Most teams aren’t short on talent. They’re short on oxygen.
The environment is noisy and unpredictable and full of things nobody fully documented, so your people spend their days reacting, translating and explaining. Then someone in a quarterly review asks why the backlog never shrinks, and half the room fantasises about a cabin in the woods with no Wi‑Fi and no ServiceNow instance.
This is the article about turning chaos into routine. Without turning your people into robots.
The quiet problem
Here’s what it actually looks like.
A $180/hr network architect is copy-pasting log entries into a spreadsheet because nobody ever built the automation. Nobody will build the automation, because the automation project is behind the other automation project that’s behind the compliance project that’s behind the thing Gary broke in January.
Meanwhile:
- alerts fire in triplicate because three tools are watching the same threshold and none of them know about each other
- incidents arrive with no context, no owner and no clear next step, so the first twenty minutes of every response is spent figuring out if this is real and whose problem it is
- changes break things because drift went unnoticed, and drift went unnoticed because the people who would have noticed left and took the institutional memory with them
- the runbook exists in Gary’s head. Gary’s phone is, functionally, a load-bearing infrastructure component
Nobody planned this. Nobody is incompetent. This is just what happens when five years of “just for now” decisions stack up and nobody gets the time to go back.
What actually helps
Clear the noise before it reaches a human
Three tools firing on the same event is not “redundancy.” It’s three tickets, three notifications, three people stopping what they’re doing to look at the same thing. Suppress the duplicates. Correlate the signals. Attach context before a person has to open it. If your ServiceNow instance is the intake, Kyndryl Bridge is the layer pulling signal from across the estate and making it legible. [source] It runs more than 100 million automations a month. That’s the scale. The principle is simpler: let systems sort so humans can think.
Automate the work nobody should be doing twice
If your senior people are spending a third of their week on tasks a configured runbook handles in seconds, that’s not a staffing gap. That’s an operating model that’s burning premium rates on clerical work and calling it “operations.”
Every hour your team spends on repeatable triage is an hour they’re not spending on the preventive work that actually reduces incident volume. Which means next quarter looks exactly like this one. Same fires, same heroes, same exhaustion, same quarterly review where someone asks why the backlog never shrinks.
Make visibility do something
AIOps dashboards are lovely. If they can’t route work, trigger a response or produce evidence for an auditor, they’re decoration. [source]
Here’s the part most ops teams don’t think about: the telemetry, incident patterns and change records your team generates every day are the same data that feeds AI-driven operations. When it’s clean and governed and flowing into the right systems, the tools you already own start doing more. When it’s not, you buy another tool. Then another. Then someone asks why you have nine monitoring platforms and still can’t answer what changed last Tuesday.
Keep the same people around
This is the one nobody puts on a slide. When the same senior technical lead stays on your account for two years instead of cycling every six months, patterns hold. Knowledge compounds. Problems get solved instead of rediscovered.
When someone rolls off and returns three months later with context intact, that’s not a staffing convenience. That’s the difference between an environment that improves and one that keeps resetting to zero every contract cycle. It’s also the difference between Gary taking a vacation and Gary leaving a binder of emergency procedures on someone’s desk like he’s not coming back.
What gets better
When this works, it’s not dramatic. It’s quiet.
Mean time to resolution drops because incidents show up with context already attached. Change becomes safer because drift is visible before it becomes an incident. Your team spends more time preventing fires than surviving them. Your best people stop living in a chat room called “urgent.”
Gary sleeps through the night. Not because the problems disappeared, but because the systems handle what used to live in his head.
Not glamorous. Measurable.
What to fix first
Catalogue what you have Tie services, owners and incident patterns together. If ops can’t see what’s real, everything else is guesswork.
Build privacy into daily work Ops telemetry is data. When your monitoring routes through a foreign vendor’s cloud, you’ve created a sovereignty question you didn’t mean to ask. [source]
Operationalise governance Response steps, approvals, exceptions. In systems. Enforceable. Reviewable. Not in somebody’s head.
Secure the data itself Operational evidence, backup artefacts and telemetry are targets. Treat them like it.
ISM builds the operating model. The intake, the runbooks, the automation layer, the ownership structure that keeps it from drifting back to chaos in six months. Kyndryl brings the platform and the patterns refined across thousands of environments. Your team gets to do engineering. Delivered locally, by people who know your change windows and will still be on your account next quarter.
