Project Commander Docs ← Back to site

Ask — what it can do, what it cannot, and how

The Ask sidebar for the CLAIM project: the question box at the top, and an answer to
The Ask sidebar for the CLAIM project: the question box at the top, and an answer to "How much is done, and are we on pace?" opening with its one-line answer above the folded Detail

What you can do with it

Ask a question about your project in plain words and get the answer in seconds, worked out from the same figures the tabs show: the same delivery forecast as the Delivery Forecast card, the same Deliverable and Overcommitted verdicts as the Sprints tab, the same capacity arithmetic as the Team & Capacity tab. One question can answer what would otherwise take several tabs: when the project will finish and how likely the target is, which sprints and which people are over capacity, what is blocked, what changed, and what to look at first.

You can also try a what-if, such as one more person, without changing anything. Nothing you ask, and nothing on the sidebar, changes anything in Jira.

It is in the app installed in Jira, on every tab, and on a Jira dashboard, and in the web app at projectcommander.app once it is connected to your Jira. Show it with the control at the right-hand end of the tab bar, past the settings gear. It is not offered in demo mode, which has no Jira to read, and not before a project is registered.

Questions it answers

The sidebar lists the questions it can answer under Things it can answer:, on an empty sidebar and again below the answers once anything has been asked (they step out of the way only while an answer is being worked out). They are in folded groups, each a heading with no count beside it that opens onto its questions: General first, then one group for each tab, named as the tab bar names it — Dashboard, Sprints, Epics, Team & Capacity, Scope, Alerts, Risks, Actions, What-If, Timeline, Projects — holding the figures that tab shows. Pressing a question puts it in the question box and asks it, exactly as if it had been typed; any of them can also be typed, and so can any other question in the areas listed further down. The headings are there even if the sidebar's first call to the app does not come back.

These are the questions inside General:

The date, the issue key and the epic name are examples; put your own in their place. On screen the sidebar has already done that: the questions under Things it can answer name your project's own target date, first open issue, first epic, people and sprints, and a question your project has nothing to fill in for (no epic, nobody on the Team & Capacity tab) is left off every list rather than shown with a made-up name.

Suggestions as you type. While you type, up to six questions whose words match what you have typed appear under the question box, each with its own Ask button. Where a question takes a name — a person, a sprint, an epic — the name is a dropdown filled with your project's own names. The suggestions go the moment a question is asked, and the question already in the box is never offered back to you.

What it is

A sidebar that slides in over the right of whichever tab you are on, starting just below the tab bar. You type a question about the project you are looking at and get an answer, with the figures you were already reading still on the screen behind it.

It is not a tab and not a chat window. One control shows it and hides it again: the last button at the right-hand end of the tab bar, past the gear, drawn as a page with a panel down its right-hand side and an arrowhead pointing the way the sidebar will move. Press it to show the sidebar and press it again to slide it back out. Its tooltip reads Show the Ask sidebar or Hide the Ask sidebar, whichever it will do next.

Because the sidebar starts below the tab bar, nothing in that bar is covered while it is showing: the control that hides it again, the project picker, Guide, Refresh and the gear all stay where they are and can all still be pressed.

The sidebar opens 320 pixels wide. Drag its left edge to make it wider or narrower, and double-click that edge to put it back to 320 pixels; the width you leave it at is remembered in that browser, on every tab and every project.

Change project while the sidebar is showing and it switches with you. The questions, answers and any what-if from the project you were on are put down, and the questions it offers under Things it can answer are filled with the new project's own first open issue, epic, target date, people and sprints. The same is true if you change project with the sidebar hidden: showing it again puts you on the project the picker is set to, never the one you were on before.

What the sidebar shows, top to bottom:

The opening line follows a few plain rules. A question asking for a yes or a no opens with "Yes." or "No." when the figure settles it ("Will we finish by Mar 31, 2027?" — "No. The delivery forecast is Apr 19, 2027, 19 days after Mar 31, 2027."; the odds of finishing by that day are under Detail). A what-if reads "The delivery date would be Apr 16, 2027 instead of Apr 19, 2027." A "which" question gives the count, then the names ("1 overcommitted: Sprint 6."). A question asking for a comparison gets the comparison word and both figures ("More: 44 pts this month against 42 pts last month"). Nothing about how a figure was worked out, what was assumed or how many runs were tried is in the opening line; that is under Detail.

A few answers draw a bar chart under Detail, and the same figures are always listed there as well: "Which sprints are overcommitted?" draws each sprint's remaining demand against its remaining capacity; a question about the people draws each person's work against the capacity left; and a question about the days draws how far over each overloaded stretch of days runs.

