A planner lives in a planning screen, a buyer in a channel, a manager in a report, an application in a service call.
Five steps; the middle three are identical wherever the question was typed.
Every place starts by proving who is asking. Teams and Slack tie a message to a person through single sign-on; a SmartApp carries the signed-in user; the Ask API names the person the call runs as, and the caller holds a credential for that.
From there the sequence is the one every answer follows: intent and objects, the goal, the mapped tables, the query across one source or several, run with that person’s rows and columns.
The shape of the reply follows the place. A channel receives a card with the figure, a short table and a link into the screen. A report receives the rows. A service receives a structured result with the terms, the sources, the timing and a reference to the query.
The record stays single. One line per question names the place, the person, the scope, the query and the row count.
The same conversation, carried into the tool each person already has open.
The full conversation: history, walkthroughs, saved questions, digests and feedback.
A question in a channel or a direct message, answered where the discussion is happening.
The same conversation for teams who live in Slack, with a link into the screen for the query.
A question beside an approval queue or an exception list, filtered to the record in front of you.
An application sends a question and reads back the answer, its shape and its query reference.
A report reads the rows behind an answer and refreshes on its own schedule.
| Place | What it is for | What travels | What holds |
|---|---|---|---|
| The ChatLima screen | The whole conversation, with history and walkthroughs | Everything, including the query and the thread | The asking person’s rows and columns |
| Microsoft Teams and Slack | A question in the flow of a working discussion | The figure, a short table, and a link into the screen | The asking person’s rows and columns |
| Inside a SmartApp | A question beside the queue or exception being worked | The answer, narrowed to the record on screen | The rules of the app and of the person |
| The Ask API | An application or a service that needs an answer | The question in; the answer, shape, terms, sources and query out | The access of the person the call runs as |
| The reporting connector | A report that keeps its own refresh schedule | The rows behind the answer | Row and column rules travel with the connection |
A short answer in the channel, with the whole thread one click away.
47 lines are past due across your categories, worth $318,000. Delta Fasteners holds 18 of them and Corvid Plastics 9.
| Supplier | Past-due lines | Value | Oldest line |
|---|---|---|---|
| Delta Fasteners Ltd | 18 | $142,600 | 11 days |
| Corvid Plastics | 9 | $64,900 | 6 days |
| seven other suppliers | 20 | $110,500 | 4 days |
11 of the 47 lines sit on materials whose planned delivery time is 10 days or more below their 12-month average lead time. Those 11 lines are worth $96,200.
A channel answer is short by design: the figure, a few rows and a link. A reader who wants the terms, the sources and the query opens the thread in the ChatLima screen, where the whole conversation is waiting.
The thread carries between people. P. Sharma picked up D. Okafor’s 47 lines and asked about the planned delivery time behind them, and the answer kept the supplier, the plants and the day.
Each answer runs with the access of the person who typed it. D. Okafor reads two plants and his own categories; P. Sharma reads master data across four plants. Both answers are correct, and both leave their own line in the record.
A channel is also where a saved question lands: a walkthrough posts its Monday digest into the channel that owns the review.
One call carries the question; the answer carries its working back with it.
POST /v1/ask
{
"question": "Which open purchase order lines can
still be cancelled this week?",
"as_user": "l.tan",
"shape": "table"
}{
"answer": "31 open lines worth $348,000 sit
inside their cancellation window
until Friday 18 September.",
"shape": "table",
"rows": 31,
"terms": ["purchase order line",
"cancellation window", "supplier"],
"sources": ["SAP S/4HANA", "Oracle Fusion",
"planning suite"],
"elapsed_ms": 4180,
"scope": "plants 1210, 1220",
"query_ref": "q-2026-09-16-0912"
}The Ask API is the same conversation as a service call. The caller names the question, the person it runs as and the shape it wants back. Juno answers with the figure, the rows, the terms, the sources, the timing and a reference to the query.
That reference is what keeps a service call accountable. Any answer an application received can be opened later in the screen, with its scope and its query as they were.
SmartApps use the same route. An approval queue asks its own question about the record on screen, so a buyer sees the stock position beside the line they are approving.
An answer can also become the start of an action. A finding handed to an agent becomes a recommendation, and a named person approves it before anything is written back.
One meaning, one rule set, one record: the reason a new door costs so little.
This is what makes ChatLima safe to open widely: adding a place adds a door, and the same rules stand behind it.
It also keeps the figures steady. Two people comparing a number from a channel and a number from a report find one definition, because both were built from the mapping your team confirmed.
Showcase estate.
By Thursday the cancellation work had moved into the buyers’ channel. D. Okafor asked about past-due lines where the team was already talking, and the answer arrived in 3.2 seconds with a link into the screen.
P. Sharma followed up in the same thread and found 11 lines resting on lead times set below the actual. Both questions left their own line in the record, and both read the mapping A. Novak published on Tuesday. The next step was to teach Juno the words the team had used all week: NCNR, and the lead-time fence.
Vocabulary, acronyms and abbreviations, custom aliases per table and column, and the training run that carries them to every place ChatLima works.
Yes. Teams and Slack tie a message to a Juno person through single sign-on, and the query runs with that person’s rows and columns, exactly as it does on the screen.
Yes, and that is the point. Each answer runs with the access of the person who typed it, so a buyer and a steward in one thread each read their own scope.
The channel answer carries a link into the ChatLima screen, where the query, the terms, the sources and the whole thread are open.
Send a question, name the person the call runs as, and read back the answer, the shape, the rows, the terms, the sources, the timing and a reference to the query.
Yes. Vocabulary, mapping and policies live once. A term taught on Tuesday reaches the screen, the channels, the apps and the connector at the next training run.
Each one builds on the one before it.
Reads your rows in place.
Connect and move ↗One record per real thing.
Master and govern ↗Apps for every decision.
Operate and decide ↗Asks and acts, at your level.
Ask and act ↗Bring two systems and one decision your planners make every week. In the first week we register the data, confirm the mapping with your team, answer your questions in ChatLima and run the first agent at the level you choose.