In one vendor-reported ServiceNow configuration run, SNcode used 393,000 cost-weighted tokens, compared with 2.05 million for Claude Code with the ServiceNow SDK and 1.31 million for Build Agent. The figures suggest that prompts, tools, and supplied documentation can affect an agent workflow—but they do not establish typical savings. SNcode published the comparison and was one of the products tested, a conflict readers should keep in view.
What the comparison asked the agents to do
The task was to add two fields to a ServiceNow Incident record, place them on the Incident form without replacing existing fields, and create a business rule that blocks resolution in a specified case. The source says all three approaches used Claude Sonnet and received the same initial task prompt. It also notes that the recorded Claude Code and Build Agent runs had an additional instruction not to overwrite the existing form.
As an Amazon Associate I earn from qualifying purchases.
The comparison was published by SNcode on September 28, 2025, under the byline “Pavlo for SNcode.” SNcode was both the publisher and one of the compared tools. The article reports a single task run per approach, not a repeated benchmark.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What each run reported
| Approach | Reported rounds | Elapsed time | Cost-weighted tokens | Reported task outcome |
|---|---|---|---|---|
| SNcode | One | 7 minutes 44 seconds | 393,000 | Made the changes and tested them |
| Claude Code with ServiceNow SDK | Two | 22 minutes 30 seconds | 2.05 million | Business rule initially failed; a general follow-up was used to fix it, then the work was tested through the API in round two |
| Build Agent | Two | 21 minutes 50 seconds | 1.31 million | Article says it did not test the work; a form-layout issue was later corrected with duplicate fields |
In its own calculation, SNcode reported 5.2× fewer cost-weighted tokens than Claude Code with the SDK and 3.3× fewer than Build Agent. Those ratios describe this run only. The elapsed times, number of rounds, testing, and whether the requested form state was preserved are important alongside token use: a smaller number does not by itself mean a better or safer result.
#1 Best Overall
What “cost-weighted tokens” means
The comparison does not use raw total tokens for its headline metric. The article’s custom calculation weights output tokens at 5×, cache-write tokens at 1.25×, and cache-read tokens at 0.1× an input token, with the weights derived from Claude pricing ratios. It should therefore be read as a price-oriented composite, not as a count of all tokens exchanged or a universal measure of efficiency.
The article also says Claude Code fetched 13 ServiceNow SDK documentation topics totaling more than 85 KB of text, much of which remained in context, and accumulated 14.8 million cache-read tokens over two rounds. These are figures reported by the vendor article; they are not independently audited measurements. Documentation retrieval and retained context offer a plausible explanation for higher usage, but the comparison does not isolate their effects.
Why prompts and tools could change the result
With the model held constant in this comparison, the surrounding workflow differed: system prompts, tools, skills, documentation access, and execution details were not identical. The SNcode author’s interpretation is that a focused ServiceNow-specific prompt and compact tools helped reduce unnecessary context and tool activity. The run illustrates that possibility; it does not prove that any one prompt or tool choice caused the token gap.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Keep recurring instructions close to the task
For repeated ServiceNow work, instructions about the instance, expected form behavior, and verification can be encoded in a concise system prompt rather than rediscovered each time. State preservation should be explicit: for example, require adding fields without removing or replacing existing ones.
Prefer focused tools and targeted knowledge
A small set of tools that returns compact, task-relevant results can limit noise. Likewise, short skills for specific ServiceNow operations may be more economical than repeatedly loading broad documentation. These are practical design recommendations from the vendor’s interpretation, not independently validated savings claims.
Test whether added context earns its cost
Additional skills or memory may improve some tasks, but their tokens recur. A ServiceNow AI Research paper dated October 2026 reports that, across three WebArena domains and three models, a token-matched vanilla baseline matched or exceeded three augmentation methods in aggregate success rate while often using fewer total tokens; it reports a similar trend on WorkArena-L1 with Qwen 3.6-27B. The study is not a test of the three products in the SNcode comparison, and it notes material run-to-run variance. Its useful lesson is to compare augmentation with a baseline under a matched token budget rather than assume more context is automatically better. ServiceNow AI Research’s October 2026 publication.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why one task is not a general savings estimate
The case comparison has several limits that matter when interpreting its multiplier:
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →- It reports one small task run for each approach, so it does not establish typical savings or run-to-run variance.
- The Claude Code and Build Agent runs had an extra instruction about preserving the form, despite the article’s statement that they shared the same initial prompt.
- The approaches differed in documentation retrieval, rounds, testing behavior, and reported form outcome; token totals are not a comparison of identical execution paths.
- The article does not report an independent statistical estimate of expected production savings or isolate the effect of prompts, tools, or skills.
Broader benchmark context is a reminder that completing a narrow task once is not equivalent to reliable automation. ServiceNow AI Research’s WorkArena describes 29 ServiceNow-based tasks and concludes that “while current agents show promise on WorkArena, there remains a considerable gap towards achieving full task automation.” WorkArena does not validate the SNcode article’s token figures, but it helps frame why task success and verification belong beside efficiency metrics. WorkArena, ServiceNow AI Research, July 2024.
How to make a fairer comparison
For a useful evaluation of ServiceNow agent workflows, match the conditions that can change both the work and its token footprint:
- Use the same exact prompt, model and version, ServiceNow instance and starting state, permissions, and available tools.
- Define acceptance criteria in advance, including the business-rule behavior, field placement, and preservation of existing form content.
- Repeat runs to measure variation; report the number of rounds and retries, elapsed time, raw input/output/cache token counts, and any weighting formula separately.
- Record documentation and skill payload size, then verify the final instance state with the same tests for every approach.
- Disclose product affiliations and distinguish measured outcomes from the vendor’s explanation of why they occurred.
That approach makes it possible to tell whether an optimization reduces context or retries without trading away correctness, verification, or regression safety.
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.