The same question typed into it gives the same numbers every time.

Nothing is read from Jira before an answer. A question is answered on the figures the app already holds for the page you are on, however long ago they were read. To read the project from Jira again, press Refresh in the tab bar (the circular arrows, whose tooltip reads Read this project from Jira in full now), then ask.

What it can answer

The dates. When the project will finish: the middle run, the best case, and one in ten either side. The chance of hitting a target date you give it. What the same team would finish at the pace they have actually delivered. What capacity the answer was built on and where that came from. How much work was counted, and how much backlog was left out of it.

The plan. Every sprint: its own work, what carries into it, the capacity still ahead of it, what spills forward, and whether it is Deliverable or Overcommitted. Every person: their work, their capacity, their spare, and what cannot be placed at all. Who finishes last. What is blocked by work scheduled later, and anything blocking itself in a circle. The chain of work that decides the finish and who holds it. Everything the Alerts tab finds, including work blocked by work that is already finished ("Which issues are blocked by work that is already finished?", offered under Alerts). Where the data is too thin for the dates to be trusted. The running sprint is the one Jira has open, and only when none is open the sprint whose dates hold today, so "this sprint", "next sprint", "Who is overcommitted?", "What sprint are we in?" and "How many sprints are left?" all name the sprint the Sprints tab shows as Active (2 November 2026). So does every question about the current sprint that names no sprint — "How far through the current sprint are we?", its demand, its capacity, whether it is overcommitted, its confidence, when it ends and "How loaded is the team this sprint?" — even when that sprint has run past its end date and has not been completed in Jira (6 November 2026): on a board whose Active sprint ended last week, the answer is that sprint's Progress, as its card shows it, and not the next sprint's. A question asking for the demand, the capacity or the load of "the current sprint" or "this sprint" ("What is the demand this sprint?") is answered with that figure for the sprint running, not with which sprint is running and its dates (12 November 2026). Which sprint is most overloaded follows each sprint card's own verdict: a sprint asking for work with no capacity left ahead of it comes first, and a sprint whose work fits its capacity is named as the fullest, never as overloaded ("No sprint is overloaded; the fullest is Sprint 33 at 54%: 60 hrs asked of it, against 112 hrs of capacity still ahead.").

A person or an epic, named. Name a person or an epic in a question about the dates or the plan, such as "When will Diana finish?" or "When will the Payments epic finish?", and the answer is about that work alone: its first line opens "About Diana Lee's work" or "About the Payments epic", and the work counted, each sprint's figures and the dates are that work's. An epic's finish is given as a range, from its work taken first in every sprint to its work taken last behind its owners' other work, because the plan does not say in which order people take their work within a sprint; the two ends are one date when the order makes no difference, and on a project without sprints there is no order to vary, so the finish is one date. A person's work is planned on their own capacity from the Team & Capacity tab; where that tab holds none for them, the project's capacity is used and the answer says so. The Delivery Forecast card's date, how much is done, whether it is on pace and the epics' rows are about the whole project, so such an answer leaves them out and says why. What blocks the work is still read across the whole plan. Ask without the name for the whole project.

The people's load, as the Team & Capacity tab counts it. "Who is overcommitted?", "Who has spare capacity?", "Is Diana overcommitted?", "How many people are overcommitted this period?", "How much is available this period?", "Which team members don't have assignments?" and the team's capacity are the Team & Capacity tab's own figures (5 November 2026): each person's Demand is the open work whose dates fall in the stretch asked about, an issue's own dates first, as the tab's Demand column counts it; their Net is rounded to a whole number and their Remaining to one decimal, as the tab's rows show them; the team's capacity is those added up, as the tab's Total row adds them; and the word for a person is the tab's Status column's. With no time named, the stretch is the sprint that is running — what the tab shows with its period set to Sprint — or this week where no sprint runs; name a week, a month or a sprint and the answer is about that. Where people give only part of their time to the project, the figures are the ones the tab shows on Per selected project. Somebody holding only work with no date at all is not said to have nothing: the answer lists that work as having no date, as the tab lists it beside them as undated. Each person's total and completed work in a sprint ("What is the total in Sprint 9 by person?") is still the sprint card's own table on the Sprints tab. Over a sprint, each person's capacity is counted over the days the sprint card counts (12 November 2026): when the next sprint starts on the day this one ends, that day is capacity in the next sprint only. And somebody with work in the sprint and no capacity left in it is named in the sprint card's own words, never counted as overcommitted: "1 of 4 on the team is overcommitted in Sprint 7: Diana Lee by 9 pts; Alice Chen has work and no capacity left." Asked about that one person, the answer gives the two figures and closes "No capacity left", with neither a Yes nor a No. Over a week or a month the word stays the Team & Capacity tab's Status column's. A question naming a person ("Test1 workload") is answered about that person, never with the whole team's load, and a person's name is never taken from words like "got" or "taking": "whos got time off this week" is answered with who is off this week (12 November 2026).

