Your back-and-forth
becomes items

You clearly listed the three things you needed and even got the reply “Okay, will do.” Right after that, the talk turned to the quote and the lunch, and six days later only two came in.

How many steps does it take to hand off one task?

These steps vary a little from team to team. On the TODAYS side, your written note and attached files are split into items by By when · What · How, and the sender checks them once before sending. Dates that aren't in the files are left blank, not made up. The schedule is created once you've agreed on it together.

In the request window, write your request as you usually would, or paste in what you asked, and check the recipient. When you tap [Next], Haru, the AI assistant, reads it and splits it into items, and the sender checks them first before sending. Until it's sent, only the sender can see it.

Each item gets a title and separate fields for By when · What · How, plus completion criteria if the request includes them. Each line gets a source showing where its wording came from. Dates and completion criteria that aren't in the email are left blank, not made up.

Alex gets a “You received a new request” notification. The sender and the recipient see the same document.

Tap any line and Ask Haru and Ask directly appear.

Organizing uses Haru Energy. A gray line at the bottom of each item shows that it was organized by AI.

After-hours notification blocking is on from the start. A request sent outside working hours arrives right away, and its notification goes out when the next working hours begin, regardless of this switch.

Haru looks for answers in the request text and attached files. Only the person asking sees their chat with Haru.

If the answer isn't in the document, Haru asks, “Ask the other person?” Tap [Ask], and only that one-sentence question (plus any file you attached yourself) goes to the requester, with no need to contact them separately. You can edit the sentence before sending, and Haru's answer and the surrounding conversation aren't passed on.

The question is shown on the line you tapped, so it's clear right away what it's about. People in this project see it too, and the organization owner can also see it.

When the requester writes an answer and taps [Reply and update items], Haru uses that answer to update the document right away and shows it. Until it's sent, only the requester sees the changes and can edit or undo them. Tap [Send] to deliver it as a new version, or use [Post reply only] to just answer the question. It works the same way when the requester tells Haru about something that changed partway through.

Once applied, it becomes a new version, and changed lines get “Changed in v5” and a source. If there are two or more sources, tap “2 sources · Flow” to see how it changed, in date order. The same goes while the schedule is underway. New versions, questions, and answers made outside working hours become visible to the other person when the next working hours begin, and the notification goes out then too.

Asking Haru uses the asker's Haru Energy. Even with no Haru Energy left, you can ask directly without Haru. Having Haru update the document from an answer uses the requester's Haru Energy, and the requester can also just post a reply without Haru.

What you see first on the dashboard

1

• 1

Real screen

1 Overdue work and work in progress show here

Work whose recorded progress is behind the time elapsed from its start date (or creation date, if there's none) to its deadline gets a Caution label. It's an estimate based on dates, so it may differ from what's actually happening.

Overdue work is grouped separately in the red section at the top.

When all the to-dos are done, you get a notification: “Your completed plan was filed into the library.”

How much time are you spending on this now?

Files you receive pile up separately, wherever they came in. To find one, you first have to remember where you got it and what you wrote down at the time.

Then a week like this comes along. When a tutoring center consultation, a building maintenance fee transfer, a dentist appointment, a meeting with a business partner and an order deadline all land within three days, taking care of one pushes another back. You only notice what slipped in the evening.

Like the first week of a new school term, a week when an order deadline and a tax filing overlap, or the morning after you get back from vacation. In weeks like these, you miss what's scattered first.

In Onboarding near the bottom of the left menu, choose the Schedule tab, upload a photo, screenshot, PDF or ICS file, then press [Start extraction].

AI reads the text in the file and turns each line into one event. Lines the AI isn't sure about are marked “Needs review”, but please also give unmarked lines a quick check that the date, time and place came across correctly.

Once the events appear, AI groups similar ones on its own (to regroup, use [Group at once]). Press [Make a project] on each group, then press [Apply]. Press once more in the confirmation window, and they become to-dos on your calendar.

You don't need to create folders or come up with titles.

Reading events from photos and PDFs uses your Haru Energy (E); ICS files don't.

Type in the Home box in your own words. You can search, as in “Find the things due this week” or “Where is last month's quote?”, or ask, as in “Find the kitchen work in last month's quote — how much was it?” With the box on Auto, the button turns into [Find with Nuri] when it sees you want to find something. Words like yesterday, last week and last month work too (for files, they count from the upload date). A short name of a document that starts with words like my, the or last, like “the quote” or “my receipts”, is searched too, but if you type only a few words, like “quote”, it goes to a chat with Haru; in that case press Auto at the bottom right and choose Search.

Nuri, the AI assistant, searches your projects, to-dos and files. If you asked a question, Nuri reads what it found and shows you an organized answer, and when you expand sources used, you see the files the answer used. If the supporting sentence is found word for word in the original, it's shown too.

