Skip to content

Fixes and safety

Validated fixes

What happens to a finding after you approve it, through to the billing data that confirms the saving was real.

Each finding lands with evidence and an approve or reject decision. During an engagement, an architect prepares each approved change.

Your team reviews and merges the change through its own pipeline. Nothing writes to your production estate automatically.

From finding to verified saving, with a key noting that dashed means roadmap, not built. A week axis runs week one, week two, weeks 3 to 4 across two bands. The cyan Infralign band, marked read-only, holds step 1, the finding lands, with a document glyph and tagged evidence attached; step 3, an architect plans the fix, tagged done by a person; and step 5, the saving is verified, tagged done by a person. The amber band below (labelled your team, your pipeline, and marked every write to your estate happens here) holds step 2, you approve or reject, tagged in the app; step 4, you review and merge, tagged your pipeline; and a dashed slate card, automated pull requests, tagged roadmap, not built. Arrows run from the Infralign band down into your band in week one and in week two. From finding to verified saving, with a key noting that dashed means roadmap, not built. A week axis runs week one, week two, weeks 3 to 4 across two bands. The cyan Infralign band, marked read-only, holds step 1, the finding lands, with a document glyph and tagged evidence attached; step 3, an architect plans the fix, tagged done by a person; and step 5, the saving is verified, tagged done by a person. The amber band below (labelled your team, your pipeline, and marked every write to your estate happens here) holds step 2, you approve or reject, tagged in the app; step 4, you review and merge, tagged your pipeline; and a dashed slate card, automated pull requests, tagged roadmap, not built. Arrows run from the Infralign band down into your band in week one and in week two.
Figure 1: Every write to your estate happens in your own pipeline. The two steps marked done by a person are delivered by an architect, not by the platform.

Steps 1 and 2 of figure 1 happen in the app. Steps 3 to 5 are the engagement.

StepActorObservable outcome
1. Finding landsInfralignThe finding carries impact, evidence, owner, effort, risk, and next action.
2. Approve or rejectYour teamThe app records the decision.
3. Plan the fixSenior Azure architectThe plan addresses your architecture and policy constraints.
4. Review and mergeYour teamThe approved change passes through your pipeline.
5. Verify savingsInfralign and the architectBilling data shows the realised change against the baseline.

See reading your audit report for the six decision fields.

A finding as it lands in the app, with severity, confidence, effort and basis chips above the estimated monthly saving.

A real finding from NovaPulse, a production estate whose identifiers are replaced. The euros are that estate’s own.

Approve or reject each finding in the app’s Recommendations workflow. Every finding there carries an estimated saving, a confidence level, a risk level and a decision status. The tabs separate what is still pending from what your team has already approved or rejected.

The Recommendations page. A green banner reads "263 pending recommendations — €29.8K a month potential savings, €357.0K a year". Under it, tabs for all, pending, approved and rejected, a search box, and a sortable table with columns for title, resource, detection type, estimated savings, confidence, risk and status. Every row shown carries an amber "pending" status.

The decision list from the same estate, captured 4 August 2026.

Read that €29.8K as candidates, not as savings. It is Azure Advisor’s raw list at retail rates, before anyone has checked whether a change is safe on this estate. The sample audit sizes the same estate at €226 a month because it ran from the cost record alone, with no tenant access, so it could only count what the billing data proves. Expert review is what moves a line from the first number to the second.

Done when the finding shows the selected decision.

During an engagement, a senior Azure architect designs each approved change plan. The plan can include infrastructure as code.

The architect validates the plan against your policy and architecture constraints. Your team then reviews and merges it through your pipeline.

This expert-led method is what an engagement buys. See the track record. Platform-generated pull requests are roadmap, not built, as figure 1 shows.

If you asked Infralign to wire up a Teams channel, it receives the monthly summary and cost alerts. No self-service field exists for it, and nothing is posted to you unless you ask.

Teams carries operational notifications only. Make approve and reject decisions in the app.

Infralign verifies savings against the billing data used by your dashboards.

Done when the realised cost change reconciles with the agreed baseline.

Your team’s merge is the last of three gates. The safety model covers the two that run before a change plan ever reaches them.