Scope since the start needs a Start Date. "What was the scope at the start?", "How much has scope grown?" and "Is scope growing faster than we deliver?" are measured from the project's Start Date, which is set on the Scope tab's Start Date card (its gear, Date source). Left as it opens, on Earliest start, that card is on the day the first issue was created, when nothing was in scope yet, so there is nothing to measure from: the answer says "The scope at the start can't be worked out because no Start Date is set; the scope is 638 pts now. Set a Start Date on the Scope tab's Start Date card (its gear, Date source)." (5 November 2026; until then it read "0 pts" at the start and all of the scope as growth, on every such project). A question that names its own day — "Has the scope grown since 1 July?", "What was the scope on 1 September?" — is answered from that day whatever the card is set to.

When the Delivery Forecast card has no date. Where the card has nothing to forecast from — for example an unfinished issue that carries no story points — a question about the date, the odds, the ends of the range, the earliest and latest finish across the models, the P85 date or a what-if opens with the card's own reason. A question about something else keeps its own answer (5 November 2026): how far through the sprint the work is, this sprint against the average and the remaining work by epic give their own figure, or their own reason for having none ("Nothing is committed to Sprint 9, so there is no progress to measure.", "No epic on this project."). A figure that needs the forecast — the capacity before the target date, the capacity short, the capacity after the target date, who clears their work last — says that it could not be worked out, then gives the card's reason.

The capacity gap across every project. "What is the combined capacity gap?" is answered for every project on the Projects tab and every open sprint (5 November 2026): how many projects are short of capacity before their target date and by how much — each project's remaining work less the capacity counted before its target, the two figures its Utilization column on the Projects tab divides — then the sprints whose cards on the Sprints tab read Overcommitted, each with how far it is over. "1 of 2 projects is short of capacity before its target: BETA by 12 hrs. 2 sprints are overcommitted: Sprint A1 in ALPHA by 5 pts, Sprint B1 in BETA by 10 hrs." Every figure is in its project's own unit; points, hours and days are never added together. Under Detail a table gives each project's remaining work, capacity to target, shortfall and overcommitted sprints. The page works the figure out from what the Projects tab already holds, so nothing is read from Jira for it. Asked of the one project on screen — "What is the combined capacity gap across the current project?" — it is that project's own day-by-day reading, as before.

Demand against capacity, day by day. The finer reading of the same question the plan answers sprint by sprint: every working day in the window, what the open work booked on it adds up to against what the people have that day, the score the app puts on it, and every stretch where the demand has run past the capacity, with its dates and how far over it goes. A sprint can hold exactly its capacity with every issue in it due in the same three days, and only this shows that. It comes with how much there is to go on: how much of the work is estimated, how much has dates, how much has somebody on it.

When each issue lands. The dates the Timeline tab draws, from the same calculation: when each issue starts, when it finishes, and which of five things is holding it up — its own start date, a sprint that has not begun, work that blocks it, the person being busy on something else, or nothing at all. Ask about one by its key and it answers about that one. Ask generally and it leads with what finishes after a date somebody set. Whether one issue is late is a Yes or No from its Timeline finish against its own due date: "Is CLAIM-3 going to be late?" opens "Yes. CLAIM-3 finishes Nov 23, 2026 on the Timeline, past its due date of Oct 19, 2026", and "Is CLAIM-1 on time?" turns the yes round.

The tree. Epic, story and sub-task, with each epic’s figures the Epics tab’s own. Three things only the shape shows are called out: work under no epic, so nothing rolls it up; a parent estimated smaller than the sub-tasks beneath it, so every roll-up reading the parent is short by the difference; and a sub-task still open under a parent somebody has already closed.

