> For the complete documentation index, see [llms.txt](https://portavi.gitbook.io/portavi-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://portavi.gitbook.io/portavi-docs/ai/the-assistant.md). # The assistant The assistant answers questions about your workspace in plain language: what is at risk, which tasks are overdue, how the current sprint is going, when the next strategy review is. It can also log a risk, an issue or a lesson for you, but only once you confirm it. There are four ways to reach it: * **The assistant button** in the bottom corner of any page. On a computer it opens as a panel beside the page, so you can keep working, and the button steps out of the way until you close the panel; on a phone it fills the screen. * **Assistant** in the navigation, under My Focus (under Overview for executives), which opens the [assistant page](#the-assistant-page). * **The field at the top of** [**My Focus**](/portavi-docs/tracking-delivery/my-focus.md)**.** Ask there and the assistant page opens with your question already being answered. * **Ask more** on a project that is at risk or delayed, at the end of the [explanation in its header](/portavi-docs/tracking-delivery/project-detail.md#why-a-project-is-at-risk). The panel opens already asking why.
The assistant page: past chats in a side menu and a large question field in the middle
## What you can ask It reads the same data you can see, and nothing more: * **What's waiting on you**: everything on your [My Focus](/portavi-docs/tracking-delivery/my-focus.md) page, from slipping dependencies and due reviews to requests waiting for your decision and your own due tasks. Ask "what's on me?" or "show me my pending actions" * **The RAID log**: risks, issues, assumptions, decisions and dependencies * **Lessons learned** and **change requests** * **Status**: a project's health and why it is at risk or delayed, with its sprint, phase, cadence or milestones; and a programme's or portfolio's projects by status, which need attention, and its benefits or objectives * **Delivery**: synced tasks, flagged items, sprints, cadences, phases and milestones * **Signals** from your chat tool: blockers, issues and progress updates * **Strategy**: objectives and key results, programme benefits, and when each review is due * **Budgets** and **team capacity**, where your workspace uses them * **Stakeholders** * **Intake requests**, where your workspace uses Intake: what's waiting to be screened or decided, the highest-scoring requests, how your own are doing, and why one was approved, rejected or deferred. It sees only the requests you can open in [Intake](/portavi-docs/intake/requesting-new-work.md), and on the Intake page its suggestions ask about the pipeline A question about an area your workspace has switched off gets a one-line answer saying so. As elsewhere in Portavi, capacity is visible to PMO managers and executives only, and stakeholder analyses are not shown to executives. **On a project, programme or portfolio page, "this project" means that one.** The suggestions in an empty panel follow the page too: on a scrum project they ask about the sprint, on a waterfall project the phase. **Ask how to do something** ("how do I invite someone?", "how do I connect Jira?") and it shows the help article that covers it. Select the article to open it beside the page, at the section that answers your question. When a question could mean several things ("show me the risks" when you look after six projects), the assistant asks which one, with a list to pick from. Every answer shows the records themselves, laid out as they are on their own pages, and each one opens where it lives. A task opens in your project management tool. {% hint style="info" %} **It cannot tell you what a chat message said.** Portavi never stores message text (see [communication tools](/portavi-docs/connecting-your-tools/communication-tools.md)), so the assistant shows who raised a signal, where and when, with a link to the message itself. {% endhint %} ## Making a change Ask it to log something and it shows you exactly what it will save: the title, the fields, the project it goes on. **Nothing is saved until you select Confirm.** Select Cancel, or ask for something different, and nothing changes. It can: * log a **risk**, an **issue**, an **assumption** or a **lesson learned** * change a risk's **impact** and **likelihood**, or mark it as **occurred** * add an **objective** to a portfolio, with its key results * add a **benefit** to a programme * raise a **change request**, as a draft or straight into approval It follows the same rules as the pages. You can add to a project or programme you own; a PMO manager can also add objectives and benefits. Executives can ask anything, but the assistant does not make changes for them. A proposal is good for ten minutes. After that, ask again. It does not delete anything, record decisions, decide or apply change requests, or change tasks. Tasks belong to your project management tool, and Portavi only reads them. For those, use the page itself. ## The assistant page The assistant page gives a conversation the whole screen. Your past chats are in the menu on the side, with the **+** beside their heading to start a new chat, and the question field sits in the middle of the page until you ask something; then the conversation fills the page and the field moves to the bottom. On a phone, the clock at the top left shows your past chats in place of the conversation, and the **+** at the top right starts a new chat. To move a conversation from the panel to the page, select **Open full page** (the two diagonal arrows) in the panel's header. It is the same conversation: an answer still being written carries on there. The assistant button doesn't show on the assistant page. ## Past chats Your conversations are saved as you go. Select the clock in the panel's header to see your past chats, newest first and grouped by date (on the assistant page they are always in the side menu), and open one to carry on where you left off. **New chat** starts a fresh one; reloading the page does too. * **Only you can see your chats.** Not your PMO manager, not the workspace owner. * **Each chat is deleted 90 days after its last message.** Delete one yourself at any time: point to it (or tab to it) and choose **⋯** > **Delete**. **⋯** > **Rename** gives it a name of your own. * **Leaving a workspace deletes your chats in it.** * **An answer keeps going when you move on.** Start a new chat or open another while one is still answering, and it finishes in the background: its chat shows the animated AI mark while it works and a dot once the answer is ready. Reloading or closing the page stops it, and keeps what it had written so far. A change it proposed in an old chat can no longer be confirmed, because proposals expire. Ask again in the chat to get a fresh one. ## Good to know * **One answer at a time.** If you send another message while it is answering, it waits and goes as soon as the answer finishes. **Stop** ends an answer early. * **It stays on topic.** It answers questions about your workspace and how to use Portavi, and politely declines anything else. * **It replies in the language you write in**, whatever your Portavi language is set to. Names of projects, risks and people are quoted exactly as they are stored. * **It can be wrong.** It reads the same records you can open, so check one before acting on it. ## Which plans include it The assistant is part of AI in Portavi, on **Growth** and above. It has its own monthly allowance for each workspace (1,500 messages on Growth, 5,000 on Scale and Enterprise), separate from the one that report summaries and rewrites share, so a busy month of questions never stops a report summary. PMO managers can see how much of it is used under **Assistant messages this month** in **Settings → Billing**. If you send a lot of messages in a few minutes, it asks you to wait a moment.