Sees your structure
Tasks, statuses, priorities, time per row, deadlines, table logs, comments, notes. The AI walks in already knowing what's done, what's risky, and where your hours actually went. No briefing, no setup.
14 featuresBuilt around your real work
AI built around your real work- not bolted onto a chatbot.
SelfManager's AI reads your tables, rows, statuses, priorities, time tracking, deadlines, table logs, comments, and notes - on the days they happen - not just the words you type into a prompt. That's the difference between an answer that quotes your actual work and an answer that has to guess what you meant.
14features4surfaces2models · Fast & Thinking0manual context
Output goes into your tables
Plans become dated tables. Summaries come back as copy-ready text. Reviews cite real tasks and real time spent. There's no manual translation step between an AI sandbox and your work.
Nothing saves without you
AI output is yours to review first. Plans land in an editable preview before you commit. Generated tasks can be removed any time. You stay in control of what enters your data.
Fast or Thinking, your call
Two modes per request. Fast for quick wraps. Thinking for harder questions like "what blocked me this month?" Currently powered by Google's Gemini 3.1 Pro and Gemini 3 Flash.
Soon: more model selectors, plus an AI chat/voice assistant called Jarvis who can do anything the UI can do, just by talking to him. He's in the plan.
Jump to any feature below.
- 01Voice to Text
- 02AI Plan
- 03AI Review
- 04Team AI Review
- 05Instant AI Review
- 06Follow up on a period summary
- 07Turn any text into a to-do list
- 08Chat with AI about a table
- 09Summarize any table
- 10Follow up on a table summary
- 11Chat with pinned tables
- 12Summarize pinned tables
- 13Follow up on pinned summaries
- 14Listen to any AI response
01
Voice to Text - Speak Instead of Typing
The fastest way to get something out of your head and into SelfManager.
Tap the microphone anywhere you would normally type and just talk. AI transcribes what you said and drops the text straight into the box, on its own line, ready to edit. It works in table comments, in table notes, and in every AI prompt box - AI Plan, AI Review, table chat, pinned-tables chat, and task generation.
Where it makes sense to keep the audio, SelfManager asks. After a comment or a note is transcribed you choose Keep or Delete audio - the transcript stays either way. Kept recordings get their own player and show up on the new All Recordings page, newest first, each with its transcript and a link back to the table it came from.
In AI prompt boxes nothing is stored at all. The audio is transcribed and discarded, because a prompt is something you send, not something you keep.
The cost, plainly: 1 AI credit for a recording up to 5 minutes, 2 credits for anything longer. Ten minutes is the longest single take, and the time remaining is always on screen while you record.
Great for:
- Capturing a thought while walking, driving, or away from a keyboard
- Long comments you would never type out in full
- End-of-day notes, spoken faster than you can write them
- Briefing AI Plan or AI Review out loud instead of composing a prompt
- Keeping your own voice on a decision, so the reasoning survives the summary
Notes:
- Up to 10 minutes per recording, with the remaining time always on screen
- 1 AI credit up to 5 minutes, 2 credits over 5 minutes
- Works in Chrome, Firefox and Safari - recordings are converted so they play back anywhere
- Delete any saved recording at any time; the transcript stays where you put it
02
AI Plan - Generate a Plan for Any Day, Week, or Month
The feature in SelfManager.ai that looks forward.
Pick a period - a single day, a week, a custom range up to 31 days, or a full month - describe what you want to achieve in plain English, and AI Plan returns a fully structured plan: one table per day, each with up to 20 prioritised tasks, ready to edit or ship as-is.
Optionally feed in up to three months of your past work as context, and the AI shapes the plan around your actual cadence - your recurring habits, your active projects, the days you usually rest.
Every plan opens in a sliding review pane: one editable card per day. Rename tasks, change priorities or statuses, delete or add rows, skip whole days you don't want to commit. When you're happy, hit Approve and SelfManager creates the tables on the right dates, then drops you on day one ready to start.
Great for:
- Side-project sprints with daily deep-work blocks
- Workout schedules with proper rest days
- Daily habits that recur across a week or month
- Two-week launch plans for a project or campaign
- Studying, exam prep, or learning a new tool
- Turning a vague goal ("read more", "ship the redesign") into concrete, dated tasks
Example briefs:
- "Two-week launch plan for the new pricing page, ~4h/day, weekdays only. Day 1 is research, day 14 is GA."
- "Plan a strength + cardio split for the next month. Alternate days, one full rest day per week, lighter sessions on Sundays."
- "Build a daily reading and journaling habit. 30 minutes reading, 10 minutes journaling, every weekday morning."