The project. How much is done and whether it is on pace. The remaining work by epic ("Remaining work by epic: Payments 60 pts, Search 8 pts."), one epic's remaining work and its issues with no estimate, and whether an issue is under an epic ("Yes. CLAIM-7 is under Payments (CLAIM-100)."); an epic the project does not hold is said back by name ("There is no Reporting epic in this project.") whatever else the question asks. Who is over capacity in any way ("Are shared people double booked?"): every person with more work than their capacity in any open sprint, as the Alerts tab's Person over capacity in a sprint check lists them, and every person whose shares across projects add to more than 100% in the Projects tab's Team Allocation Matrix, saying which way. Which sprint is in the most trouble: the sprint most over its capacity and the sprint holding the most issues due after their epic's due date, each with its reason ("None is over capacity (the nearest is Sprint 9, 21 pts asked of 21 pts still ahead), and no sprint holds an issue due after its epic's due date."). Which epics have no owner, and which are done except for one issue (an epic holding a single issue, still open, counts). Velocity: the average, the trend, the best and worst sprints. How much is finished a week. Every epic and how it is doing. What is past its due date. Whether the scope has grown since a date and by how much. Who changed an item and when. Finding work by its words, by a field, or by the state it is in.

Releases. Every release, how much is in it, how much is done, and the earliest it can land. Name an epic and a release together, such as "How is Identity affecting Mercury release?", and the answer says how much of what is left in that release is the epic's work and what the release's earliest landing would be without it; a release is named the way the project names it, with or without the word "release", and in any form the page's own release list holds it — "release v2.1" on a project that calls it 2.1 is that release, "R4" is R4 — while a release the list does not hold is said back by name ("No release called v3.2 is on this project, so nothing was answered about it."), never answered with another release's figures or the project's remaining work. "Will release 2.1 make the deadline?" is answered too (below).

Capacity against the work, as a shortfall. "How much capacity are we short?", "How much extra capacity is needed by the end of the quarter?", "What is the capacity gap?": the work still to do less the capacity before the day named, both the Delivery Forecast card's own figures, with the target date standing in when no day is named. "7 pts more capacity is needed before Oct 19, 2026: 61 pts of capacity against 68 pts of work still to do", or "No capacity is short before the target date, Nov 13, 2026: 101 pts of capacity against 68 pts of work still to do, 33 pts to spare." Capacity to a day named ("How much capacity is there before Oct 19?") is the same card's figure for that day.

Time off against the date. "Would PTO affect delivery?", "Would time off affect the date?": the Delivery Forecast card's date with the time off booked on the Team & Capacity tab beside its date without it, as a Yes or No. Name a person and days as a what-if ("What if Diana is off from Oct 13 to Oct 16?", "How would the forecast be affected by Diana taking PTO from Oct 13 to Oct 16?") and the plan is run again with that time off added: "The delivery date would be X instead of Y." On sprints whose capacity is a typed number the date cannot move, and the answer says so under Detail: set the sprint's capacity to Team capacity for time off to count. Time off somebody would take, with nobody and no days named ("Would adding PTO affect delivery?"), is asked back for the person and the days.

A thing you name is read, never asked back for. A label no issue carries is answered "No issue is labelled regulatory."; a name at the start of the question ("Zoë's capacity?", "Müller's workload?") is read as a name, accent and all, and a person the project does not hold is refused by name; a sprint the project does not hold, by number or by word ("cycle 12 status", "What blocks Sprint Blue?"), is said back ("This project has no Cycle 12, so nothing was answered about it."); and "Which sprints have ended and are still open?" is one question, not two.

Time actually spent. Hours logged against the estimates on the same issues, how far off they were, and the hours per person.

Whether a release makes its deadline. "Will release 2.1 make the deadline?", "Will release 2.1 make Oct 30?": three dates against the target date, or the day named — the version's own release date in Jira, the latest due date of any issue tagged with the release, and the end of the last sprint holding its unfinished work — as a Yes or No, each date said with which side of the deadline it falls. The versions are read from Jira when asked; a version with no date in Jira is said so.

Whether an epic slipped since a day. "Did Payments slip since Monday?", "Has the Search epic slipped since last week?": read from the history Jira keeps for the epic and every issue under it, when asked — a due date moved later, or an issue moved to a later sprint, on or after the day named. "Yes. Search slipped since Sep 21, 2026: CLAIM-4 moved from Sprint 9 to Sprint 11 on Sep 27, 2026." or "No. Nothing under Payments moved later since Sep 28, 2026: no due date and no sprint changed in Jira's history." The app keeps no history of its own.

The last closed sprint. What it held, what it delivered, what carried over, what was added after it started, and the app's own scoring of it.

What needs your attention. A short brief, worst first, one line each, nothing that is fine: work nothing can place, a target the runs did not hit, work past its due date, a sprint over its capacity and by how much, a person with more work than they can do, a blocker scheduled after the work it blocks, scope growing, and the gaps that weaken every date above.

Why an answer is what it is. The engine's own factors, largest first, each with the figure that makes it one.

What it is going on. Everything in use that Jira did not supply — the target, the capacity, the estimation mode, scope growth, the backlog setting, a what-if in force — each saying where it came from and the words that change it.

