Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsInfrastructure plan review stops functioning as a meaningful control when the volume of proposed changes outstrips the attention reviewers can give them. The approval step may still appear in the workflow, but an approval record alone cannot show that someone carefully checked compliance, risk, intent, and evidence. In his September 24, 2026 article for DevOps.com, Mariusz Michalowski argues that teams should automate checks that can be expressed as rules, use constraints to limit blast radius, preserve human judgment for whether a change fulfills its request, and make enforcement produce an audit record.
Why an approval can remain visible but lose its value
Infrastructure plan review asks a person to inspect a proposed change and decide whether it should proceed. As Michalowski describes it, that approval step has often been expected to do four jobs at once:
As an Amazon Associate I earn from qualifying purchases.
- Check the change against organizational rules.
- Estimate its potential blast radius.
- Decide whether it implements the request that prompted it.
- Leave evidence that an evaluation took place.
All four jobs require attention, but they do not all call for the same kind of judgment. When review volume grows beyond what people can inspect carefully, clearing the queue can become the practical priority. The workflow still records approvals, yet the record no longer establishes that each change received meaningful scrutiny. Michalowski captures this distinction with the line, “A saturated control emits the same signals as a working one.” That is his diagnosis, not a finding attributed to an independent standards body or regulator.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →What makes review capacity saturate
Generated changes can be harder to recognize
Michalowski points to generated code as a challenge: reviewers may lack a familiar baseline against which to recognize patterns or anomalies. The issue is not simply the number of lines in a plan, but whether a person can interpret what the change means in context.
#1 Best Overall
Slow reviews can encourage workarounds
If waiting for review takes too long, people may turn to an unplanned console change rather than wait for the governed path. That creates a gap between the intended workflow and the infrastructure state teams need to manage.
Routine records may conceal shallow review
A standard audit event can look identical whether an approval followed close inspection or a quick click to clear a queue. That makes approval counts a weak proxy for control quality when the review process is overloaded.
The article’s examples of “40 pull requests before lunch” and “90 seconds” illustrate the pressure scenario; they are not reported measurements or statistics.
Rank #2
Give rules, constraints, and human judgment separate jobs
Michalowski’s proposed redesign starts by matching each review responsibility to a mechanism suited to it. Machine-evaluated rules can handle requirements that are explicit; constraints can limit the damage a change can cause; people can judge whether the proposed change actually answers the request.
| Review responsibility | Proposed mechanism | What still needs attention |
|---|---|---|
| Compliance with written rules | Evaluate policy during the infrastructure run, and evaluate again after state changes. | Teams must define and maintain the rules they want enforced. |
| Blast-radius management | Set operational boundaries such as expiry, spending caps, permitted resource types, and separation from production data. | Choose limits that fit the environment and the risk being controlled. |
| Whether the change fulfills the request | Keep this as a human review responsibility. | Compare the proposed change with the actual request and its context. |
| Evidence of evaluation | Generate records from enforcement, including the rule, input, decision, and time. | Use scheduled drift detection and track how long unintended state persists. |
Automate explicit policy checks
Rules that can be written down should not depend solely on a reviewer remembering to notice a violation. The article recommends evaluating compliance as part of the infrastructure run and reevaluating it after state changes. It also proposes deny-by-default treatment for costly or difficult-to-reverse areas such as identity and access management, networking, and data. Exceptions defined by policy can then be routed to people rather than silently accepted.
Keep request intent with reviewers
A policy engine can inspect resource attributes, but those fields alone do not reveal whether the plan implements what a person asked for. The proposed division of responsibility therefore does not eliminate review: it reserves human attention for intent verification, where understanding the request matters.
Rank #3
Use constraints instead of relying on a quick risk estimate
A reviewer has limited time to predict the consequences of a proposed change. Michalowski argues that teams can reduce reliance on that short-time estimate by imposing boundaries on what experimental infrastructure can do.
- Expire experimental infrastructure rather than leave it running indefinitely.
- Cap spending to limit financial exposure.
- Restrict which resource types an experiment may create or change.
- Keep experiments isolated from production data.
These controls do not determine whether a change is correct or appropriate. They bound its possible impact while leaving request-specific judgment to the reviewer.
Separate production traceability from experimental speed
The article advocates a two-path model rather than forcing every change through the same workflow. Production infrastructure-as-code and GitOps remain the system of record; experimental work can use a governed fast path that permits speed within defined limits.
| Dimension | Production path | Experimental path |
|---|---|---|
| Primary purpose | Operate production infrastructure through the system of record. | Enable faster experimentation within governance boundaries. |
| Traceability and speed | Traceability through infrastructure-as-code and GitOps. | Governed speed, subject to the controls set for experiments. |
| Decision split | Machine-evaluated rules check defined policy; human review checks whether the change meets the request. | Rules and constraints define permitted activity; the article does not specify a separate human-review requirement for every experimental change. |
| Blast-radius approach | Manage changes through the production system of record. | Bound risk with controls such as expiry, spending caps, resource restrictions, and production-data isolation. |
The distinction is about governance design, not a quantified comparison: Michalowski’s article provides no measured outcomes for the two paths.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Make audit evidence an output of enforcement
If an audit record must show more than that someone clicked approve, the enforcement process can record what rule was evaluated, what input it received, what decision followed, and when that decision occurred. This ties the record to the control’s actual operation rather than relying on an approval event as a proxy for scrutiny.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThe article also recommends scheduled drift detection. A raw drift count says how many unintended states were found; drift mean time to repair (MTTR) adds how long those states persisted before correction. That duration can make the exposure more visible, although the article does not provide a target MTTR or comparative results.
Best Value
- Durable hardcover with concealed wire-o binding
- Archival, acid-free paper helps preserve your information.
Product capabilities described in the article
Michalowski names Spacelift Intelligence, Intent policies, and Infra Assistant Build mode. He describes attached Intent policies as being evaluated before create, update, delete, import, and refresh operations. In that account, a matching deny rule produces an explicit denial reason, while an operation with no matching rule is held for review. These are capabilities as described in the DevOps.com article; they are not independently verified here against official product documentation.
The broader governance argument does not depend on those named products: teams need explicit rules for machine checks, bounded environments for experiments, human review of intent, and records that show enforcement decisions.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
