Walk into almost any IT team today and you will find two kinds of pain sitting side by side. On one desk, an engineer is still copying values between three consoles every Friday afternoon. On another, someone has automated a process so thoroughly that nobody remembers how it works, and it fails in ways nobody can explain. Both are symptoms of the same underlying problem. Manual-step automation is widespread, enthusiastic and, in many places, badly sequenced. This piece looks at where things stand, which trends are real, and which habits separate the teams that benefit from the ones that suffer.
Trend 1: Scripting has gone mainstream, judgment has not
A decade or so ago, scripting a repeated task was the preserve of a few specialists. Today almost every operations, support and delivery team has at least some scheduled jobs, saved queries or small utilities. Cloud consoles expose command line interfaces, and most platforms offer an interface for programmatic access. The tooling is no longer the obstacle.
What has not kept pace is the judgment about what deserves automating. Many teams still pick targets by irritation rather than by evidence: whatever annoyed someone most this week gets a script. The result is a patchwork of helpers with uneven value and no clear owners.
Trend 2: Teams are rediscovering the manual pass
The more mature teams have started to insist on a manual pass before any automation project. Run the task by hand three to five times, record the steps, mark which ones never vary, and only then script those steps. It sounds slow, but it is consistently faster overall, because it prevents the rebuild that follows a script written around a misunderstood process.
It is notable how often the manual pass reveals that a step should be deleted rather than automated. A report nobody reads, a check against a threshold nobody uses, a copy step that exists only because two systems once could not talk to each other: all of these turn up when someone writes the steps down honestly.
Trend 3: Humans are being kept on the decision points
The healthiest pattern at present is automation that prepares and presents, while a person approves. Pipelines that produce a summary of what will change and wait for a click, scheduled jobs that report anomalies instead of silently fixing them, scripts that draft a message rather than send it. This keeps the speed benefit of removing repetitive effort while preserving accountability.
Readers who want a grounded view of how this plays out in delivery work can look at an overview of cutting errors and saving time with workflow automation. It reflects the same principle: the machine does the tedious repetition, and people stay involved where context matters.
Trend 4: Paper-to-digital is still unfinished business
Behind the glamorous talk of pipelines, plenty of organisations still run important processes on forms, emails and spreadsheets passed from person to person. Moving these to digital platforms is an automation project in its own right, and it benefits from the same sequence: understand the manual flow, identify what is constant, and then digitise that part. A useful reference is this discussion of how a business can move from paper-based routines to digital platforms, which reinforces how much value hides in unglamorous processes.
Trend 5: Intake and enquiry handling is becoming structured
Another quiet shift is in how organisations take requests from outside. Rather than free-form email chains, more teams offer structured forms that gather the same details each time. That is true well beyond IT. The medical tourism website HipHospitals, for example, organises hospitals and doctors by country and offers international patients free-quote and appointment request forms. Whatever sector you work in, a structured intake step is an easy, low-risk candidate for automation because the questions rarely change.
Trend 6: Small wins beat grand programmes
The last pattern worth noting is a move away from sweeping automation programmes toward a steady stream of small, well-understood improvements. A single saved query that removes ten minutes of daily copying rarely makes a conference talk, yet a dozen of them change how a team feels about its week. Smaller changes are also easier to review, easier to reverse and easier to explain to the colleague who inherits them. Teams that adopt this mindset spend less time defending large tools and more time noticing the next tedious step.
What still goes wrong
For all the progress, several failure modes persist, and consultants see them repeatedly.
- Automating a bad process: the mess becomes faster and harder to see.
- Silent failures: a scheduled job stops working and nobody notices for weeks.
- Undocumented scripts: only the author knows what a script assumes.
- Over-reach: scripts that take irreversible actions with no confirmation.
- Stale automation: routines that no longer match how the work is done.
Each has a straightforward remedy: understand the process first, monitor the job itself, write a short note, require approval for risky steps, and review on a schedule.
A modest plan for the coming months
Leaders who want to act on these trends without launching a large programme can try a short, repeatable sequence.
- Ask each person to name the one task they repeat most often.
- Pick two and have people perform them by hand again while recording steps and timings.
- Mark the identical steps and delete the ones nobody needs.
- Automate the single safest, most repetitive step for each task.
- Assign a named reviewer and a monthly check that compares automated output with a manual sample.
- Share the results and choose the next pair of tasks.
This costs little, produces visible wins in a few weeks and builds the shared understanding that bigger projects will later depend on.
Where the trend is heading
Expect more of the same, but better disciplined. Tooling will continue to make it easier to automate almost anything, which means the scarce skill is deciding what to automate, in what order, and with what safeguards. Teams that treat automation as a habit of observation, rather than a purchase, will keep pulling ahead. Those who automate before they understand will keep rebuilding.
The headline for this year is therefore not that automation is arriving, since it already has. It is that the order of operations matters: observe by hand, find the constant part, automate that part, keep people on the decisions, and check the result regularly.