Notes:
- Capped at 31 days per generation
- From 1 credit on the Fast model, from 5 on Thinking - pick the model on the input pill. Price rises with how much data the plan carries, and you see the exact number before it runs
- Tables created from a plan have time-tracking off and logs on by default - flip either later from Edit Table
03
AI Review - Chat About Any Week, Month or Quarter
Pick a time window. Ask anything about how it actually went.
This week. Last month. Q2. A custom range you define. The AI reads every table you touched in that window - tasks, statuses, time, comments, logs - and lets you ask anything about how the period really played out.
This is your retrospective surface. The data is in your tables; you just have to ask the right question.
Hero example:"I felt slow last week. Is that real, or am I imagining it?" → the AI compares your last week's completion + time-tracked output to the prior 4 weeks. You either get validation or a gentle reality check, both useful.
- "What patterns do I see in my Mondays this month?" - habit / cadence read
- "What % of my time was on client work vs internal?" - tag-free time accounting
- "Was my high-priority work actually high-priority? Or did I default to easy tasks?" - calibration check
- "Which week of the month am I most productive?" - cadence pattern
- "What recurring task did I keep snoozing? Why?" - avoidance signal
You can optionally include table comments for even more context.
04
Review your team’s shared tables
The same review, scoped to the work your team can see.
Choose My team in AI Review to focus the analysis on tables you’ve shared with your team. Select a week, month, quarter, or custom date range, then let AI summarize progress, priorities, discussions, and potential blockers.
Not a scorecard: the review is organised by table and by theme - what moved, which themes emerged, where blockers were raised. Deliberately not a per-person breakdown, a ranking, or a count of tasks per head. It is a status digest your team could read together.
- Share one or more tables with your team.
- Open AI Review and select My team.
- Choose the period you want to review.
- Include comments if you want insights from team discussions.
- Generate the review and continue with follow-up questions.
Completed tasks record who finished them, so the review can name who unblocked what rather than only that something moved. Tasks completed before this existed carry no name, and the AI is told never to guess one. Available on the Team plan.
05
Instant AI Review of Any Week, Month or Quarter
No questions in mind? Click Review and get the report.
Pick the window - this or last week, a month, a quarter, or a custom range of up to 90 days - and the AI reads all of it, then produces a structured report: what shipped, what slipped, where time went, what to focus on next. The closest thing to having a manager who actually pays attention.
Hero example: Friday afternoon. Hit Review for the week. Two minutes of reading replaces twenty minutes of trying to remember.
What the review surfaces:
- A clear recap of what happened
- Main completed items
- Ongoing or overdue tasks
- Suggested focus areas for the next period
Great for:
- Weekly self-review - without the awkwardness of writing one for yourself
- Monthly retro for solo operators - the manager-style debrief you don't have a manager for
- Quarterly performance check - pick this quarter, the previous one, or any quarter you choose, and walk into a raise conversation with structured evidence
- Project-end debrief - set a custom range matching the dates the project actually ran, up to 90 days
- Habit accountability - monthly review of whether the routine you committed to actually happened
06
Continue the Conversation After an AI Review
The AI review is loaded. The data's there. Now ask anything.
Reframe the review for an audience, drill into a specific date inside the window, or turn the insights into the input for next week's plan - all without re-explaining what the AI already knows.
Hero example: Monthly review says "You spent 47% of your time on client work this month." Follow-up: "Generate an AI Plan for next month that protects 50% for internal work." The handoff goes straight to /plan, with the period data baked in.
Best follow-up moves after a period review:
- "Turn this into a journal entry" - first-person voice for personal notes
- "Write an invoice summary for [client] based on time spent" - billing draft
- "Compare this month to last month" - longitudinal lens
- "Generate an AI Plan based on what's still open" - direct handoff to /plan
- "Pull the highlights for my LinkedIn weekly recap" - external-content draft
- "What's the one habit I should change next month based on this?"
- "Surface only the work I'd brag about in an interview"
- "Find the day I was most productive - what was different about it?"
- "Highlight only the tasks related to [client/project name]"
- "Create a bullet list of goals for next week based on this review"
AI stays inside the context of your chosen period and your own data, so insights become action.
07
Turn Any Text into a To-Do List
Paste it, describe it, or chat it out. The AI hands you a clean, prioritised table - ready to ship.
A Zoom transcript, a long client email, a brain-dump after a planning call, a Slack thread you barely remember scrolling through. The AI reads it, pulls out the actual action items, and lands them as priority-tagged rows inside the table you're standing in.
When the brief is hazy, open the optional chat. Type a rough idea and the AI asks the clarifying questions a sharp colleague would ask before rolling up their sleeves - scope, priorities, deadlines, dependencies. Refine in the back-and-forth, then hit Generate from this chat. The conversation is the brief. No second prompt, no double-typing.
