Private beta · for consultants, fractional executives and owner-operators

The planner that learns how long your work actually takes.

Tell it what you've promised. It plans each day around your meetings, says why each piece of work is where it is, and tells you what to do next — and every finished task teaches it how long your work really takes.

  • Never edits your calendar events
  • The AI never moves your schedule
  • No ads, no analytics scripts

Five steps on one example day — the screens and sentences are the product's own.

How it works

Five steps, the same every day

You decide what matters. The engine does the arithmetic — and shows its working.

  1. 1Capture

    Say it in one line

    Type it, or say it, the way you would to a colleague: “Draft the proposal by Friday 5pm, about an hour, in one sitting.” A model reads the line into a draft — what, how long, by when — and it waits in Inbox until you decide: Do, Defer, Delegate or Drop. Nothing becomes a commitment without you, and no field is ever required.

  2. 2Plan

    It finds the time — and says why

    Press Do, and a deterministic engine — not a model — places the work in the free time your meetings, lunch and working hours leave. Beside every sitting the margin says why it is where it is: the meeting it follows, how its length was sized, the stretch that was too short for it.

  3. 3Adapt

    When the day changes, it moves the minimum

    A meeting lands on a planned hour, and the plan moves the minimum — here, just the sitting it displaced — and says what it did and why.

  4. 4Focus

    Know what to do right now

    Now picks the best use of the time you actually have and shows its reasons. One click starts the timer — and with the Windows companion it floats over your work until you stop or finish.

    Draft the proposal24:13
  5. 5Learn

    It learns how long your work takes

    You guessed the proposal at 1h; the plan sized it at 1h 32m from your record of deep work; it took 1h 22m. The timer, the tracker and what you mark done put every finished sitting on that record, and the measure for its kind of work moves with it — so the next plan sizes your estimates by what your work really takes.

Why it's different

A plan that shows its working

Some planners pack your day full; others leave the planning to you. Intenday plans it for you — and never asks you to take its word for anything.

intenday.com/today
TodayTuesday, October 27
80%of today's envelope

Demand 2h 55m · envelope 3h 40m left today (4h 54m free minus the 25% reserve, the default)

Working hours · meetings from 1 calendar subtracted.

Last plan at 9:05 AM — 1 sitting placed, 6 kept.

The day5
  1. 9:00
    Reply to the tender thread
    #3

    20m left · until 9:20 AM

    Placed at the first free time of the day — your hours begin at 9:00 AM. Placed to end before the 9:30 AM–10:00 AM meeting and its 10-minute padding.

  2. 9:30
    Stand-upNext · in 25m

    Calendar · 9:30 – 10:00 AM · 30 min

  3. 10:10
    Review the contract
    #1

    Deadline-criticalAt risk1h left · due tomorrow 5:00 PM (hard) · until 11:10 AM

    Placed at the first free time after the 9:30 AM–10:00 AM meeting and its 10-minute padding. The stretch at 9:00 AM was too short for 25 unbroken minutes.

  4. 11:1050m free
  5. 12:00Lunch · kept free
  6. 1:00
    Client call

    Calendar · 1:00 – 2:00 PM · 60 min

  7. 2:10
    Draft the proposal
    #2

    1h left · due Fri 30 5:00 PM · until 3:45 PM

    Its work was sized at 1h 32m when placed: 1h of your estimate × deep work's ×1.52 (7 samples, 30% starting assumption). Placed at the first free time after the 1:00 PM–2:00 PM meeting and its 10-minute padding. The stretch at 11:10 AM was too short for 95 unbroken minutes.

  8. 3:451h 15m free
An example day. The sentences are the product's own; the day is made up.
  • Every sitting says why

    Not “the AI decided”. Each sitting stores the factors it was placed from and says them back beside it — down to why #3 sits before #1 on the day here: the stretch at 9:00 AM was too short for the contract.

