A8gent
GlossaryAutomation mechanics

What is serviceNow assist?

Also called: now assist consumption, servicenow assists, now assist pricing unit, how many assists does now assist use

An assist is the unit ServiceNow meters Now Assist in - roughly one per skill invocation, with third-party LLM calls charging one assist per call and consuming additional assists past 1,000 output tokens - and ServiceNow publishes no dollar price per assist.

Written by A8gent
Contents
  1. At a glance
  2. What an assist is
  3. The mechanics that are documented
  4. The part that is not public
  5. How to build a usable estimate anyway
  6. FAQ
  7. Related terms

At a glance

Unit
One assist - a per-skill weighted unit of Now Assist consumption
OOTB skills on Now LLM
1 assist per request as standard, rising past output thresholds
Third-party LLM calls
1 assist per call at base rate
Long outputs
Calls exceeding 1,000 output tokens consume additional assists; 3,000 tokens is charged 3 assists
Errors
Not charged - "we do not charge an assist" when the LLM returns an error
Guardrail trips
Not charged
Per-skill rate card
Referenced but not published inline - obtain it from your rep
Price per assist
Not published. ServiceNow directs you to your account representative (checked September 2026)

What an assist is

An assist is ServiceNow's consumption unit for Now Assist. ServiceNow's own framing is deliberately loose: "Now Assist consumption is measured by assists. Different skills and uses consume different numbers of assists."

That sentence is doing more work than it appears to. It means the assist is not a fixed quantum of work like a Zapier task - one successfully completed action, always - and not a fixed quantum of time like a Latenode CPU second. It is a per-skill weight, and the weights live in a rate card rather than in the definition. Two features that look equally trivial to a user can carry different assist costs, and there is no way to reason your way to the number from first principles.

The mechanics that are documented

ServiceNow's Now Assist FAQ does publish the structural rules, verified in September 2026:

Out-of-the-box skills using Now LLM consume one assist per request as the standard rate, with consumption rising when output exceeds defined thresholds.

Custom skills built with Skill Kit or the Generative AI Controller against a third-party LLM charge one assist per call as the base rate. Calls exceeding 1,000 output tokens consume additional assists, scaling with output length - ServiceNow's own worked example is that output reaching 3,000 tokens is charged three assists.

Errors and guardrail trips are not charged. When a third-party LLM returns an error, ServiceNow states "we do not charge an assist", and a Guardian guardrail trigger likewise avoids the charge.

Those three rules are the actual design, and they are enough to tell you where the risk is. Output length is the variable. Input length is not mentioned in the escalation rule; output is. A skill whose prompt encourages long, thorough answers costs multiples of the same skill constrained to a short answer, and the multiplier is roughly the number of thousand-token blocks it emits. If you are building custom skills, the single highest-leverage cost control is an output length instruction in the system prompt and a hard max-tokens cap on the call.

The error exemption is the mirror image of Zapier's "successful actions only" rule and works out the same way: a broken integration is cheap, a working one is not. It also means a retry storm against a failing provider will not silently drain your entitlement, which is a genuinely good property. See retry and backoff for the design that keeps that true.

The part that is not public

Two things are missing from the public record, and both matter.

First, the per-skill rate card. ServiceNow's FAQ references an assists rate card and links out to a separate document rather than stating the weights inline. So the answer to "how many assists does this specific feature cost" exists, is written down, and is not on the page that defines the unit. Per-skill figures do circulate on third-party blogs. We are not repeating them, because we could not verify them against a ServiceNow-published rate card, and a per-skill weight that is wrong by a factor of ten is worse than no number at all in a capacity model.

Second, the price. There is no public dollar price per assist. ServiceNow's guidance is to contact your account representative. Now Assist entitlements are packaged into enterprise agreements alongside licensing you are probably already negotiating, and the effective rate is a function of that whole deal rather than a list price you can look up.

So the honest summary of ServiceNow's model is: the unit is defined, the escalation rules are defined, the per-skill weights are in a document you have to be given, and the price is in a contract you have to negotiate. This is the same shape as the Zendesk resolution allowance and for the same reason - enterprise deals price on the account, not the unit.

How to build a usable estimate anyway

You cannot calculate an assist budget from public information, but you can build the model that a rate card slots into. Do it in this order.

Count the invocations, not the users. For each Now Assist skill you plan to enable, estimate how many times a day it will actually fire - summarisations per incident, question-answering sessions per agent shift, code generations per developer. Invocation counts are yours to measure and are the only input to the model that no vendor can give you.

Separate out-of-the-box skills from custom ones, because the escalation rules differ. Then, for every custom skill, estimate typical output length and express it in thousand-token blocks, since that is the axis consumption scales on.

Take that table to your account rep and ask them to fill in the weights and the rate. Ask specifically: what is the assist weight for each skill on our list, what is our effective cost per assist at our committed volume, does the entitlement reset monthly or annually, what happens at overage, and does anything roll over. Then ask for the rate card itself in writing, because skill weights are the number most likely to change between renewals.

And keep the model bill separate in your head. Where Now Assist calls a third-party LLM, provider token costs may sit outside the assist meter depending on how the model is provisioned - confirm which side of the line yours falls on before you assume the assist covers it. See total cost of ownership for the implementation and change-management costs that no consumption meter reflects.

ServiceNow assist FAQ

What is an assist in ServiceNow Now Assist?

It is the consumption unit Now Assist is metered in. ServiceNow states that consumption is measured by assists and that different skills and uses consume different numbers of assists, so the assist is a per-skill weighted unit rather than a fixed quantum of work.

How many assists does a Now Assist skill use?

It depends on the skill. Out-of-the-box skills on Now LLM consume one assist per request as the standard rate, and third-party LLM calls charge one assist per call at base rate. The full per-skill weights live in a rate card that ServiceNow references but does not publish inline - ask your account representative for it.

Do long outputs cost more assists?

Yes. ServiceNow documents that calls exceeding 1,000 output tokens consume additional assists, and gives 3,000 tokens as being charged three assists. Output length is the variable that drives cost escalation, so capping max tokens and instructing brevity in the prompt is the main lever on a custom skill's consumption.

Are failed Now Assist calls charged?

No. ServiceNow states that when a third-party LLM returns an error it does not charge an assist, and a Guardian guardrail trigger likewise avoids the charge. A failing integration does not drain your entitlement.

How much does a ServiceNow assist cost?

There is no published price per assist. Now Assist entitlements are packaged into enterprise agreements and ServiceNow directs customers to their account representative. Any per-assist dollar figure you find online is third party and not sourced to ServiceNow.

How do I estimate my Now Assist consumption?

Count planned invocations per skill per day, separate out-of-the-box skills from custom ones because the escalation rules differ, and estimate typical output length in thousand-token blocks for the custom ones. That table is the part only you can produce. Take it to your rep to have the skill weights and the rate filled in.

Primary source

ServiceNow - Now Assist FAQs

Checked September 2026. Anything priced or rate-limited moves, so verify against the source before you budget.

Want this built rather than explained?

Knowing what servicenow assist means is one thing; having it running in your business is another. We scope one workflow at a time, quote a fixed price in writing after a free consult, build it, and hand it over with documentation you own.