Remote work tools: pick the category first, the brand second
Remote work needs six categories of tool, not six particular brands: a task board, a shared document store, a calendar, a channel, time tracking, and a safe way to receive access. The client picks the brand, and the free plans usually cover all six, as long as you know each ceiling before you promise anything.
- Lesson 3 of 6
- Beginner
- Free, no signup
The six categories of tool remote work needs
The client decides the brand in each cell. What stays fixed are the cells themselves, and someone who knows the category loses nothing when the brand changes.
None of these six replaces an agreement. A tool is only the place an agreement is written down, and deciding what is urgent stays a conversation with the client.
Last checked: Facts and tool names in this lesson are re-checked against their sources on this date.
Pick the category first, then the brand
Every tool you have heard of falls into one of six categories: the place where tasks sit with dates on them, the place where files and documents stay, a calendar, a channel to talk on, time tracking, and a way to receive access without a password crossing a chat window. Someone who knows the categories loses nothing when the brand changes; someone who knows one brand starts again with every new client.
The most important rule in this lesson is not technical at all: the client tool wins. If the work lives in one person WhatsApp today, that is where you work, and only then do you slowly propose one single place for anything that carries a date. An assistant who forces a migration in week one has done the exact opposite of what they were selling, which was less work for the client.
When the client has no tools at all the situation flips and the choice is yours. Do not start with six there either. Two will do: one place for dated tasks and one place for files. Add the rest when a real problem asks for them, not because they were on a list.
One thing about free plans holds this whole lesson together. For an assistant with one or two clients the free plans are almost always enough, but their ceilings sit in odd places. The cap is usually not on the number of features; it is on things like how many boards, how large a single file, how long the history is kept, and how many people. The numbers on this page were read off each vendor own pricing page on the date printed at the top, and they are what separates a promise you can keep from one you cannot.
A tool that reduces the client work, against one that only moves it
A tool that reduces work
- The client already knows it or learns it in five minutes
- You have read its free ceiling before starting
- Every task in it carries a date and an owner
- If you switch tomorrow, the content comes back out
A tool that only moves work around
- The client has to create an account and be trained
- You discover its free ceiling in the middle of the job
- The same information is kept in two places
- You are the only one with access to it
These tests make the choice simpler and do not replace a conversation. If the client is comfortable with their current tool, that is the best tool, even when you know a better one.
Boards and documents: where do I write so nothing gets lost?
A task board is the place where every task sits with a date and an owner on it. Whether that is Trello or Notion or a plain Google Sheet matters much less than there being exactly one of them. On the Trello free plan, their own pricing page says up to 10 collaborators per Workspace and up to 10 boards in that Workspace, with unlimited cards. For one assistant and a few clients that ceiling stays far away for a long time.
The real ceiling is somewhere else and few people see it coming: on that same free plan storage is unlimited but each individual file is capped at 10MB. So the client 30MB PDF catalogue simply will not attach to a card. The fix is not to argue with the tool: the file lives in the document service and the card holds a link to it. That one habit keeps the board light and searchable.
For documents and files, what people in Iran actually use is a personal Google account. Google own help says a personal account has up to 15 GB of storage and that this space is shared among Drive, Gmail and Photos. Read that sentence twice, because the trap is right there: a full mailbox also stops file uploads. An assistant who suddenly cannot upload a client file usually assumes the problem is Drive.
And one simple rule so that nothing gets lost: anything that is a decision lands in the document, not in the chat. Chat is for coordinating; the document is for the thing that has to be found three weeks later.
Calendar and channel: which words have to survive?
A calendar does two jobs and only one of them is visible. The obvious one is holding appointments. The more important one is showing the hours you have blocked for work, because otherwise a client assumes every empty hour belongs to them.
If your work involves booking time with people outside the company, a scheduling tool buys back a lot of time: one link instead of five messages back and forth. The free ceiling here sits in a specific place too. The Calendly pricing page says the free plan gives one event type and one calendar connection. So do not waste that one. Make it the meeting that actually repeats, such as the weekly client check-in, rather than three half-used types.
About the channel, one thing has to be said plainly: chat is not an archive. The Slack pricing page, for a tool built specifically for work conversation, says its free plan keeps 90 days of searchable message history and that you upgrade to reach the previous 12 months. Now consider what a tool not built for this does. The practical consequence for you is one sentence: anything that is a decision, an amount or a deadline lands in the document the same minute, and its link goes into the chat.