  • It never pretends a day fits

    An overcommitted day reads 113% — the number is never capped at 100. Work that does not fit is listed with the reason it found no place, never dropped and never squeezed in, and assumed hours say “assumed” until you confirm them.

    An overcommitted day, as TODAY draws one — an illustration, not the example Tuesday.
  • Plans hold still

    A sitting in the next ninety minutes never reshuffles without a real conflict, and a repair moves the minimum — so the hour you planned around is still there when you get to it.

  • AI reads. Code decides.

    Models read messy text and write words. Placement, priority and repair are plain deterministic code that gives the same plan from the same day, every time.

  • A two-minute close

    Each evening the day is judged against the plan it started with, in one line; then its open loops are answered and the next working day is set up. No streaks, no scores.

The record

Every week it knows your work better — here is the founder's own

The one chart this claim rests on, live from the account Intenday was built in. Each week: what the work took over what was sized for it — the first guess, and the plan's sizing of it with what the learning knew when the week began. Aggregate ratios only; no task is named.

×0.5×1×2×331 Aug · took ×1.46 the founder's first guess, over 13 measured tasks7 Sep · took ×2.35 the founder's first guess, over 15 measured tasks14 Sep · took ×1.56 the founder's first guess, over 14 measured tasks21 Sep · took ×1.18 the founder's first guess, over 7 measured tasks28 Sep · took ×1.36 the founder's first guess, over 6 measured tasks5 Oct · took ×2.17 the founder's first guess, over 5 measured tasks12 Oct · took ×1.17 the founder's first guess, over 5 measured tasks19 Oct · took ×1.67 the founder's first guess, over 1 measured task (under three: no verdict)26 Oct · took ×1.24 the founder's first guess, over 7 measured tasks2 Nov · took ×1.21 the founder's first guess, over 3 measured tasks9 Nov · took ×0.81 the founder's first guess, over 6 measured tasks16 Nov so far · took ×1.00 the founder's first guess, over 1 measured task (under three: no verdict)31 Aug · took ×1.39 the plan's sizing, over 13 measured tasks7 Sep · took ×2.10 the plan's sizing, over 15 measured tasks14 Sep · took ×1.44 the plan's sizing, over 14 measured tasks21 Sep · took ×1.05 the plan's sizing, over 7 measured tasks28 Sep · took ×1.11 the plan's sizing, over 6 measured tasks5 Oct · took ×1.90 the plan's sizing, over 5 measured tasks12 Oct · took ×1.08 the plan's sizing, over 5 measured tasks19 Oct · took ×1.03 the plan's sizing, over 1 measured task (under three: no verdict)26 Oct · took ×1.04 the plan's sizing, over 7 measured tasks2 Nov · took ×1.05 the plan's sizing, over 3 measured tasks9 Nov · took ×0.67 the plan's sizing, over 6 measured tasks16 Nov so far · took ×1.00 the plan's sizing, over 1 measured task (under three: no verdict)31 Aug5 Oct16 Nov so far
Plan accuracy over the founder's last 12 weeks: the first guess and the plan's sizing, each as what the work took over what was sized
WeekMeasured tasksTook over the founder's first guessTook over the plan's sizing
31 Aug13×1.46×1.39
7 Sep15×2.35×2.10
14 Sep14×1.56×1.44
21 Sep7×1.18×1.05
28 Sep6×1.36×1.11
5 Oct5×2.17×1.90
12 Oct5×1.17×1.08
19 Oct1 (under three: no verdict)×1.67×1.03
26 Oct7×1.24×1.04
2 Nov3×1.21×1.05
9 Nov6×0.81×0.67
16 Nov so far1 (under three: no verdict)×1.00×1.00

the plan's sizing, with what it knew when the week beganthe founder's first guess×1 is what the work took; faint is a week with under three measured tasks, which counts in no figure

Over 10 weeks with 3 or more measured tasks (81 tasks), the founder's first guesses were off by 47% of what the work took, and the plan's sizing by 45%. The plan's sizing was fewer minutes off than the first guess in 8 of 10 such weeks with a record to size by.