Searching and answering from the text inside files and in photos works only for files that have finished analysis. If Nuri can't find an answer, it says so and shows related files. Please check important numbers against the original.

Click a result, and that file or to-do opens right away.

Asking Nuri to find something or answer a question uses Haru Energy (E).

In the input box under Haru picks on Home, write what you want to do in your own words. If you ask, as in “Make a plan from this: …”, or attach related files, the button turns into [Create with Haru] when the box is on Auto. You don't need to set a name or deadline separately. A short line on its own goes to a chat with Haru; in that case press Auto at the bottom right and choose Create.

Haru, the AI assistant, first shows what it got from your request under “What you asked” and “Summary,” then suggests what to create on a card below.

Press [Create plan] on the card, and a window opens filled with your written note and what Haru took from it. Leave it unchanged and press [Start AI analysis]; the window closes and the plan outline appears first. On a computer it opens right away at the bottom right; on a phone, press [Create project] at the bottom right and it opens full screen.

It first shows the final goal and the order of stages. If it looks good, press [Create as is]. This creates tasks that follow these stages, and the to-dos and dates for each stage are set during creation. If it can't follow the stages, it tells you after creating. For short requests where questions come first, the outline is only a provisional overview, so pressing the button re-analyzes what you wrote and splits the tasks anew. The plan name stays exactly as written in the window. If you want to change something, write it in Extra instructions. The button changes to [Revise with this]; press it, and the plan is reworked with what you wrote and shown to you once more. If you don't press it, no tasks or to-dos are created.

If there's more to ask, questions appear below the plan. You don't have to answer them. If a lot of information is missing, the title changes to “Just a few more things.” If you write answers, they're applied just like extra instructions when the plan is created.

While it's creating, it shows which step it's on: Structure, Tasks or Schedule. The more files there are, the longer creation takes. You can do other things in the meantime, and when it's done, the number of tasks and to-dos appears in the same spot.

Press [View result] to open the result screen. See the stages and items in Strategy doc, each to-do's due date and time needed in List view, and the schedule by date in Timeline view. Haru fills in each to-do's AI note, completion criteria and execution guide after everything is created (click a to-do to expand it). Fix anything with [Edit], then press [Start project]; the to-dos start and you go to the new project.

Haru Energy (E) is used when you press [Create with Haru], when the plan is shown and when it's created. Applying what you wrote means analyzing again, so it's used once more. If you close the window midway, what has already been used isn't returned.

The request was three things in one sentence. Right after that, the talk turned to the quote and the lunch.

What needed doing was only in the chat, so six days later you took care of only the two you remembered.

The notification says “You received a new request.” Tap it to open the same document the sender sees.

The sender checked the items once before sending. That's why each card shows “Organized by AI · Sender reviewed” at the bottom.

Each item has separate fields for By when · What · How, plus completion criteria if the request includes them. If it doesn't, the field is left empty with “The request has no completion criteria.”

Tap “Original” or “Written note” next to a line, and the place that line came from opens right below it.

Lines marked with a dotted line can be tapped to ask about them. Continued in chapter 04.

Anything the sender adds along the way comes as a new version. Added lines show “Added in v2” and the source “Follow-up · Oct 8.”

[I'll do it as written] means you'll do it as described in what you're looking at now. When you tap it, the sender gets a notification: “The assignee reviewed the items.”

When the sender confirms too, it's agreed, and you get a notification: “Your schedule was created.” If the sender asks for changes, you get “Changes were requested,” and you go back to the step of checking the items.

The Schedule tab has only to-dos and dates. Each to-do shows its item number. Deadlines are changed by the sender. If one needs to change, tap the “By when” line and say so directly. Haru, the AI assistant, builds the schedule and won't place to-dos on weekends, public holidays, or outside working hours on its own.

If anything changes even after it's agreed, it becomes a new version, and changed lines show “Changed in vN.” If you see a changed line differently, say so on that line. If you say nothing, work follows that version.

Tap a line and [Ask Haru] and [Ask directly] appear. Dates, deliverables, each line of the method, and even completion criteria can all be tapped.

Haru looks for answers in the request text and attached files, and when it finds one, it also shows which file it came from. Only the person asking sees their chat with Haru.

If the answer isn't in the document, Haru asks, “Ask the other person?” Tap [Ask], and only that one-sentence question (plus any file you attached yourself) goes to Sam, with no need to contact Sam separately. You can edit the sentence before sending, and Haru's answer and the surrounding conversation aren't passed on.

The question is shown on the line you tapped, so Sam can tell right away what it's about. People in this project see it too, and the organization owner can also see it.

Questions and answers sent outside working hours become visible to the other person in the next working hours. Continued in chapter 05.

