Use case
Someone asks what ships when, and the answer lives in a spreadsheet that was right in week one. Spectr reads every requirement in the project and plans milestones and epics from them, dated from what your team has actually finished. Every story is filed under an epic, and each bar fills as stories close in Jira, Linear or Shortcut.
How it works
Spectr reads every requirement in the Partner Dashboard project, three documents, including the target date one of them states (Sep 30). One run returns three dated milestones and seven epics, planned from the team's pace and what is already done in Linear, and files the project's stories under the epics. Each bar fills as its stories finish in the tracker; cancelled stories are left out. When the security addendum changes, the PM runs Update with AI and asks to finish by the end of September. It suggests changes: an epic added, one renamed, and the milestones' dates at the team's pace, which now run to October 2. The dates squeezed to September 30 are offered as a separate choice that starts unselected. Nothing changes until the PM applies it.
A roadmap is built once, from the documents as they stood, and kept up by hand. Then a requirement changes, a story gets cancelled, a section moves to phase two. Nobody re-plans, because nobody has the afternoon. So the dates on the slide come from the kickoff, and the honest answer to "what ships when" is a guess.
An agency PM. A partner dashboard, three requirement documents, 132 stories in Linear. The client wants dates.
Old way
You read the three documents again, group the stories into epics in a spreadsheet, guess a length for each and colour the cells. Most of a day, and out of date at the next requirement change.
10:00am
Now: on the Delivery page you pick the project and click Generate milestones and epics. Spectr reads all three documents, the draft billing spec included.
10:02am
Three milestones and seven epics come back, dated. The first runs from July 6, when its first story started in Linear. The rest are laid end to end at the pace the team has kept.
10:03am
128 of the 132 stories are filed under the epics. Four fit none, so they are left for you to place rather than guessed.
10:10am
The target, September 30, was read from the PRD with the sentence it came from. You screen-share the roadmap on the client call.
Two weeks on
The security addendum changes and the page says so. You run Update with AI and ask to finish by the end of September. It suggests an Audit log epic and new dates that run to October 2, and offers dates squeezed to September 30 as a separate choice, unselected.
Next call
You apply the epic and the at-pace dates, and tell the client the audit log moves the date by two days. Or you cut it. Either way it is a decision, not a surprise.
A day of re-reading documents and colouring a spreadsheet has turned into ten minutes of reviewing a plan, and the re-plan after a requirement changed took one review instead of another day.
One run reads every requirement in the project, working drafts included, so a feature described across a PRD and an addendum lands in one epic. Very large documents go in as outlines rather than being cut off. Epics come back grouped into milestones in delivery order, and the engineering groundwork gets an epic of its own.
Finished milestones keep their real dates from the tracker. Work under way starts when it started. The rest is laid end to end from today, at the pace your team has actually kept. A target date stated in the requirements is drawn on the chart. The plan aims for it, but never shortens work to fit.
Existing stories are filed when the plan is made, and new ones as they are generated. A story that fits nothing is left unfiled, not guessed. Choose a story's epic yourself and Spectr never moves it. Progress comes from Jira, Linear and Shortcut, and rejected or cancelled stories are left out of it.
When a requirement changes, or stories finish or get cancelled, Update with AI re-reads everything and suggests changes: epics added, renamed, merged or archived, and new dates. Give it an instruction and it follows it. Nothing changes until you apply it, and an edit you made by hand since is not written over.
Nothing on the page says late, at risk or on track. The bars show what is done and where the rest sits, with today and the target as lines. Whether to cut scope, add people or move the date is your call.
Anything else? Email hello@crowdlinker.com and a person will answer.
No. Nothing is planned until you click Generate, and an update runs only when you ask. The page tells you when a requirement has changed since the last plan.
Product Delivery is on every plan, Free included, with a monthly allowance of delivery revisions: 5 on Free, 25 on Starter, 100 on Pro and 500 on Business. A first plan and each Update with AI is one revision, and a run that fails costs nothing. Reviewing, applying and editing by hand are free.
Yes. Add, rename, date, reorder and delete milestones. Add, move, archive and restore epics. Change a story's epic from the story itself.
No. It shows how much of each milestone is done and where the rest falls against today and the target date. Whether that is a problem is for your team to decide.
The plan works from your requirements alone. Progress and pace come from stories' status in Jira, Linear or Shortcut. Until five stories are done there, the plan assumes the pace of a small product team.
