Now Assist Consumption Explained: What Every Platform Owner and CIO Needs to Know Before Renewal

You bought Now Assist. Your teams started using AI-powered skills like summarization, case resolution, and knowledge article generation. The productivity gains are real. But here’s the question nobody asked during the sales cycle. And, it’s now keeping platform owners up at night:

How exactly does Now Assist consumption work, and what happens when you exceed your entitlement?

If you’re a ServiceNow Platform Owner, CIO, or part of the procurement team managing your organization’s AI investment, this guide breaks down the Now Assist consumption model in practical terms. No vendor spin. No jargon soup. Just the operational knowledge you need to govern usage effectively and avoid surprises at contract renewal.

This article draws on insights from my LinkedIn series on ServiceNow AI adoption, distilled into one comprehensive reference for decision-makers. If you want to skip ahead, use the table of contents below.

In this article:

• What is an Assist — and why it’s not a token
• The new burndown model for Now Assist consumption
• Why account-level pooling matters for your Assist budget
• How to monitor Now Assist consumption effectively
• Five questions every CIO should ask about Now Assist consumption
• How to optimize your Now Assist consumption
• What procurement teams need to know

What Is an “Assist”. And Why Now Assist Consumption Is Not Measured in Tokens?

Let’s start with the distinction that trips up even experienced ServiceNow teams: an Assist is not a token.

In the broader AI industry, providers measure usage in tokens — the raw units of text processed by a language model. ServiceNow chose a different path. Instead, they introduced the Assist as a proprietary unit that abstracts away the underlying token mechanics.

So when does Now Assist consumption happen? Every time a skill action is invoked, one or more Assists are consumed. The exact number depends on the complexity of that action.

For example, a simple summarization costs just 1 (one) Assist. However, a document Q&A interaction costs 10. And here’s where it gets critical for organizations scaling AI Agents: agentic workflows consume between 25 and 150 Assists per execution, depending on their complexity tier.

As a result, the tiered Now Assist consumption model has significant implications for capacity planning. If your roadmap includes heavy use of AI Agents — and given ServiceNow’s strategic direction, it almost certainly will — you need to understand the cost multiplier before committing to a specific entitlement size.

Here’s a simplified breakdown of what common skill actions cost:

Skill Action	Assists Consumed
Summarization	1
Custom skill call (under 1,000 tokens)	1
Attachment summarization	5
Document Q&A	10
Knowledge article generation	10
Agentic workflow — Small	25
Agentic workflow — Medium	50
Agentic workflow — Large	150

The full and authoritative reference is the Now Assist Overview document. This legal document is referenced directly in your Order Form. If you haven’t read it yet — and in my experience, many platform owners haven’t — that should be your first action item.

The New Burndown Model: How Now Assist Consumption Measurement Changed in 2026

As of January 2026, ServiceNow fundamentally changed how Now Assist consumption is measured. This is not a minor update. It affects how your annual entitlement works, how overages are calculated, and how you should plan your rollout cadence.

The old model: Rolling 365-day lookback

Under the previous model, Assist consumption was calculated as a rolling total of the last 365 days. Usage never truly reset. As long as your rolling total stayed below your annual entitlement, you remained compliant.

The problem? This rolling window made precise forecasting extremely difficult. A consumption spike in month three could haunt you for the next twelve months, even if usage dropped significantly afterward.

The new model: Annual reset burndown

The new burndown model works differently. Your Assist pool now resets annually on your contract anniversary date. Usage accumulates from the reset date forward, and that cumulative total is compared against your annual entitlement. Consequently, if Now Assist consumption exceeds entitlement at any point, ServiceNow flags it as non-compliance.

Why does this matter for decision-makers? There are three key reasons.

First, predictability improves significantly. You now have a clean annual cycle aligned to your contract. As a result, budgeting and forecasting become much easier.

Second, the reset mechanism creates breathing room. Under the old rolling model, organizations that adopted Now Assist aggressively early in their contract could find themselves squeezed later. With annual resets, however, each year offers a fresh start.

Third, legal and operational alignment is cleaner. Your procurement team can now map Now Assist consumption directly to contract periods without the ambiguity of a rolling window.

There is one critical prerequisite, though. Your instances must be running Yokohama P12, Zurich P6, or later, with Subscription Management 6.0.2 installed. If you have multiple instances on mixed versions, consumption measurement will be inaccurate. This is not a “nice to have” upgrade — it’s a compliance requirement.