The structure matters. The output IS the table, with inferred priorities and clean phrasing. You skim, edit anything that needs nudging, and you're moving.
Three modes, one toggle:
- Auto (default) → the AI reads your input. List? Preserve it verbatim. Brain dump? Expand it into a plan.
- Structure as-is → you've already written the tasks. We split them into rows, infer priority, and keep your wording 1:1. Zero rewriting.
- Plan from idea → you've got a goal, not a list. We expand it into 4-10 concrete, action-oriented tasks.
Hero example: Paste a client email that says "hey can you also pull the Q3 numbers, oh and the deck still needs the new branding, plus we should probably talk about pricing soon" → 3 tasks: pull Q3 numbers (high), update deck branding (medium), schedule pricing conversation (low). Done.
More patterns:
- Voicemail or voice-note transcripts → a phone-call to-do list with the social pleasantries stripped
- An RFC, spec, or design doc → an implementation checklist
- Meeting notes from a 45-minute call → only the decisions, not the "we'll figure this out later" lines
- Documentation or SOPs → a step-by-step you can actually follow
- A WhatsApp screenshot OCR'd into text → today's grocery list, priorities and all
When the chat earns its keep:
- "Plan today as a 32-year-old male, balanced career and fitness." - the AI asks about deadlines, gym days, and goals before generating.
- "Help me break down a launch checklist for next Friday." - the AI clarifies scope, audience, and dependencies first.
- "I have a vague idea for a side project." - the AI pulls the goal out of you in 2-3 turns, then drafts the table.
You stay in control: edit, reorder, and re-prioritise tasks like any other table in SelfManager.
08
Chat with AI About a Specific Table
Open any table. The AI walks in already knowing every task, status, priority, comment, and minute logged.
No briefing, no setup. The AI sees every task, every status, every priority, every comment, every log entry, and the time you've spent on each row. You don't paste context. You ask the question.
This is the workhorse of the AI suite. Most days, this is the feature you'll touch most.
Hero example:"I've been on this table for three days. What am I actually avoiding?" → the AI pulls the rows that haven't moved, ranks them by priority, and tells you the pattern. Hard to lie to yourself when the data is right there.
More queries that work well:
- "What's a realistic ETA based on how I've been pacing?" - reads time-tracked data and projects forward
- "Write a 4-line client update from this table" - drafts a status note grounded in real completion percentages
- "What did I learn from this project that I should take to the next one?" - reads comments + logs for retrospective signal
- "Which tasks are high priority and still not started?" - quick triage
- "Help me break the biggest open task into 5 smaller ones" - planning help, in-context
What the AI has access to in this chat:
- Overall completion progress of the table
- Status and progress of each task
- Priority of each task
- Time tracking per task (if enabled)
- Notes added to tasks
- Comments (if you tick the Include comments checkbox)
- Table logs - edits, status changes, history (if Track table logs is enabled)
09
Summarize Any Table with AI
Skip the chat. Get a one-glance recap of where the table actually stands.
Hit Summarize. You get what's done, what's open, where the priority weight is, what's stuck. Built for the 30 seconds between two meetings when you need to know where you stand without having to think about it.
Hero example: Open the project table. Hit Summarize. Read the "Risks" line aloud at standup three minutes later.
What you get:
- A quick recap of what's done vs not done
- Highlights of high-priority items still open
- A short "status update" you can copy into email, Slack, or a meeting
- A list of possible next steps or suggestions
Great for:
- Pre-standup brief - a 30-second read before your daily so you're not winging it
- Handoff to a teammate - paste the summary into Slack when you're tagging someone in for the day
- The two-week-old table you forgot existed - three sentences of "this is what you were even doing"
- Monday morning re-entry - come back from a weekend and remember which tables matter
- Pre-meeting prep - open the relevant table, summarise, walk in informed
10
Follow Up on Any Table Summary
The summary lands. The conversation doesn't end there.
Ask the AI to reshape the recap for whoever needs to read it next, drill into a specific risk, or pivot from "what's the state" to "what should I do tomorrow." It already has the table loaded - no re-explaining.
Hero example: Summary says "3 high-priority tasks blocked, waiting on client feedback." Follow-up: "Draft a polite nudge email I can send to unstick them." You get a ready-to-send draft in 5 seconds.
Best follow-up moves on a single-table summary:
- "Make this client-friendly" - tone shift for external sharing
- "Highlight only the blocked items" - narrow the surface
- "Show me the rows that haven't moved in 7 days" - aging-task scan
- "Plan tomorrow based on this" - pivot from review to forward action
- "Write a one-line update I can paste into Slack" - shrink and ship
- "Rewrite this in second person - talking to me, not about me"
- "Turn this into a journal entry I can save in my notes"
- "What single decision would unblock the most rows?"
