Now Advisory · Buyer side guide · 2026 edition
Now Assist Assists Explained: A Buyer Side Guide
Now Assist assists explained from the buyer side, what an assist actually is, how metering works, why large agentic actions consume far more, and how to forecast consumption before it drives your renewal, with benchmark data from real enterprise renewals.
Section 01Now Assist assists explained: what an assist is
With Now Assist assists explained simply, an assist is the unit ServiceNow meters AI usage in: a single chargeable AI action on the platform, whether a summarisation, a generated reply, a search, or a step taken by an agentic workflow. Understanding the unit is the first step to controlling the cost, and on the buyer side it is where most consumption surprises begin. This guide sets out how assists work, with benchmark data from real enterprise renewals.
We are independent advisors with no vendor partnership and nothing to resell, so the angle is plain. AI is bundled across the 2026 tiers, but the assists that power it are metered, which means consumption, not licensing, is now the line that moves a renewal. For the wider context, start with our guide to Now Assist pricing. Figures below are typical negotiated ranges based on benchmark observations, not official list prices.
The key point is that an assist is not a flat per user cost. It is a metered event, and the number of events your estate generates depends on how the platform is used, not just how many people are licensed.
Section 02How assist metering works
Assist metering counts chargeable AI actions against an entitlement included with your tier. Each action consumes from that allowance, and once consumption passes the included volume, additional assists are billed as overage or drawn from a top up.
The mechanic matters because it decouples cost from headcount. Two enterprises with identical fulfiller counts can generate very different assist bills depending on how heavily they automate, summarise and deploy agentic workflows.
This is why a renewal that counts only seats and modules now misses a fast growing line. The metered assist allowance and its consumption have to be tracked separately, which is the foundation of our Now Assist consumption guide.
Section 03Simple prompts versus agentic actions
Not all assists cost the same in practice, because not all actions consume the same volume. A simple prompt, a single summarisation or a generated reply, draws a small, predictable amount. A large agentic action, where the platform plans and executes a multi step workflow, consumes materially more.
This is the single most important thing for buyers to understand. As estates move from simple AI features toward agentic automation, consumption per action rises sharply, and a forecast built on simple prompt volumes will understate the real bill.
Model the two separately. The mix between simple prompts and agentic actions, not the raw user count, is what determines whether your assist consumption stays inside entitlement or runs into overage.
Section 04Where assist consumption accumulates
Consumption accumulates in predictable places. High volume queues where summarisation is enabled, automated workflows that fire on every record, and agentic agents deployed across multiple processes each generate assists continuously, often without a human initiating each one.
The risk is that consumption grows quietly. An automation that seemed inexpensive at pilot can dominate the assist line once it runs at production scale, and an agentic workflow expanded to new processes multiplies its draw.
Mapping where assists are generated is the buyer side equivalent of mapping shelfware. It shows which workflows drive the cost and where consumption can be tuned before it drives the renewal.
Section 05Overage and top up charges
When consumption passes the included allowance, the assist line moves into overage, and overage triggers top up charges. These top ups are where an unmanaged assist line becomes expensive, because they are billed above the rate of the bundled entitlement.
The exposure is asymmetric. A single busy quarter, a new agentic rollout, or a seasonal spike in volume can push consumption over the line and trigger top ups that were never budgeted. The vendor tracks this in real time; the buyer often meets it at the invoice.
Managing overage is therefore a contract question as much as an operational one. An overage grace band and a pre priced top up rate, negotiated at renewal, turn a penalty into a managed line. Our Now Assist overage guide sets out the protections.
Section 06The 2026 model and bundled AI
The 2026 commercial model bundles AI across Foundation, Advanced and Prime, which can read as if AI is now free. It is not. The capability is included; the consumption is metered. The distinction is the whole of the buyer side story.
Bundling removes the separate AI license line that buyers used to negotiate, and replaces it with a consumption line that scales with usage. For estates that adopt agentic workflows heavily, this can be a larger and less predictable cost than the old per user AI add on.
