Ask
Read, explain and cite. Ask why a milestone slipped, what is overdue, or what changed this week, and every answer links back to the activity, task, post or note it came from — nothing changes.
Planrill AI
Planrill AI works
inside your projects.
An assistant that reads what you can see — activities, tasks, the Feed, notes, meetings, files and links — explains what needs attention with links to the real items, and proposes changes you approve. Nothing moves until a person applies it.
Two modes
Ask, or act — you choose
The same assistant, two clearly different jobs.
Act
Propose changes for your approval. Reschedule, reassign, add tasks, post an update, append to a note — Planrill AI drafts the exact changes as a proposal, and you apply them with one click or decline them.
Nothing changes until a person clicks Apply. Every applied change is logged with who requested and who approved it, and can be undone while the item is untouched.
Where it works
Global, project, or the page you are on
Open it anywhere; it already knows where you are.
Across your portfolio
At Global scope it looks across every project you can see — what is at risk, what is overdue, what needs you today.
Inside one project
In a project it reads that project's schedule, tasks, Feed, notes, meetings, documents and links, and answers with the evidence.
On the page you are on
Page quick actions send the question that fits the screen — explain this schedule, summarise this thread, draft a status update.
Every morning, at no cost
A morning brief and a nightly check for milestones at risk and work gone stale are built from your data directly — no model call, no credits, just the facts and a link to each item.
Approval built in
It proposes, you apply
Four steps, none of them automatic.
A proposal, not a change
In Act mode the assistant files a proposal: the exact changes, item by item, with a plain-language summary.
One click to apply
You review each item and apply the proposal — or hand it to a colleague who has the permission to apply it and let them decide.
Logged with provenance
Every applied change lands in the project's activity log with who requested it, who approved it, and a 'via Planrill AI' mark.
Undo while it is untouched
Applied a proposal by mistake? Undo it. The assistant refuses to undo anything a person has since edited.
Beyond one question
Automations and plans, on your terms
Approve once; it runs within what you approved.
Approved automations
A rule is one trigger and one action — when an activity is completed, post to the Feed; when a task goes overdue, assign it. A project admin approves the rule, it runs under that person's permissions, and it pauses itself if it fails or that person loses access.
Delegated plans
Give it a goal and it lays out up to six steps inside an envelope you approve: one project, non-destructive changes, a bounded size. Steps inside the envelope run; anything outside it waits for your click.
What it knows
Your workspace, and only what you can see
The assistant reads through your own permissions — never more.
Search and files
It searches posts, comments and notes, and reads the text of PDF, Word and plain-text files uploaded to your projects, so an answer can cite the document, not just the task.
Memory only when you ask
Tell it to remember something and it will; tell it to forget and it does. Project memories are on the record; personal ones are private to you.
Private by design
Conversations are private to you. Questions and the content the assistant reads go to the AI processor only to produce the answer, and are never used to train models. The privacy policy names every processor.
Provider-agnostic
Answers come from Google's Gemini API today through a neutral layer, so the model can change without your workspace changing. Which models are offered is configuration, not code.
Credits and plans
Included in every plan
Usage is metered in Planrill AI credits — a monthly pool for your organization, visible in Settings.
- Free: Planrill AI with a starter allowance
- Pro Flex: Planrill AI included
- Pro Unlimited: Planrill AI included
- Enterprise: Planrill AI included, more by agreement
The morning brief and the nightly checks never spend credits. Your organization can turn the assistant off entirely.
Compare plansQuestions people ask
Does Planrill AI change my projects on its own?
No. In Ask mode it only reads, explains and cites. In Act mode it files a proposal, and nothing is applied until a person clicks Apply. Every applied change is logged and can be undone while the item is untouched.
What can it see?
Exactly what you can see, and nothing more: it reads through your own permissions, in the scope you open it in — across your portfolio, inside one project, or on the page you are on. Your conversations are private to you.
Which files and content does it read?
Activities, tasks, the Feed, notes, meetings, calendar events, links, and the text of PDF, Word and plain-text documents uploaded to your projects. It searches posts, comments and notes too.
Which model answers, and is my data used for training?
Answers come from Google's Gemini API through a provider-agnostic layer. The content it receives is used only to produce the answer, never to train models. The privacy policy names every processor we use.
What are Planrill AI credits?
A monthly pool for your organization, included in every plan and shown in Settings. Each answer spends a small number of credits; the morning brief and the nightly checks are built from your data directly and spend none.
Can it run things automatically?
Only within what a person approved: an automation is one trigger and one action approved by a project admin, and a plan is a short sequence of steps inside an approved envelope. Both run under the approver's permissions, log every change, and pause if something fails.
Ready when you are
Bring Planrill AI into your next project.
Start free, ask it what needs attention, and apply its first proposal in minutes.
Get started Reads what you can see. Proposes. You apply.