And one access note that concerns Iran, sourced from Google own page. The Google Workspace help says Google restricts access to some of its business services in certain countries and regions, naming Iran among them, and then adds that certain Google services might be available in those countries for personal use, but not for business or education use. So a personal account and a client paid domain are two separate questions, and if the client is on Workspace with their own domain, check with them before promising anything. About the other tools on this page we make no claim: we found nothing about Iran on their own pages, so our answer is unknown, neither yes nor no.
Do you track time for the client or for yourself?
Most people start tracking time when they are paid by the hour and a client wants an invoice. That is the smallest of its uses. The main one is for you: until you know how many minutes a repeated task really takes, you can neither price it properly nor see which task is quietly bankrupting you.
An example every assistant has lived: in everybody head, cleaning up an inbox takes twenty minutes. The first time you actually start a timer you usually see a different number. That number is what serves you later in pricing, and pricing is the subject of lesson six of this path.
The free ceiling is real here too. The Clockify pricing page says its free plan is zero dollars for up to 5 users and that tracking on it is unlimited. For one person working for a few clients, that means the heavy part is free and what costs money is professional reporting and invoicing rather than the timer itself.
And a note that matters just as much: tracking time is not the same as being monitored. If the work is priced per task and the client pays for a result, you do not owe them a report of every minute, and it is better to say so plainly at the start. A timer that is for you stays for you.
How do I receive a client password without it landing in a chat?
This is the sixth category and the only one where getting it wrong cannot be repaired with an apology. A password sent over WhatsApp stays on both phones, in both phone cloud backups, and on every device where that account is open, and you control none of them.
The right answer has two steps and the first one is not about passwords at all: ask whether the service allows a separate account with limited access. For WordPress, social accounts, shop tools and most modern services the answer is yes. With a separate account the client can close your access whenever they want without changing their own password, and you are no longer holding something worth more than the whole contract.
The second step is for when there genuinely is only one shared credential, such as an old service with a single login. A password vault is the right tool there, and against the common assumption, its free plan is built for exactly this situation. Bitwarden own help says the individual Premium plan does not include secure sharing and that sharing requires an organization, then says a Free organization lets you share data with one other user in up to two collections. That is precisely the client and assistant pair, at no cost.
And three rules to state in the first meeting, which end the argument later: no password crosses the chat, access closes when the work ends and reminding them is your job, and if a password arrives in the chat anyway, ask the client to change it after the work and put that in writing. If you want to know why these three matter so much, the lesson on what cybersecurity is explains the same subject from the attacker side.
Two doors, and neither of them is the chat
The client says they will send their password
Have them create a limited account
- only the part where the work happens
- the client closes it whenever they want
- their personal password is never in your hands
Use a shared password vault
- a free Bitwarden organization is for two people
- the password never sits in a chat or an email
- after the work, the share is removed
This chart is about handing over access, not about trust. The route is the same with a long-standing client, because the main risk is not bad intent; it is a lost phone and a cloud backup.
The fast path, with AI
Starting with a new client is usually a call with twenty questions, half of them answered inaccurately because nobody has ever described their own daily work out loud. The faster route is to arrive at that call with a draft map: take what already exists in the client tools, strip the identifiers, and have the model pull the recurring processes out of it. Then the call is five questions rather than twenty.
- Ask the client for read access to one place, not to everything: the task board or the file folder. The same limited account this lesson described.
- Take the raw material: the last forty task titles, or the folder tree two levels deep, or a week of message subject lines. Names, numbers, amounts and anything that identifies a person are removed before it leaves your machine.
- Give the recipe below along with that material. A frontier model does better here because the pass is inference rather than mechanics; our current pick stays up to date in the RGB AI reference.
- Treat the output as a draft, never as a finding. Turn its list of unknowns into the questions for the call, and get the client to confirm every process it named before you write it down as fact.
Copy-ready recipe
I am a virtual assistant and I have just taken over a client account. Below is raw material from their own tools. Names, numbers and amounts have already been removed.
Context: the business is a {type of business} and I am taking over {scope of work}.