The implication is that the renewal conversation shifts from licensing AI to governing consumption. The tier sets the allowance; your usage sets the bill. Our Now Assist pricing guide treats the metered line as the central negotiation, not a footnote.
Section 07Forecasting consumption before renewal
Forecasting is the buyer's main defence. Take current assist consumption, split it between simple prompts and agentic actions, and project it forward across the planned automation roadmap. The result is a consumption forecast the renewal can be negotiated against.
A good forecast accounts for growth that has not happened yet. If agentic workflows are planned for new processes, their consumption belongs in the forecast even before they go live, because the renewal will set the allowance for the term ahead.
Without a forecast, the assist line is negotiated on the vendor's numbers, which is exactly the asymmetry to avoid. With one, you can size the entitlement to genuine need and negotiate the protections around it.
Section 08Protecting the assist line in a contract
Three protections matter most. A right sized assist entitlement, set to your forecast rather than an upsell, stops you paying for headroom you will not use. An overage grace band absorbs normal variation without triggering top ups. And a pre priced top up rate caps the cost of genuine growth.
Add a true up mechanic that bills forward from an agreed count rather than retroactively, so a busy quarter does not produce a backdated charge. Each of these turns the metered line from an open exposure into a managed cost.
These protections cost nothing today and save the most later. Securing them is the core of our Now Assist consumption advisory, which models the metered line into the renewal rather than discovering it at the anniversary.
Section 09Assists in the renewal runway
The assist line is easiest to control when the work starts early. Four quarters out, baseline current consumption and its mix. Two quarters out, build the forward forecast against the automation roadmap. One quarter out, negotiate the entitlement and the overage protections into the renewal.
Held this way, consumption stops being an anniversary shock and becomes a forecast you negotiate around. Each stage sharpens the number you bring to the table.
An independent advisor who has modelled assist consumption across many enterprise renewals shortens the work, because the pattern of where consumption accumulates is already known. The aim is a right sized assist entitlement, protected against overage, set before the renewal rather than after.
The same discipline applies whether the estate runs a handful of AI features or a full agentic program; only the scale of the forecast changes, not the method behind it.
Section 10Reading consumption against benchmark
An assist forecast is only useful read against benchmark. Consumption per fulfiller, the share of agentic actions, and the overage rate all vary across comparable enterprises, and only a benchmark shows whether your numbers are reasonable or inflated.
Score the forecast and the proposed entitlement against range for enterprises of your size and automation maturity, then negotiate the entitlement and the top up rate where they sit furthest from it. A benchmark turns a consumption estimate into a position.
Where the account team prices the assist line as an opaque allowance, that opacity favours the vendor, because it hides whether you are over provisioned or under protected. Insist on the consumption picture and its pricing in one benchmarked view, so the metered line is negotiated on evidence rather than trust.
Read in full, the assist line stops being a mystery on the invoice. It becomes a forecast you sized, an entitlement you negotiated, and an overage rate you capped, each one a number you can defend.
FAQFrequently asked questions
What is a Now Assist assist?
An assist is the unit ServiceNow meters AI usage in: a single chargeable AI action such as a summarisation, a generated reply, a search, or a step in an agentic workflow. Under the 2026 model AI is bundled across the tiers but the assists that power it are metered, so consumption rather than licensing drives the cost.
Why do agentic actions consume more assists?
A simple prompt draws a small, predictable amount, while a large agentic action plans and executes a multi step workflow and consumes materially more. As estates move from simple AI features toward agentic automation, consumption per action rises sharply, so a forecast built on simple prompt volumes understates the real bill.
What happens when assist consumption exceeds entitlement?
Consumption above the included allowance moves into overage, which triggers top up charges billed above the bundled rate. A single busy quarter or a new agentic rollout can push consumption over the line, so an overage grace band and a pre priced top up rate, negotiated at renewal, are the key protections.
Are these assist figures official ServiceNow prices?
No. All ranges are typical negotiated figures based on benchmark observations across real enterprise renewals, used as internal leverage rather than published as official list prices.