Portfolio by HeroCoders is good at what it was built for: one table and one Gantt over every project, with each person's bookings laid out by the day. Project Commander was built for a different question. This is a plain account of where each one is strong, taken from HeroCoders' own help centre and Marketplace listing and from Project Commander's own screens. Sources are listed at the end.
Every delivery team eventually gets asked the same thing: will this be delivered by the date, with the people we actually have — and if not, where does it break?
Portfolio by HeroCoders lays every work item out in a spreadsheet-like table and a Gantt chart across projects, rolls up progress and dates to the parent, and books each person's hours between a start and an end date, marking the days they are overbooked. Project Commander takes the work actually assigned in the sprints, the real people and the date, does the arithmetic, and tells you whether the plan holds — sprint by sprint, person by person, with odds — and keeps telling you as the plan moves. With or without sprints: a project with Sprint Mode off is read from its Issue filter and judged week by week, so a Kanban or date-planned team gets the same answer.
| Built to answer | |
|---|---|
| Portfolio — Table and Gantt | Where does every epic and work item sit across projects, how far along is each parent, and which child runs past it? |
| Portfolio — Capacity Planner | On which days is each person booked past their hours, and what would moving a booking do? |
| Project Commander | Can this be delivered by the date with the people we have — and if not, where, by how much, and what do we do? |
1. It plans with the people you actually have. Capacity is worked out per person — hours or points a sprint, how much of their time this project gets, booked time off, company holidays — and added up to the sprint. Teams that do not run sprints get the same arithmetic: with Sprint Mode off in the project's settings, a Kanban or date-planned team's work is read from its Issue filter and weighed week by week against capacity for a period of a length you choose. Portfolio also works per person, with hours per day, working days, per-person schedules, time off and public holidays. It spreads a work item's Original Estimate across the days between its booking start and end dates, so every item needs those two dates before it counts. Project Commander works per sprint, in the team's own unit — story points or hours — and needs no booking dates: the sprint is the window.
2. Every sprint carries a verdict, and it means what it says. Deliverable, Tight or Overcommitted, with the load as a percentage, on every sprint row. For the sprint running now, it compares only the work still left against only the capacity still ahead — half a sprint gone means half the capacity gone. Portfolio marks a person's overbooked days red. There is no verdict on the sprint, no percentage of the sprint's capacity, and no allowance for the part of the running sprint already gone.
3. It forecasts a finish date, with odds, not a wish. The forecast walks the plan sprint by sprint, carries overflow forward, and names the person who becomes the bottleneck. Then it runs the plan two thousand times against the team's real variation and reports the probability of hitting the target and a likely-finish range. Portfolio's calculated end date for a parent is the latest end date of its children — a roll-up of dates people typed, not a forecast. Its pages describe no probability.
4. It will tell you "never." If scope is growing faster than the team clears it, the forecast says the work does not finish — not a comfortable date two quarters out. Portfolio does not make that call; it has no forecast to make it from.
5. It sees scope creep as it happens. From Jira's own change history it rebuilds what was added, re-estimated or removed after the plan was agreed — per sprint and over time — and flags a running sprint that has quietly grown past its commitment. Portfolio shows the plan as it stands today. It does not read Jira's change history.
6. It finds the dependencies that break a plan, not just draws them. A task blocked by work that is not scheduled until later. Two tasks that block each other, so neither can ever start. A sub-task due after its parent. Each is an alert with a way to resolve it. Portfolio's Gantt draws dependencies with lag times, and its Delivery Alert flags a child whose end date runs past its parent's. It does not detect two items that block each other, or an item blocked by work scheduled in a later sprint.
7. It names the data that would poison a forecast. Missing estimates, missing dates, unassigned work, items marked done with time still remaining, overdue items — thirteen kinds of alert in all, each from a rule. A forecast built on half-filled fields is worth nothing, so it says which fields. Portfolio's warnings are an overbooked person, a child ending after its parent, and an item past due that has not started. Missing estimates, missing dates and unassigned work are not among them.
8. It shows you who is overloaded from the work, not from a timesheet. Per person, per sprint, against their own capacity — read from what is actually assigned to them. Across several projects, one person's shares add up and turn red past a hundred per cent. Portfolio reads the Original Estimate, which is a fair start, but spreads it over booking dates by the day. Project Commander reads what is assigned to the person in each sprint and compares it with that sprint's capacity for that person.