Raw material (one item per line):
{task titles or folder names or message subjects}
What I want:
1) Find the tasks that repeat and give each one a process name. For each process, say how many times it repeats in this data.
2) For each process, say what starts it and what ends it.
3) Say which processes depend on each other, meaning one cannot start until another finishes.
4) Put everything that cannot be worked out from this data in a separate list headed "I do not know". Do not guess and do not fill the gaps.
5) Finish with five questions whose answers would remove the most remaining uncertainty.Before you trust the output: Everything that comes out is inference from titles, and a title lies more often than you expect: a task called "follow up" may be three different jobs. So treat no process as real without the client confirming it, and read the repeat counts as a reading of this sample rather than a fact about the business. More important than any of that, this material is client data: until the client has said which tool is allowed, you either strip the identifiers or you paste nothing at all.
AI in this kind of work
AI has the smallest share in this lesson, and saying so is more honest than inventing a use for it. Two places it genuinely works: reading a service limits and pricing page and telling you which ceiling hits your case first, and writing the automation rule or the spreadsheet formula that would otherwise cost you an afternoon. Choosing the tool itself is not one of the two, because a model opinion about which tool is better is a summary of marketing pages.
Tools that actually help
- Claude Paste the service pricing page and ask which ceiling hits first at your number of people and your volume. Iran is on neither of Anthropic two supported country lists; we read that on Anthropic own page rather than measuring it.
- Gemini It understands Persian itself, so a Persian task list needs no translation first. Google own page says the Gemini web app works in more than 70 languages and over 230 countries and territories, and Iran is not on that list.
- NotebookLM The closest thing to a searchable memory for a client scattered documents: you supply the sources yourself and it answers only from them. It runs on the same Google account and Iran is not on the list of available countries.
Where it backfires
A model will state a price and a free-tier ceiling in the same confident tone it uses for everything else, and that is precisely the one thing not to trust it for. Ceilings move, and every model training data carries a date. Every number on this page was read off that vendor own pricing page on the date printed at the top, and reading it once more before you rely on it is the right move. The second risk belongs to this lesson specifically: a client task list is client data. Whether a conversation is used to improve a model depends on the plan and on a setting; Anthropic own privacy page says that for its consumer products, chats are used when the user has chosen to allow it, when a conversation is flagged for safety review, or on explicit opt-in. Until the client has said which tool is allowed, the default answer is not to paste. If you want to see the same thing from the attacker side, the lesson on what cybersecurity is is the right place.
Sources: Anthropic: supported countries and regions Anthropic privacy: is my data used for model training Google: where you can use the Gemini web app
Where this advice stops
This lesson names the categories and gives one example for each; it is not a review, and we have not compared these tools against one another. The free ceilings written here are what each vendor own page said on the date at the top, and they change without notice. Two things are not solved here either: paying for a foreign subscription, which has a subject of its own and is followed in the crypto and international payments path, and reaching a tool that does not open from Iran, where we write unknown rather than guess. And the last one matters most: if the client already has tools, none of this applies and you work with theirs.
From our own work
We had to make this same rule mechanical in our own code, because remembering it was not reliable. On rgb.ir no customer ever gets a wp-admin account: inc/account.php in the theme closes the admin area and returns the person to their own panel. The part that gets less attention is the test it uses, which is deliberately not the user role. It is the absence of the edit_posts capability, so a role added next year is locked out by default rather than because somebody remembered to lock it. The right answer to a client who wants to send their password has exactly that shape: the safe version of a rule is the one that works without being remembered.
Real follow-up questions
The client has no tools at all. Where do I start?
With two, not six: one place for dated tasks and one place for files. Those two remove most of the confusion, and the client does not have to learn two new tools at once. Add the rest when a real problem asks for them.
Do I have to pay for these tools?
Not at the start. For an assistant with one or two clients the free plans are almost always enough, and what usually pushes you onto a paid plan first is not features but history and the number of seats. And if a paid plan does become necessary, the cost of the client tool belongs to the client rather than to you; say that plainly from the beginning.
The client wants everything to stay on WhatsApp. What do I do?
Work there, but build one single place for the things that have to survive and drop its link into that same chat. Chat is for coordinating and is not an archive; three weeks later nobody finds that message. The client does not have to learn a new tool for it either: one shared document that only you write in and they read is enough.