Remote work tools — how not to end up with seven apps
Chat, video, documents, tasks, data. How to choose a toolset for a distributed company, and why fragmentation costs more than the licences do.
A distributed company rarely has a problem with too few tools. It has a problem with too many.
The typical set: a messenger for conversations, a separate video app, cloud storage for files, a task tool, spreadsheets for data, an inbox for customer contact. Six places. Each good at its job. Together — a machine for losing context.
The real cost of fragmentation
Licences are the smallest item. Three bigger ones:
Carrying context. A decision is made on a call. Someone retypes it into a task. The task refers to a file on a drive. The file concerns a customer whose details are in a spreadsheet. Four hops, each one a chance to lose something — and something regularly is lost.
Losing the history of decisions. Three months later someone asks "why did we do it this way". The answer was in a call nobody recorded, or in a thread two hundred messages back. Reconstructing it takes an hour, or fails.
Onboarding cost. Six tools mean six accounts, six sets of permissions and six things to learn. In a company with any turnover that is a standing tax, paid on top of everything else that the first thirty days of a remote hire already have to cover.
The organising principle: context follows the data
The point is not to have one tool for everything. The point is for the conversation, the file and the decision to be attached to the matter they concern — rather than to the channel they happened to occur in.
In practice that means several things:
Chat at the data, not beside it. A discussion thread pinned to a specific job rather than a "projects" channel mixing twenty matters. A year later you open the job and the whole history is right there. Channels and mentions get a message read; they do not give it an address, which is why comments on the job outlive chat messages.
Files attached to the matter. Not in a drive folder whose structure one person remembers.
Calls started from the matter. Clicking "call" on a job instead of hunting for a link and pasting it into chat. That saves a minute on every call and — more importantly — means it is clear afterwards what the call was about.
Conclusions recorded where they were reached. A note on the matter after the call, not in a private notebook.
What this means when choosing tools
The cheapest tool is the one you do not buy — running projects on data the company already holds often keeps an app out of the set instead of adding one.
Three questions worth asking before you buy:
How many times a day will people jump between apps? If a typical task needs two tools, that is a design for future problems.
Can the data be exported? Remote work rests on tools that accumulate the entire memory of the company. Worth knowing the cost of exit before that memory grows.
Does it cover people outside the office? A distributed company is not only people working from home. If you have field crews, the tool has to work for them too — otherwise half the information is created outside the system anyway. What that group requires is covered in the guide on managing field teams.
Where AI genuinely helps distributed work
Three uses where the gain is clear:
Meeting summaries. A conversation turned into a note with a list of decisions. This solves the most expensive problem of remote work — decisions that were made and evaporated.
Answers about the state of things. "What is happening with this customer's order" asked once, instead of three messages to three people. The condition: the data has to be in one place, or the model answers from a fragment.
File descriptions on upload. A document uploaded to the system gets an automatic description, so it can be found without remembering its filename. Sounds minor, solves the most common question in a distributed company.
A shared condition for all three: the model must respect permissions. A summary of a meeting you had no access to is a leak, not a feature. We develop that thread in the guide on connecting AI to company data.
What to avoid
Do not migrate everything at once. Moving six tools over one weekend ends with a month of nothing working and the company reverting to the old set; the order in which the old tools get switched off decides whether the consolidation holds.
Do not introduce a tool without retiring the previous one. Two parallel places for the same thing guarantee the information will be in both and complete in neither.
Do not assume video replaces a written record. Talking is faster than writing, but it leaves no trace. A distributed company needs both, and it pays to know when a written thread beats a call before the calendar fills up.
Where to next
- Managing field teams — for crews outside the office
- Connecting AI to company data — plugging a model into your own information
Frequently asked questions
- How many tools does a remote company need?
- Fewer than it usually has. A typical company of 11–200 people runs separate tools for chat, video, documents, tasks and data — five places between which context has to be carried by hand. The cost of that carrying exceeds the cost of the licences.
- What is the biggest cost of tool fragmentation?
- Not licences but lost context. A decision is made on a video call, the task is created in another tool, the file sits in a third — and a month later nobody can reconstruct why the decision was made. That cost never appears in the IT budget.
- Does a company of 11–200 people need a separate video tool?
- It needs video, but not necessarily a separate tool. A call started from the data it concerns skips the 'let me send you a link' step and keeps the conclusions attached to the matter rather than in someone else's calendar.
- How do you include field workers in a remote company?
- By giving them access to the same data in a form that is usable on a phone — a map instead of a table, a form instead of a document. Remote work and field work are the same distribution problem solved with different views of the same information.
Read next
Comments on the work, not messages in the chat
An agreement from three weeks ago is technically in the chat and practically gone. Why a comment on the job stays findable and a channel message does not.
Onboarding a remote employee — the first thirty days
Remote onboarding breaks on the knowledge nobody wrote down. What to prepare before day one, why a buddy matters, and checkpoints at days 7, 14 and 30.
Running projects without buying a project tool
A project is a view of data you already hold: tasks, dates, people, files, decisions. What has to be recorded, what is decoration, when a separate tool pays.