The records the app keeps. The risks on the Risks tab, the actions on the Actions tab, and the baselines taken on the Dashboard with what the forecast has done since each. Which risks have nobody on them is read from the Owner of each open risk: "Who has unowned risks?" opens "1 open risk has nobody on it: Data migration gap (6)." - each with its score, and "any critical risks without an owner" says how many of them are critical.

What-ifs, which change nothing. More people, epics taken out of the plan, and any of the four What-If variables (velocity, issue estimation, scope or capacity) moved by an amount you name: "What happens if velocity drops 20%?", "What if scope grows 10%?", "What if the estimates turn out 15% larger?", "If capacity falls 30%, when do we finish?" The answer is the date with that change, instead of the date as the plan stands, and the change is named in the amber line while it is in force. These build up across a conversation, so asking for one more person and then another is a question about two. A second amount named for the same slider takes the place of the first ("What if scope drops 10%?" after 15% is 10%), while "another 10%" adds to it (15% and another 10% is 25%), and the amber line shows the total. What is in force is shown on screen until you put it down with Back to the plan as it stands or Clear, or change project. The sidebar does not save what-ifs: asked to save one by name, it says so, and points to the What-If tab, which keeps scenarios by name.

What re-planning would do. The same work packed into the earliest sprint each issue can go in, run on a copy.

What it cannot do, and why

It changes nothing in Jira. It cannot move an issue, change an estimate, an owner, a status or a date, or create or close anything. Asked to move work, it says "This panel does not move work. On the Sprints tab an issue is dragged to another sprint, and Auto-level moves work so no sprint is over its capacity." Every change to the plan is made on the tabs.

It will not tell you what people meant. Comments, descriptions, whether the team seems blocked, the mood — it refuses these and says so: "That asks what people meant, and this does not read what people wrote." Reading a comment and concluding there is a risk cannot be checked by anybody, and something nobody can check is the one thing this product must never produce. The refusal is the product, not a limit on it.

It will not judge a person. Who is the most reliable, who to promote, whose fault a slip was, whether somebody is unhappy, off sick or struggling — it refuses these too: "That asks for a judgement about a person, which this product never makes."

It will not explain how to use the app. "How do I save a baseline?", "Where is the Include backlog switch?" are refused: "That asks how to use the app, which this panel does not explain." Each tab's guide and the online help say where every setting is.

It will not do things outside the app. Sending an email, booking a meeting, ordering lunch, looking up an exchange rate or a public holiday, advising on Scrum, comparing other tools, writing code — "That asks for something this product does not do."

It will not read anything outside the project's Jira data. "Check Slack and tell me why we're late", "Look at the Teams chat and tell me who is blocked", "summarise the standup notes", "What risks do you think are forming that aren't recorded anywhere?", "What is not in Jira yet?" are refused: "That asks the panel to read something outside the project's Jira data, which it does not do." It answers from what Jira records for the project and does not notice a risk nobody has written down; record it in Jira or on the Risks tab and it answers about it from there. The measure called slack is still answered when it is asked for as a measure: "How many days of slack do we have?" is the plan's slack.

It will not advise how to recover the date. "How do we get back on track?", "What do I cut to make the date?", "What would it take to be back on target by March?", "Quickest way to pull the date back?" are refused: "That asks what to change to recover the date, which this panel does not advise." It works out what the plan as it stands will do; to see what a change would do, ask it as a what-if — one more person, a later date, less scope — and "What would one more person do?" is offered first under the refusal. A what-if that names its change ("If we add a developer, do we make the date?") is answered as a what-if.

It will not answer about a person the project does not hold. Ask what somebody is carrying and the project has nobody of that name — not on the Team & Capacity tab, not assigned any of its issues — and it says so by name: "Nobody called Zoë Brennan is on this project, so nothing was answered about them." The people it knows are the Team & Capacity tab's members and the assignees of the issues it holds.

A question it cannot place is refused, not guessed. When the words place a question nowhere, the answer opens "I could not tell what this asks for.", says beneath it the nearest reading it found, and offers questions it can answer, each a press away. A sentence that only brushes the plan with a word like "team", "epic" or "risk" and names nothing of the plan itself is refused the same way, instead of being answered with the delivery forecast. Under a refusal too, the line beneath the question box says why Claude's reading was not used: it did not answer in time, it picked no figure, or its reading could not be fetched (12 November 2026).

