask RIA to help fix these | The validation drawer, with a check failing | Closes the drawer and puts the failing checks in the box |
| A Suggested next action | Model home | Puts its prompt in the box, or goes to its screen with the prompt ready |
| Approve on a request to build | Home inbox | Puts the request in the box and opens the model in New Model |
| Talk to RIA about this data | The load data step | Sends your note about that table straight away |
| A what-if that reads as a question | Anywhere you write a what-if | Sends it to RIA as a chat message |
Everything except Talk to RIA about this data waits for you to press Enter. Other Ask RIA buttons — in Develop's review, data entry, the widget builder, and the workspace and workflow drafting windows — answer in their own panel and don't use the console.
Know when RIA has finished
The diamond carries the only signal. Its status dot pulses while RIA works (RIA is working…) and while a model is registering or uploading. With the console closed, an accent dot appears when RIA leaves a proposal or starter what-ifs you haven't seen (New from RIA); a plain answer, a finished build or a receipt doesn't light it. There are no pop-ups, sounds, tab-title changes or browser notifications. Requests from RIA's background agents and long solves show on Home.
Limits
- Pane width 300 to 640 pixels, 380 by default. Height 7/8 of the window.
- The diamond keeps 8 pixels from every edge.
- Messages have no length limit.
- Data files:
csv,tsv,txt,md,json,tab, up to 1,000 KB, one per message. The chip counts lines up to 99,999. - Suggested chips show only on an empty conversation.
- The step list keeps the last 20 steps.
- The console reconnects after 60 seconds of silence, or after a 5-minute answer.
- 30 messages per minute per person. See Troubleshooting.
Related
- RIA's proposals: the cards between the conversation and the box
- Conversations: the menu under the header, one conversation per model
- Troubleshooting: every error the console shows
- Window tour: where the diamond sits in the window