iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Yes—when a person triages or reviews an agent-generated pull request (PR), that attention is spent whether the change eventually merges, is rejected, or is abandoned. But there is no established universal figure for the minutes spent on non-merging agent PRs, or for their share of review effort across organizations. Teams should measure reviewer time alongside PR outcomes, latency, and review quality rather than infer cost from submission counts alone.
Why a PR can cost review time without shipping
A PR can require human attention before anyone decides whether its code is worth merging. A reviewer may assess its purpose, check whether it duplicates or conflicts with other work, request changes, or determine that it should not proceed. If the PR is later abandoned, superseded, or rejected, that earlier effort does not disappear from the team’s workload.
The cost is not necessarily the same for every submission. A PR screened out quickly may take little time; one that receives substantive review may take considerably more. The available sources do not establish an average number of minutes for an agent-generated PR that never merges. The defensible claim is about accounting: effort spent on triage or review remains effort even when no change ships.
What the available evidence does—and does not—show
Salesforce’s reported experience
In a January 29, 2026 article, Salesforce Engineering authors Shan Appajodu and Ravi Boyapati described changes within their own engineering organization as AI-generated code increased. They reported that code volume rose by approximately 30%, that PRs regularly grew beyond 20 files and 1,000 lines of change, and that review latency rose quarter over quarter. These are Salesforce-specific observations, not industry-wide estimates.
#1 Best Overall
The authors also said review time on their largest PRs began to plateau or decline. They wrote: “This indicated that reviewers were no longer meaningfully engaging with changes.” That was their interpretation of the trend in Salesforce’s setting; lower review time elsewhere could have other explanations, such as effective early triage or a change in PR mix. The article does not provide a population, sampling method, baseline period, or estimate of time spent on PRs that never merged. Read the Salesforce Engineering article.
What remains unquantified
The evidence does not establish how many reviewer minutes are spent, on average, on non-merging agent PRs, nor what portion of organizations’ review workload they represent. Accordingly, a team should not use Salesforce’s figures as a proxy for its own costs or treat them as proof that agent submissions cause a particular amount of review delay.
How to measure review effort without hiding abandoned work
Track reviewer effort for both intake triage and substantive review, and define what counts as reviewer time consistently. For a team and a stated time period, report the following measures together:
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 reinstall- Total reviewer time, including triage and substantive review, and time per submitted PR.
- Submission outcomes: counts and shares merged, abandoned, rejected, or superseded. Define these categories so that the same PR is not counted inconsistently.
- Time per merged change: total reviewer time divided by merged changes in the period, alongside time per submitted PR.
- Review latency and a quality check, such as whether required human review occurred and whether post-merge issues changed.
Time per submitted PR describes effort attached to intake. Time per merged change makes the review effort associated with delivered changes more visible, but it is a diagnostic—not a universal productivity score. Work may span reporting periods, and multiple PRs may be bundled into a single delivered change, so state the denominator and outcome rules clearly.
Compare like with like across periods, teams, and PR classes. If time per PR falls, examine whether submissions are being screened efficiently, the mix of PRs has changed, or reviewers are spending too little time on changes that need scrutiny. A single ratio cannot distinguish those explanations; review outcomes, latency, and quality provide necessary context.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why review-time trends need context
Falling review time may reflect faster, more effective triage—or less meaningful reviewer engagement. Salesforce reported the latter interpretation for its own largest PRs, but that does not establish what a similar trend means at another organization. Likewise, rising PR volume or latency alone does not identify which submissions consumed the effort or whether they ultimately shipped.
Salesforce describes Prizm, its internal review system, as reconstructing intent, adding work-item and historical context, and delivering feedback in IDE and PR workflows. It is an example of an internal approach, not a generally available product recommendation or evidence that another team should adopt the same system. Salesforce Engineering’s article describes Prizm.
Quick Recap
Best Value
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.