If an answer changes the document, it becomes a new version, and changed lines get “Changed in v5” and a source. If there are two or more sources, tap “2 sources · Flow” to see them in date order.

Asking Haru uses the asker's Haru Energy. Even with no Haru Energy left, you can ask directly without Haru.

Working hours follow your team's settings. If none are set, they're 9 AM to 6 PM on weekdays. Weekends and public holidays are skipped.

On the poster's own screen, it shows right away, marked “Shown at the next working hours.”

A request you're receiving for the first time shows up right away even if it arrives at night, and its notification goes out in the next working hours.

If you delete a question you wrote, it disappears right away. Versions already applied and answers written by others remain.

Other team notifications don't come after work or on weekends when blocking is on, and they aren't sent later either. The work itself stays on your screen.

After-hours notification blocking is on from the start, and the organization owner or an operator can turn it off. The timing of changes to a request document (when the other person sees them and when their notifications go out) follows the team's working hours, separately from this switch.

It's not that there's nothing to write down. You don't write it down because the quote might look expensive, and you don't ask because it might look like you don't trust them.

The gaps both sides left blank come back later as charges. More than the amount, it's “I was never told that” that hurts both sides.

Sync is a type you choose when creating a new project. Choose your side (I am doing the work · I am commissioning the work) and enter a title, and it's created. Once you've sent the link, you can't change sides.

When the other party opens the link, they can write what they want, attach files and send it, with no account and nothing to install.

Auto-organize received messages is an option you turn on in the link card (it's off at first). If auto-organize is turned on, AI organizes the messages the other party sent with consent to AI tidying and shows the result on their screen too.

If auto-organize is off (the starting setting), received messages pile up in their original form in “Received”, and they're organized into a work summary when you press [Organize received messages]. Messages that arrive after you send the work summary aren't organized by auto-organize or by [Organize received messages]; edit the lines yourself to reflect them.

If the other party doesn't turn on consent to AI tidying (optional), those messages and files aren't used for AI organizing and stay in “Received” as they were sent.

Press [Organize received messages] in “Received”, and AI makes a work summary from the messages the other party consented to AI tidying. It uses your Haru-Energy, and the organized result also appears on the other party's screen.

Where a sentence from the other party's message became a line as written, it's marked “Written”. Lines not decided yet and example values you set count as △ (not decided yet). Tap a line to edit it right there.

Press [Send to the client], and a window opens for one look before sending. It shows how many lines go out as they are and how many example values go out. Sending alone doesn't confirm it. The version you send stays as it is; if something needs to change, edit the line and send it again with [Send changes].

On the link, the client can tap a line to leave a question. Write your answer under the question on that line and send it. Sending an answer doesn't change the work summary — edit the line itself to reflect it.

When the client presses [Confirm] and you press [Confirm] on that same version, it becomes an “agreed work summary” and the In progress step opens. To change it after that, you send the changes and both sides confirm again.

Each side's [Confirm] is kept as a record that they pressed it. It isn't an identity check or e-signature. TODAYS doesn't judge whether this record counts as a written contract or other document required by laws such as the Fair Transactions in Subcontracting Act or the Artists Welfare Act. If your deal needs such a document, prepare it separately as the relevant laws require.

In the In progress step, the “What” items of the agreed work summary become the items to deliver. After you send the work, press [I sent it] and leave a one-line note, and the mark, date and note show on the client's link. This mark doesn't transfer any file.

On the link, the client can press [Received and checked] for each item to receive. “Received” is a mark made on the link and doesn't mean inspection or handover is complete. TODAYS doesn't check whether anything was actually delivered.

If an item you marked with [I sent it] gets no received mark, you can send [Ask for a received mark]. The request shows at the top of the In progress section on the client's link, and a notification can go out only if the client has added this deal to their own TODAYS account. TODAYS won't contact them otherwise, so pass the link on again yourself if needed. When every item without a received mark has an [I sent it] note and 7 days have passed since your first request after the last [I sent it], you can wrap up with [Wrap up without a received mark] even without a received mark. While there's no open deal link, or while a changed work summary is waiting for confirmation, you can't ask or wrap up this way. [I sent it] notes left before it's confirmed again don't count, and requests made before you make a new link don't count. This wrap-up isn't a record that the client received anything, and it doesn't mark anything as “Received”. The client can still mark it received on the link afterwards.

When everything to receive is marked “Received”, it moves to the wrap-up step. You can note the date, amount and memo of a payment you received on your own screen (optional). It isn't shown on the client's link, and TODAYS doesn't calculate or check amounts. TODAYS doesn't issue payment requests or handle payments for you.

When you press [Wrap up the deal], the client's link becomes view-only until it expires. Wrapping up is a status shown on screen and doesn't decide inspection or payment. If needed, you can open it again with [Reopen].