If It Lives in Someone's Head, It Isn't a Process Yet: Why Documentation Comes Before Automation
Documentation turns tacit knowledge into repeatable processes. Learn why recording how work actually runs is essential before automation, tools or AI.
The gap between how work happens and how it is documented
Many organisations discover a painful truth when they attempt to automate a workflow or introduce an AI tool: the process does not exist on paper. It exists in someone's head.
A manager approves expenses by instinct. A team lead knows which invoices need chasing because they remember the payment terms. A co-founder runs payroll from a mental map of who earns what, when and how. None of it is written down. When that person leaves, takes leave or becomes the bottleneck, the process vanishes.
Automating a process that lives only in memory is almost impossible. AI assistants cannot be trained on what was never documented. Spreadsheets built on unwritten rules fail when new people join. Even the best tools cannot fix work that has never been made visible.
Documentation is not about creating perfect process maps. It is about recording how work actually runs, so it can be repeated, improved, taught and eventually, if it makes sense, automated.
Why tacit knowledge stays tacit
Most organisations do not deliberately choose to keep processes undocumented. It happens because:
- Work feels too complex or contextual to write down.
- People are too busy doing the work to describe it.
- There has never been pressure to make it explicit, someone has always known how to do it.
- Early-stage or rapidly growing teams prioritise action over process.
- Documentation feels like bureaucracy, not support.
The cost is real. Undocumented work creates dependency on individuals, slows onboarding, makes quality inconsistent and prevents any meaningful improvement or automation.
What documenting actual process looks like
Documentation does not mean lengthy process manuals or formal flowcharts, though those have their place. It means recording, clearly and practically, what actually happens.
For example, imagine a small charity processing grant applications. The work happens: someone reads each application, checks it against the funder's criteria, makes notes, discusses uncertain cases with the team lead and either approves or requests more information. That is the process. It should be written down in a way that a new team member could follow: what documents to read, what criteria matter, when to escalate, who makes final decisions. Once it is visible, inconsistencies become obvious. Perhaps the team lead applies stricter rules than the person processing applications thinks. Perhaps some criteria are guessed at because no one wrote down what the funder actually requires. Documenting the process reveals these gaps.
Similarly, a founder managing client onboarding might realise that they are doing seven different activities: reviewing the contract, setting up a project folder, creating a checklist, sending a welcome pack, adding the client to the CRM, scheduling a kickoff call and creating an invoice. Each step happens in someone's head. Written down as a simple sequence with decision points, and with the template folder, email and checklist attached, it becomes teachable and automatable. A junior team member can do it. Later, some steps might be automated.
Documentation as the foundation for AI readiness
This matters especially as organisations consider AI and automation. Many attempt to deploy an AI assistant or workflow automation tool before their processes are documented. It rarely works well. The AI has nothing clear to work from. The tool automates guesswork. The result is low-quality output and frustration with the technology.
When process is documented first, even roughly, something changes. The AI assistant has clear context. The automation workflow has defined rules. A person reviewing the output can spot what went wrong because they know what right looks like. Documentation does not replace judgement; it enables it.
An example: a team might want to automate routine approval workflows. Before the tool is set up, someone should document: what requests come in, who decides whether each is approved, what information they need to decide, what happens if they say no, who gets notified and what happens next. Documenting this usually reveals that the decision is not as routine as it seemed. Three different people make similar decisions differently. The organisation does not have a clear policy. The request form is missing information that decides should have. Fixing those things first, not automating around them, makes the eventual tool far more useful.
Starting documentation without overwhelming anyone
Documentation does not have to be elaborate or time-consuming:
- Record the next time someone does the work. Write down the steps as they happen, in plain language.
- Let the person who does the work document it themselves, in their own words. They own the knowledge; formalising it is their work.
- Use simple templates: what you do, in what order, with decision points and where to find templates or tools.
- Start with the work that matters most: processes that are repeated often, involve money, affect customers or consume senior time.
- Review and improve as the process is documented. The act of writing reveals what needs to change.
The goal is visibility, not perfection. A roughly documented process is more useful than an undocumented one.
The foundation before tools
Documenting how work actually runs is foundational work. It comes before automation, before AI, before new tools. It clarifies what is happening, reveals what is inconsistent and creates the clarity that makes change possible.
If a process lives only in someone's head, it cannot be improved, taught, scaled or automated. Making it visible is the first step to making it stronger.
If your organisation is carrying undocumented processes or institutional knowledge that sits with a few people, clearing that backlog and documenting what actually happens is practical and essential work. VAxAI specialises in helping organisations recover backlogs and document processes so that work becomes visible, repeatable and ready for improvement or automation. A free Admin Review can help you understand where documentation gaps are and what support would help most. Start a conversation with us about clearing your backlog and documenting your foundations.
Get in touch
Have a question about this post or want to explore working together? Your enquiry will be linked to this content.