The latest week with 3 or more measured tasks was sized 67% from the record and 33% from the starting assumption, 17% of its guessed minutes being work with no kind, which is ×1 whatever the record. Regenerated about every fifteen minutes.

Privacy

Private by design

  • The tracker sees apps, not content

    Off until you switch it on, machine by machine. Then the desktop app records which application has focus and how long — never a window title, never anything you typed or read. Site names only if you switch them on too, and then the domain alone.

  • Your events are never modified

    It reads your calendar and writes only to a separate calendar it creates for you; there is a mode where no meeting title is ever stored.

  • Models never see your browsing

    The models read what you write and the notes you paste. Your browsing domains are never sent to any model at all.

  • Delete is not a request

    Settings has a delete button: everything the app stores goes at once, and the sign-in record is removed by hand, which the policy says plainly. No advertising trackers, no analytics scripts.

The whole of it, written from the data model rather than a template: the privacy policy.

Pricing

What it will cost

Nothing can be bought yet — the beta is closed, and the first invitations go to the waitlist. When it opens, it is one plan with everything in it.

One plan

  • A plan for every day, around your real calendar
  • A reason beside every sitting
  • Now — the best next thing, with a timer
  • Repairs that move the minimum
  • Sizing learned from your own record
  • The two-minute close and the weekly review
  • The Windows companion: capture hotkey, floating timer, tracker

$16a month, billed yearly

or $20 a month, month to month

Founding price: $12 a month, yearly

For the first 100 people — held for as long as you stay.

A 30-day trial, no card

Join the waitlist

The models cost real money per account, so every account has a spending ceiling, stated in the terms; the price is set so a heavy month stays inside that ceiling. Cancelling will be as plain as deleting.

Questions

What people ask first

Anything else goes to a person: kevin@simplifywithai.io. For how it works, in full: what this is.

Who is it for?

Independent consultants, fractional executives and owner-operators — people with dates promised to clients and days cut up by other people's meetings.

Will it change my calendar?

No. It reads your calendar and writes only to a separate “Intenday” calendar it creates for you; your own events are never modified. Connecting is optional, and there is a mode where no meeting title is ever stored — only that you are busy.

What does the AI actually do?

It reads and writes words. A messy line becomes a structured draft; pasted meeting notes become candidate commitments, quoted from the line they came from; the note you leave at the close is read for commitments; and the morning briefing and the weekly reflection are written from the record. It never moves your schedule — placement, priority and repair are deterministic code.

What happens to work that doesn't fit?

It is listed with the reason it found no place — never dropped, never squeezed in — and the day shows its real load, past 100% when that is the truth. “Do it anyway” pins it where you say.

How does it learn how long things take?

From the record: the timer, the tracker spans you answer, and what you mark done. Each kind of work gets its own measure, and the plan says when it used one — “1h of your estimate × deep work's ×1.52 (7 samples, 30% starting assumption)”. Your estimate stays as you gave it.

Do I need the desktop app?

No. The web app is the product and runs in your browser. The optional Windows companion adds a capture hotkey, the floating timer and an opt-in tracker; a macOS companion is planned.

Can I delete my data?

Yes. Settings has a delete button, and it is not a request: everything the app stores goes at once. The sign-in record is removed by hand, which the privacy policy says plainly.

When can I get in?

The beta is closed. Invitations go out a few at a time, by hand, each with a twenty-minute call to set up your hours and your first plan. Join the waitlist.

Get in

Get your first plan

Invitations go out a few at a time, by hand, with a twenty-minute call to set up your hours and your first plan. Leave an address and nothing else.

Only the address is kept, until you have an account or ask for it to go. No newsletter.

Already invited? Sign in.

The web app is the product. The optional Windows companion — Intenday 0.5.0 for Windows — adds the tracker, the capture hotkey and the floating timer. It is offered in Settings; Windows warns that the publisher is unknown until the installer is signed.