Additionally, if you hold multiple Now Assist entitlements with different start dates, the system uses the oldest entitlement’s start date as the contract anniversary for pool resets.

Account-Level Pooling: Why Sub-Production Instances Affect Your Now Assist Consumption

Here is a detail that catches many organizations off guard: Assist usage is measured and aggregated at the account level, not per instance.

In other words, your production instance, your staging environment, and your QA instance all draw from the same Assist pool. The only excluded instances are those explicitly designated as demonstration or evaluation environments.

What does this mean in practice? Extensive testing and development work in sub-production environments directly eats into your entitlement. If your teams run load tests, training sessions, or UAT cycles with Now Assist features enabled, those Assists count toward your total Now Assist consumption.

Therefore, multi-instance organizations need a governance conversation early on. Who owns the Assist budget? How is it allocated across environments? Without clear answers, you risk a scenario where a development team burns through a disproportionate share of your annual pool before production users have even ramped up.

How to Monitor Now Assist Consumption — And Why You Should Start Immediately

ServiceNow provides built-in tooling for tracking Now Assist consumption. The challenge is not data availability — it’s that most organizations don’t establish a monitoring cadence until it’s too late.

Subscription Management: Your primary Now Assist consumption dashboard

The most important tool is Subscription Management, accessible via Admin > Subscription Management. This application provides both account-level and instance-level views of your Assist usage. Specifically, it shows purchased vs. allocated quantities, over/under-allocation status, skill-level breakdowns, and monthly trend data.

If you’re a Platform Owner and haven’t yet reviewed your Now Assist consumption in Subscription Management, do it today. Not next quarter. Today.

Recommended monitoring cadence:

• Weekly — during active rollout phases or when approaching entitlement limits
• Monthly — for steady-state customers with comfortable headroom
• Quarterly — the absolute minimum for any Now Assist customer

AI Agent Analytics Dashboard: Tracking agentic Now Assist consumption

For organizations deploying AI Agents, the AI Agent Analytics Dashboard (AI Agent Studio → Analytics) provides dedicated visibility into agentic consumption. The Assist Consumption tab lets you filter by individual AI Agent, date range, and complexity tier. Moreover, it visualizes trends and identifies which agentic workflows drive the most consumption.

This dashboard runs on standard Platform Analytics components and works out of the box — no extra configuration needed.

Key ServiceNow tables for advanced Now Assist consumption analysis

For platform teams that want deeper insights, three tables are essential:

Generative AI Usage Logs (sys_gen_ai_usage_log) — Records the raw Assist consumption for every Now Assist feature call. This is the source table behind all usage metrics.

Usage Analytics Counts (usageanalytics_count) — Stores the results of ServiceNow’s counting definitions, including account-level totals (DEFN2006541) and instance-level totals (DEFN2018702).

Generative AI Logs (sys_generative_ai_log) — Captures detailed interaction data including prompts, responses, processing times, and user feedback.

Together, these tables enable custom reports, alerting mechanisms, and forensic analysis when Now Assist consumption spikes unexpectedly.

Five Questions Every CIO Should Ask About Now Assist Consumption

Based on patterns observed across dozens of Now Assist implementations, these are the questions that separate well-governed AI programs from those heading toward a compliance surprise.

1. “How much Now Assist consumption have we used so far?”

If you can’t answer this with a specific number right now, your governance gap is already wider than you think. Monitoring should never be reactive.

2. “What happens if we exceed our Assist entitlement?”

Exceeding your Assist pool triggers non-compliance status. While ServiceNow doesn’t immediately shut off functionality, this has real implications for your commercial relationship — especially at renewal. Procurement teams need to understand this dynamic early.

3. “Which skills are driving the most Now Assist consumption?”

Not all AI features consume Assists equally. If 80% of your Assist spend goes to a single skill that delivers marginal value, that’s an optimization opportunity. The Assist Ratio metric in Subscription Management is specifically designed to surface these imbalances.

4. “Are we tracking value alongside consumption?”

Consumption data without context is a dangerous metric. High consumption is only a problem if it isn’t delivering value. ServiceNow’s Value in Hours (VIH) Analytics Dashboard provides outcome-based metrics like hours saved and productivity gains. Always present VIH alongside Now Assist consumption data.

A high Assist count paired with high VIH is a success story. A high Assist count with low VIH, on the other hand, signals misconfiguration or under-adoption and warrants investigation.

5. “Are we scaling efficiently, or just scaling?”