"Did you mean …?" When the AI reader (below) finds no figure for a question but names the nearest one, the refusal asks whether that figure was meant: under "I could not tell what this asks for." it reads Did you mean "Who is overcommitted?" and Press it below to ask it; the others are what it can tell you about this project instead:, and that question is the first of the questions offered beneath, the others after it. The question is the panel's own wording for the figure, filled with the project's own names — a person, sprint or epic the question named where the project holds it, the project's first otherwise — never words the reader wrote; a figure whose question the project cannot fill with its own names is not offered. It appears only where the panel would otherwise have refused: where the question is answered, nothing is offered beside the answer.

Which figure a question asks for may be read by an AI; no figure and no sentence comes from one. Inside Jira, every question is sent — except one the panel refuses for a stated reason and one of the panel's own questions picked from its list — with the list of figures the panel can answer, to Claude through Atlassian's own service, which says which one figure the question asks for and which names it carries — or that none fits, in which case the plain phrases' reading stands. Every figure is still worked out by the app from the project's own data, and every sentence around the figures was written into the app itself; the reader never supplies a number or a word of an answer, and the line under each answer says how the question was read. Where the plain phrases' reading stands, that line says why, in the words for what happened: "this site's monthly allowance of questions read by Claude is used up until next month"; "Claude did not answer in time" (it took longer than eight seconds); "Claude's reading could not be fetched" (Atlassian's service returned an error, and the question is not counted against the allowance); "Claude's reply could not be read" (the service answered but its reply held no figure the app could read; it is counted); or "Claude picked no figure for this". After "Claude picked no figure", where the plain phrases placed the question nowhere, Claude is asked once more for the nearest figure, and the line says what that gave: the figure it named (offered as "Did you mean"), none, "it did not answer in time", or "its reading could not be fetched" (an error from the service, not counted against the allowance). When the Jira site has reached Atlassian's limit on questions read by Claude in one minute, no answer is given unless the question is in the app's own wording for a figure: the panel shows one line under the question, "This Jira site has reached Atlassian's limit on questions read by Claude in one minute. Ask again in a minute.", with no figure, no Detail and no questions offered under it, and the next question is read as a fresh one. A question in the app's own wording for a figure is still answered, and its reading line says the limit was reached and to ask again in a minute. When the limit is reached on the second call, the one for the nearest figure, the refusal stands and its line says so in the same words (13 November 2026). When Claude cannot be reached or does not answer in time and only the broadest of the word rules placed the question, the panel refuses and offers the nearest questions, and does not guess with the delivery forecast (12 November 2026). The web app's public copy has no reader and reads by the plain phrases alone.

One question of up to 500 characters, in English. A longer one is answered with how long it is and a request to ask the one thing you want to know; one in another language is answered "English only: this reads questions in English."

One project read at a time. Where a question names several projects by their keys, the panel answers them one after another, each as its own answer, with a line such as Reading SHOP… while each one is worked out. A project other than the one on screen is not held by the page, so it is read from Jira for that question, and reading one project takes ten to fifteen seconds, so two do not fit in the time a call is allowed. Where the question asks for them together, such as whether all of it will be delivered by a date, one more answer adds their forecasts together.

Asked while the projects are still loading, an answer across them says so (5 November 2026). The app reads every registered project when it opens, one after another, and the Projects tab shows Loading on a row until its figures arrive. A question across the projects asked in those seconds — which are at risk, which has the least slack, how many are on target, which finishes first or last, the scope across every project — opens with how many are still to come and which, and counts only the ones read: "Still reading 2 of the 4 projects (HRB, NOVA), so this covers only the 2 read so far. No project is at risk: all 2 read healthy in the Projects tab's Health column." It does not open Yes or No on part of the list. With every project still loading it says there is nothing to answer from yet; a question about one project that is still loading says that project is still being read. Ask again once the rows have loaded for the whole list.

A project without estimates is counted in issues (3 November 2026). Where nothing is estimated, the counts are said in issues, as the Dashboard's Progress card counts them ("6 of 25 issues"), never as points, and the dates are only as good as a count of issues allows. A figure that needs something measured is said as not measured, never as 0: on closed sprints that planned nothing, "Efficiency is not measured yet: the 2 closed sprints planned nothing to measure it against, so nothing is discounted." On a project that does estimate, work carrying no estimate is in none of the totals, and the answer says how much of it there is rather than quietly leaving it out.

It will not say how accurate the estimates were. The app does not measure it (Don, 21 October 2026), so "How accurate are our estimates?" in any words is answered "That asks how accurate the estimates were, which this app does not measure.", with "How much of what we committed did we deliver?" offered first (or "How much do we finish a week?" on a project with Sprint Mode off).

