What's Next

On Home, What's Next lists the handful of things most worth your time today: deals that have gone quiet or are past their close date, new leads nobody has contacted, meetings nobody followed up, people who just clicked your email, goals falling behind and past customers worth a catch-up. The assistant reads each one in context first, keeps what is worth doing, says exactly what to do and drafts the email. Nothing changes until you approve it.

What you get

• What your orders say — Products your customers buy together more often than chance, and how long they usually wait before buying the same thing again, each turned into a suggestion: one click builds a live segment of the people it applies to, ready for a campaign or a flow.
• Goals in plain words — Type a goal the way you would say it, such as “Book 30 meetings this quarter” or “Win 50,000 in new business this quarter”. Solstral works out how to measure it from your own data, shows you exactly how it will be counted, and saves nothing until you press Set goal.
• Read in context, not just matched by a rule — Before a suggestion reaches anyone, the assistant reads the meeting and the answers given when booking, the last few notes and emails, and the deal behind it. A routine call with a customer or a supplier's enquiry is set aside with the reason; what is left comes with the specific next step and, where an email is the move, a draft you send yourself.
• A button that does it — For your own quiet deals and new leads, Do it builds a plan of follow-up tasks, one per record, that you approve on a card before anything changes. Other suggestions open the assistant with a prompt that drafts the emails or plans the catch-up, and it never sends, schedules or publishes without asking.
• Every number is a list of names — Show me lists the records behind each suggestion, checked again at that moment and limited to records you are allowed to see. Members see suggestions about their own records; admins also see the team's.
• Ordered by what matters to you — Suggestions are ranked by severity, the money or number at stake, urgency and whether they move a goal that is behind. What has worked in your own workspace before nudges the order. Other customers' data is never used.

Capabilities at a glance

• A short ranked list on Home, refreshed each morning at 5am in your workspace's timezone
• Deals gone quiet: no contact and no stage move for 21 days or more
• New leads from forms or chat that nobody has emailed, called or set a task for within 48 hours
• Goals that are behind pace
• Past customers gone quiet: won 180 or more days ago, no contact for 120 days or more, and still able to be emailed
• Meetings nobody followed up: ended in the last fortnight with no email, call, task or later meeting since, sent to the host
• Deals past their close date: still open, with a close date (any date field named for closing) that has passed
• People who clicked a campaign email this week and nobody has contacted since
• Goals set in plain words, shown with exactly how they will be measured, saved only when you press Set goal, and editable under Reporting ▸ Goals like any other goal
• Do it: a plan of tasks (one per record, up to 25) for your own quiet deals, new leads, meetings, overdue deals and clicks, approved on a card before anything changes; closing the card without building hands the suggestion straight back
• Other suggestions hand off to the assistant, which drafts emails or plans a catch-up but never sends, schedules or publishes without asking
• Show me: the records behind every number, checked again live and filtered to records you can see
• Snooze or dismiss; a dismissed suggestion only comes back if it gets clearly worse (its severity rises or its count grows by half)
• Members see suggestions about their own records, admins also see team-level ones, and suggestions about a module you can't open are hidden
• A one-line reason for each suggestion in plain English, written by AI and checked so it can't invent numbers
• Recently done: what happened afterwards to the records it touched, such as replies, meetings and wins, with a 30-day summary of what acting on suggestions led to
• The assistant reads each candidate in context before it is counted: the meeting and booking answers, the last ten notes, emails and calls on the record and the records linked to it, the deal's stage and value, and whether they are already a customer
• The assistant's picks under each suggestion: the record, the specific next step, why, the ones for today first, and a draft email where an email is the right move, sent only when you press Send
• Set aside with a reason, never silently: Show me lists what the assistant set aside and why, and any of it can be brought back
• Meetings, leads and clicks from your own team (a member's address or one of your sending domains) are never suggested
• Not needed in one click, or in your own words, which the assistant remembers and uses next time; stop suggesting a booking page, a company or a record entirely, and undo it in Settings
• What counts as gone quiet is learned per pipeline from the deals your team actually won (the default 21 days until there are enough wins to learn from)
• Tasks built by Do it carry the assistant's next step as the title, with the reasoning and the draft in the notes
• Reading in context can be switched off per workspace; suggestions then come from the rules alone. Each read uses one AI credit, at most 30 new reads a day
• Off by default; an admin turns it on per workspace in Settings ▸ AI assistant ▸ What's Next
• Live audits: broken or redirecting links, ended offers and products that can't be bought in live flows and scheduled campaigns, and flow emails that underperform, each opening the exact step
• Hold back part of a suggestion's work and see what it caused; only a holdout says 'because of'