As you roll out AI Agents, consumption increases non-linearly. For instance, a single large agentic workflow at 150 Assists per execution — triggered hundreds of times per day — can exhaust your entire annual pool in weeks. This isn’t theoretical. It’s happening to organizations that didn’t model their Now Assist consumption before going live.

How to Optimize Your Now Assist Consumption: A Strategic Framework

Monitoring consumption is necessary. Optimizing it, however, is where the real value lies. Here are the strategies that matter most.

Map every skill action to its Assist cost before go-live

Every planned skill deployment should come with a consumption estimate. Multiply the per-action Assist cost by the expected invocation volume, then model it against your annual entitlement. If a basic skill at 1 Assist can achieve the same outcome as a complex agentic workflow at 50 Assists, the decision is clear.

Audit for unused or low-value skills regularly

Skills that were enabled during a pilot but never governed can silently consume Assists in production. Therefore, conduct a quarterly review using the sys_gen_ai_usage_log table. Look for skills with high consumption but low user engagement or poor feedback scores.

Phase your rollouts to control Now Assist consumption growth

Not every user needs access to every AI capability from day one. A phased rollout — starting with high-impact, low-complexity skills like summarization, then progressively enabling more expensive capabilities — lets you validate consumption patterns before committing your full pool.

Build a governance cadence around Now Assist consumption

Make Assist consumption a standing agenda item in your platform governance meetings. Don’t treat it as an afterthought triggered by a breach notification. During active rollouts, review weekly. In steady state, review monthly at minimum.

Always combine consumption with value metrics

Never present Now Assist consumption data in isolation. Instead, always pair it with VIH or other outcome metrics. This reframes the conversation from cost management to ROI optimization — a far more productive discussion, whether you’re talking to your CFO or your ServiceNow account team.

What Procurement Teams Need to Know About Now Assist Consumption

If you’re on the procurement side, the Assist consumption model has specific implications for contract negotiation and management.

Right-size your entitlement at purchase. The Assist pool is defined in your contract and pooled at the account level. Under-purchasing creates compliance risk, while over-purchasing wastes budget. Use the consumption rates in the Now Assist Overview document, combined with your planned use cases, to build a bottom-up estimate before signing.

Negotiate with data at renewal. If you monitor Now Assist consumption throughout the year, you’ll arrive at renewal with concrete usage data and ROI metrics. This creates a fundamentally stronger negotiating position than relying on ServiceNow’s account team to tell you what you’ve consumed.

Understand the burndown model’s impact on multi-year deals. With annual resets, each contract year stands on its own. This simplifies multi-year planning but also means you can’t “bank” unused Assists from year one to cover overages in year two.

Factor in AI Agent growth. If your organization’s ServiceNow roadmap includes AI Agents — and it almost certainly will — ensure your entitlement accounts for the 25x to 150x consumption multiplier that agentic workflows introduce compared to basic skills.

The Bottom Line on Now Assist Consumption

Now Assist consumption is not a technical footnote. It’s a financial governance issue, a capacity planning challenge, and a strategic lever for your AI program.

The organizations that treat it as such — monitoring proactively, optimizing continuously, and governing rigorously — will extract significantly more value from their ServiceNow AI investment. Those who discover their Now Assist consumption reality only at contract renewal will not.

The shift to the burndown model in January 2026 is a reset opportunity — literally. Use it to establish the monitoring cadence, reporting infrastructure, and governance framework that your Now Assist program deserves.

Free Assessment Tool: Now Assist Consumption Governance Checklist

Take control of your AI investment. We’ve distilled the key principles from this article into a practical, one-page governance checklist designed for Platform Owners and CIOs. It covers the necessary areas for your next platform governance review. You can share it with your procurement team ahead of your next renewal discussion.

Need Help Optimizing Your Now Assist Consumption?

If you're looking for hands-on support with consumption monitoring, optimization strategy, or implementation governance, let's talk. We work with Platform Owners and CIOs to build AI governance frameworks that maximize ROI and minimize compliance risk.

Whether you're just getting started with Now Assist or scaling across your enterprise, a structured approach to consumption governance is the difference between AI as a cost center and AI as a value driver.

This article is part of an ongoing series on ServiceNow AI adoption. [Follow me on LinkedIn][LINK_LINKEDIN] for regular updates on Now Assist, AI Agents, and platform governance best practices.

Kostya Bazanov, Managing Director, Mar 04, 2026

Eager to take the next step? Contact us today!

Privacy policy (GDPR)

* Required fields

Submit