It will not answer a different subject. A question naming risks, actions, labels or holidays whose nearest figure is about something else is not given that figure: "That asks about risks, and no figure about risks answers it, so nothing was given.", with the figure it passed over named under Detail and that subject's own questions offered.

Nothing infers. It will not read Slack, guess an undocumented dependency, or tell you a risk is forming. Everything it says is arithmetic on what Jira holds, and every figure can be checked against a screen.

How it works

The words place the question first. In the web app, and inside Jira whenever Claude names no figure, a question naming one of the figures the app shows — by any of the plain phrases that name it — is answered with that figure; a sentence no figure claims is read by a list of word rules that decide which kind of question it is; and a sentence those place nowhere is refused, saying how it was read, with the nearest questions offered. Short forms and typing shortcuts are read the one way first: "cap" is capacity, "wk" is week, "hrs" is hours, "whats" is "what is". The project's own name in a sentence names nothing but the project on screen, so "What percentage of Web Shop is complete?" asked on Web Shop is its progress.

A question about what a figure means gets its definition first. "What does slack mean?", "How is velocity worked out?", "What counts as remaining work?" — and "What is velocity?" on its own — open with the figure's definition (what it adds up, what it subtracts, where it comes from and which screen sets it) and carry the project's own figure beneath as a second answer. "What is our velocity?" is the figure. The question can be put in any words and at any length: "What does 'capacity to target' mean on the forecast card?", "feasibility score meaning", "KPI Load definition?", "hrb load 132%, what's that worked out from", "odds 20% — odds of what exactly?", "OPTIMAL means?" all get the definition, and the screen's own words for a figure — Load on a sprint row, odds on the Dashboard, Net on the Team & Capacity tab, "Scope outpacing delivery" on the Delivery Forecast card, P85 on the Scope tab's confidence band — are known to it. A figure the panel cannot fetch (P85) or one asked about with no person or sprint named ("What does the Net column mean?") gets the definition alone, with no second answer under it. The figure defined is the one the question names (5 November 2026): "the feasibility score" is the Deliverability score (on no screen since 15 November 2026, when the What-If tab's Simulation lost its Deliverability box), and a question asking for it as a figure ("What is the feasibility score?") is answered with that score, named Deliverability as that box named it; "What does the 132% on Team Capacity mean?" is the Dashboard's Team Capacity row in Project Statistics, the team's load; "what does utilization on the Projects tab measure" is that tab's Utilization column; Committed size and Live estimate are the two choices of the Remaining work setting; Effective team capacity and Team capacity, named together, are two of the Delivery Forecast card's capacity models. A question asking for the difference between two figures — "What is the difference between velocity and effective capacity?" — opens with both definitions, in the order asked, and carries no second answer.

Four lines that say what their figure is (12 October 2026). The latest finish is said as odds: "There is a 90% chance of finishing no later than Apr 21, 2027", and the earliest as "There is a 10% chance of finishing before ...". When more than one run in ten never fits the planned capacity there is no 90% date, and the Delivery Forecast card shows no "likely … – …" range either; the line then gives the date that does exist, the share of runs that fit and the latest of them ("There is an 85% chance of finishing no later than May 3, 2027"), with the reason under Detail. Slack says what the days are: "The plan has 2 days of slack to meet the target date, Apr 21, 2027", or "is 5 days short of" it. "What would move the date most?" names the four it tried, velocity, issue estimation, scope and capacity, each at a 15% swing. And "Is the target likely?" uses the app's one on-time band, the one the What-If tab's "Finish on time?" box used until it was taken off on 15 November 2026: 70% and over is Likely, 50% to 65% Maybe, under 50% Unlikely. The odds are said to the nearest 5%, as the Delivery Forecast grid's squares show them (12 November 2026; the Delivery Forecast card has shown no odds since 15 November 2026): 250 runs cannot tell 68% from 72%. A "Yes." or a "No." is said only when every run, or none, met the date. "Will the work ever finish?" gets neither a Yes nor a No when the work does not fit the planned sprints (12 November 2026): it gets the Delivery Forecast card's own sentence that the work does not fit the plan, with how much does not fit.

The start date, when none is set (12 November 2026). Where the Scope tab's Start Date card reads "Not set", "What is the start date?" says so too: "No Start Date is set; the first issue was created Aug 11, 2026.", with where to set one under Detail.

Sub-tasks, as the project's Sub-task estimates setting counts them (12 November 2026). The setting is on the Projects tab's Edit form. The panel tells a sub-task by Jira's own mark on the issue type, whatever the type is called; a question naming another project is answered on that project's own setting; and the scope answers add up delivered and cancelled work the way the setting says, as the Scope tab does.