- "Create a checklist of just the open items, ordered by priority"
- "Flag the rows where I've spent the most time but the least progress"
You stay in the same context, so the AI doesn't "forget" what you were working on.
11
Chat with AI About Your Pinned Tables
Pinning is your signal that these projects actually matter right now.
Pinned-tables AI takes that signal seriously. It reads only the tables you've pinned - not your archive, not your half-started drafts - and treats them as your real active workload.
Use it when the question crosses projects: capacity, priorities, portfolio health, where you're falling behind.
Hero example:"Am I overcommitted? What can I pause without breaking anything?" → the AI reads activity, priority weight, and recent progress across every pinned project, and tells you which one you've been ignoring (and whether that's actually fine).
- "Which pinned table has the highest ratio of high-priority tasks open?" - where to focus today
- "Looking at the next 2 weeks, where am I going to run out of time?" - capacity projection
- "Compare velocity across my pinned projects this month" - pace check
- "Which project has the most blocked work?" - surface the bottleneck
- "Summarise the state of all my client-facing projects so I can send a Friday email" - outbound report draft
The AI can see across every pinned table:
- All pinned tables and their tasks
- Progress, priorities, time tracking, notes
- Comments (if you choose to include them)
12
Summarize Your Pinned Tables with AI
One click, one paragraph. Across every project that matters right now.
What's healthy, what's slipping, what needs eyes today. Built for Monday-morning coffee, not Wednesday triage.
Hero example: Monday, 8:47 AM. Hit Summarize on Pinned. Read it. Plan your week from it. Done before your first meeting.
You'll get:
- A global overview of your most important work
- Highlights of risky, overdue, or blocked items
- Suggestions on where to put your focus
- A ready-to-send status update about your "big picture"
Great for:
- Coming back from vacation - catch up on your own life in 60 seconds
- Friday wrap - confirm what shipped vs what slipped before you log off
- Pre-1:1 with a manager - walk in with a clear story across all your work
- Quarterly self-check - scan the cross-project state before quarterly planning
- End of a busy push - regain the bigger picture after a head-down sprint
13
Follow Up on Pinned Table Summaries
The summary is the starting point. Follow-ups dig in - across all your pinned projects.
Reshape the recap for different audiences, drill into a specific project, or pivot from review to forward planning - without losing the cross-project context that pinned-AI gives you.
Hero example: Summary mentions "Project Atlas is at risk." Follow-up: "Drill into just Atlas - what's actually slowing it down?" The AI zooms into that one pinned table without losing the portfolio view it had a second ago.
Best follow-up moves after a pinned-tables summary:
- "Make this a manager-ready update" - reframe for upward reporting
- "Focus on just the [project name] line" - drill into one pinned table
- "Across all of these, where am I lowest-priority?" - cross-project sort
- "Plan my Tuesday based on what's open across all pinned projects" - forward pivot with portfolio context
- "Compare this week to last week" - longitudinal read
- "What's the one thing that, if I shipped it tomorrow, would have the biggest cross-project impact?"
- "Group the open items by client, with totals"
- "Pull just the items I've been the bottleneck on"
- "Turn this into a weekly review I can paste into my journal"
- "Highlight only the work that's both high-priority AND not-started"
The conversation stays anchored in your pinned data, so you can explore your work from multiple angles without re-explaining anything.
14
Listen to Any AI Response
Press the speaker and get on with something else.
Every AI response can read itself to you - AI Review, AI Plan, table chats, pinned-tables chats, and saved conversations. Useful when the answer is long and you would rather walk, drive, or cook than sit and read it.
It is a real player, not a play button: speed up to 2×, jump back or forward 15 seconds, scrub anywhere, and download the audio.
The cost, plainly: 1 AI credit per 2 minutes of audio, always rounded up. Two minutes is 1 credit, five minutes is 3, ten minutes is 5. You see the exact number before you press anything, and replaying audio you already generated is free.
- Keeping the audio is optional - press save and it joins your recordings, skip it and nothing is stored
- Saved audio appears in All Recordings alongside your voice notes, and in your data export
- You are charged for the real length, never more than the number you were shown
- Come back to a saved response later and it plays again for free
Nothing is saved automatically. Most responses are listened to once, so the audio lives in the page and disappears when you leave unless you say otherwise.
See it on your own work.
Every AI feature above runs on your real data the moment you sign in. The trial gives you a full week to find the workflow that fits.
All 14 AI features are available on every table, across your pinned tables, and on the AI Review page.