9. Auto-Level rearranges the sprints so they fit — and shows you before it touches anything. When sprints are overcommitted, Auto-Level moves work between them to fit the capacity that is actually there. It orders the work so nothing is placed before what it depends on, treats each person's own capacity as a hard limit, uses only the capacity still left in the running sprint, and offers four strategies — by priority, by size, by due date, or balanced across people — over a horizon you choose. Two tasks that block each other are flagged, not skipped. The result is a preview: every sprint before and after, the change to the forecast date, and a line explaining each move that a constraint forced. Nothing is written to Jira until you accept it. Portfolio rearranges by drag and drop. There is no automatic levelling and no preview of the whole plan before and after.
10. What-if you can feel. Sliders for velocity, estimates, scope and capacity; the date and the odds move as you drag. Portfolio has no what-if.
11. Risks with a paper trail. A register of risks and dated actions, entered by people or proposed from patterns in the numbers with the arithmetic printed underneath, with owners and response strategies, and alerts when an escalation is waiting or a review is due. Portfolio has risk warnings on rows. It keeps no register with owners, actions or escalation.
12. Every number is arithmetic. Nothing on any screen is inferred, sensed or guessed. Where AI is switched on, it explains and suggests, and never produces a figure — a guard in the code checks every quantity an answer states against the data it was given. Portfolio's figures are arithmetic too. The difference is what they are arithmetic about: bookings by the day, not sprints.
13. Your own AI provider, your own rules. AI in Project Commander is off until you add a key of your own, and you choose the AI provider — Anthropic, OpenAI or Google — on the Settings screen. Nothing is sent at all until someone presses one of the seven places the AI can be asked from. Portfolio's Portfolio Insights agent runs on Atlassian Rovo. Your data goes to Rovo under Atlassian's terms, and you do not choose the provider behind it.
14. It runs inside your Jira, and nothing leaves. A Forge app: your project data stays in your Atlassian instance. Free during the beta. Portfolio is also a Forge app, and HeroCoders says plainly that it stores none of your data. On this point the two are on the same footing.
Be as clear about this as about the rest.
Plenty of teams keep the portfolio in Portfolio and use Project Commander to pressure-test whether each project's sprints hold. They are not in competition for the same job.
| Project Commander | Portfolio by HeroCoders | |
|---|---|---|
| Capacity per person, with time off and holidays | Yes | Yes — hours per day, spread over booking dates |
| Deliverable / Tight / Overcommitted on every sprint, with a percentage | Yes | Overbooked days per person |
| Running sprint judged on what is left against what is left | Yes | — |
| Forecast finish date | Yes, sprint by sprint, naming the bottleneck | Calculated end date: latest child's typed end date |
| Odds of hitting the date, and a range | Yes — two thousand runs | — |
| Says "never" when scope outruns the team | Yes | — |
| Scope creep rebuilt from history | Yes | — |
| Dependency conflicts: blocked by later work, circular loops, child after parent | Yes | Gantt lines with lag; child-after-parent alert |
| Missing estimates, missing dates, unassigned, done-with-time-remaining, overdue | Yes | Overbooked; child after parent; past due and not started |
| Who is overloaded, from assigned work | Yes, per person per sprint, and across projects | From estimates spread over booking dates, by the day |
| Rearranging work to fit capacity | Auto-Level: per-person limits, dependency order, running sprint's remaining capacity, four strategies, preview before anything is written | Drag and drop |
| What-if | Sliders with live odds | — |
| Programs of several projects, shared people | Yes | Yes — portfolio table, Gantt, cross-project capacity |
| Works with sprints and without them (Kanban, date-planned) | Yes — Sprint Mode on or off, per project | Yes — bookings by the day, no sprints needed |
| Risk register with owners, actions, escalation alerts | Yes | Row warnings only |
| Baselines — how the forecast has moved | Yes | — |
| Portfolio table with roll-ups, formulas, budgets; Gantt | Projects table and program roll-up | Yes — its core strength |
| AI provider — whose, on whose terms | Your own: Anthropic, OpenAI or Google, your own key, off until you add one | Atlassian Rovo agent |
| Runs inside Jira; data stays | Yes | Yes — Forge |
| Cost | Free during beta | Paid per user on the Marketplace |
Portfolio lays the whole portfolio out and shows who is booked past their hours on which day. Project Commander tells you whether the work in the sprints can be delivered by the date with those people — where it breaks, by how much, and what to do about it.
HeroCoders documentation and listing, as of 5 September 2026:
Project Commander: projectcommander.app — features, documentation, and the live demo.