A question about one thing is not answered with the delivery forecast (13 October 2026). A question about a person's work ("whats Test3 got this sprint", "list everything Test1 is assigned in the running sprint"), a person's capacity ("Test1 hours left", "how many pts can Test1 take next sprint"), an issue's status ("SHOP-41 done yet"), the team's capacity in a sprint ("How many hours does the whole team have in Sprint 33?"), a sprint looked back on ("last sprint summary", "What rolled into the next sprint?", "Was last sprint better or worse than normal?"), time off ("Does Test2's holiday overlap the end of Sprint 15?"), work depending on a later sprint, the Alerts tab's lists ("which issues are still open in Sprint 14 even though it is finished", "which issues are past their due date") and what needs attention ("What's most likely to hurt delivery?") each get their own figure, however the question is worded and however long it is.

A follow-up reads on from the turn before. "What is Diana carrying?" then "and Marc?" is read as "What is Marc Oberon carrying?"; "how much capacity does she have left" after a question about Diana Lee is about Diana Lee; "and what percent is done" is "what percent is done". After an answer about an epic, "that epic", "it" or "the same one" is that epic: "What is under the Payments epic?" then "how far along is that epic?" is read as "how far along is the Payments epic?". After a what-if that moved one of the four What-If variables (the What-If tab itself has shown only Capacity and Remaining work as sliders since 16 November 2026), "another 10%", "10% more" or "the same again" adds to the amount already in force on that slider: "What happens if scope drops 15%?" then "another 10%" is read as "What happens if scope drops 25%?". The reading line says so — "Following on from …" — and nothing else is carried between questions.

A request for a status update, summary, report or one-liner is answered with the brief of what needs attention, worst first, one line each. A request for a file is still the Report tab's.

An Alerts check asked for by its name — "which issues are done with work still on them", "does anything have a due date before its start date" — is answered from the Alerts tab's own list of what each check found, with where its list is.

The engine answers it. The same functions the tabs are drawn from — the same forecast the Dashboard shows, the same words the Sprints tab uses, Deliverable or Overcommitted, the same capacity arithmetic the Team & Capacity tab does. The panel adds no calculation of its own. If an answer ever disagreed with a screen, one of them has stopped calling the shared function, and that is the bug to look for. Every figure the panel gives is checked against the tab that shows it, on the demo projects and on real boards read from Jira (since 20 October 2026). So the counts of issues with no estimate, no due date, no epic, nobody assigned, or still open in a closed sprint are the Alerts tab's own checks; the backlog is the Sprints tab's Backlog row; the cumulative demand against capacity is the Scope tab's Demand vs. Capacity section; and every answer that names the plan's finish date names the Delivery Forecast card's date.

No AI writes any part of an answer. One call does the whole job: the question is placed (inside Jira by Claude, as described above; in the web app by the plain words), the engine works out the figures from the project's own data, and the sentences that carry those figures are the app's own, written into it. There is nothing to set up. The web app sends nothing to an AI model.

A question about the plan produces one large set of figures, and the answer is written from that set. This is why the answers feel broader than any list of questions: one question reaches the dates, the sprints, the people, the dependencies, the chain, the alerts and the data gaps at once, and you read the part you asked about.

Asking writes nothing to Jira. The panel reads the project's Target Date, Sprint Capacity and Estimation Mode from the Projects tab and never changes them. Whether the backlog counts is the project's Include backlog setting, the same one every tab reads. With it ticked, the panel plans backlog work exactly as the Delivery Forecast card does: in the open sprints, each issue by its own dates, so its dates and its "does not fit the plan" are the card's (13 November 2026). Each sprint's own figures stay the ones its card on the Sprints tab shows, which count no backlog. An answer about one sprint's demand adds a second sentence giving that sprint's demand with the backlog work the forecast plans in it - "With the 40 pts of backlog work the forecast plans in it, that's 50 pts." - and the sprint table under Detail gains a With backlog column beside Remaining demand. With Include backlog off, or no backlog work planned in a sprint, neither appears. A date, a capacity or a unit named in a question is used for that question only and is not carried into the questions after it; it is stored nowhere. The one thing carried from question to question is a what-if, and it is shown in the amber line the whole time it is in force. To change a setting for good, edit the project on the Projects tab.

No sheets in or out. Asked for a sheet to fill the team in on, the panel says it does not send out a sheet, and that the team, their capacity and their leave are typed straight onto the Team & Capacity tab. Asked for the forecast as a file, it says it does not make files and that the Report tab sends the forecast out.

© 2026 Project Commander · projectcommander.app · Support · Privacy · Security · Terms