ServiceNow Now Assist Workflow Automation Review: Capabilities, AI Features, and Costs

ServiceNow Now Assist is generative AI built natively into the Now Platform (the ServiceNow AI Platform) that converts plain-language prompts into Workflow Studio flows and playbooks for IT, HR, and customer service teams. Now Assist also runs subflows from Virtual Agent conversations and supports AI agents that plan and execute multistep tasks under human oversight. The verdict is that Now Assist suits teams already standardized on ServiceNow with clean data, and it is a harder sell where CMDB quality is weak or a metered AI bill is hard to forecast.

TL;DR

• Kissflow’s review finds that Now Assist drafts flows and playbooks from plain-language prompts in Workflow Studio, and a person reviews each draft before it goes live
• Kissflow’s review finds that Now Assist usage is metered in units called assists, which makes the AI line harder to forecast
• Kissflow’s review notes that licensing is in flux, with Rezolve reporting Foundation, Advanced, and Prime tiers succeeding older bundles since April 2026
• Kissflow’s review finds that Now Assist fits ServiceNow-standardized IT teams whose CMDB and CSDM data are already clean

Your ServiceNow account team has proposed moving the instance onto the new AI-native tiers. The decision in front of you is whether Now Assist earns that uplift for workflow automation. This review draws on ServiceNow’s documentation and demo videos, consultant write-ups, and practitioner reviews. It covers what Now Assist builds, what each AI feature does, what consumes assists, where the drawbacks sit, and which teams should buy. The final section covers where business-owned approval work fits alongside ServiceNow.

ServiceNow Now Assist workflow automation capabilities

Five capabilities carry the workflow automation case: flow generation, playbook generation, conversational subflows, AI agents, and governance through AI Control Tower. Flow generation is the capability the buying decision turns on.

Natural language flow generation

Natural language flow generation turns a described trigger, data, actions, and logic into a draft flow in Workflow Studio or Flow Designer. ServiceNow says the draft is assembled from out-of-the-box actions, retrieval-augmented generation (RAG), and data pills, and that the Now LLM applies established practices and common patterns. Typing a hashtag in the prompt points it at a specific table.

Flow generation does not hand you a finished automation. It produces an outline that you review, edit, and test before activation.

Here is an example prompt: “When a laptop replacement request item is approved, create a fulfillment task for the asset team and email the requester.” That prompt should return a trigger, a create-record step, and a notification for you to check.

Playbook generation

Playbook generation applies the same prompt-driven drafting to playbooks. A process owner describes the stages of a multistep case, and Now Assist proposes the structure. The draft reflects common patterns, so the owner still has to add the exceptions their team actually handles.

Conversational subflows through Virtual Agent

Conversational subflows let a chat request trigger a workflow. The LLM reads each input description on the subflow, matches what the user typed to those inputs, and prompts for anything missing. Precise input descriptions matter because they are what the model matches against.

AI agents and multi-agent collaboration

AI agents take on an entire task. Aelum Consulting describes AI agents as software that can take action, make decisions, and complete workflow tasks. Autonomy moves the checking work to a different point in the process. Someone still has to review what the agents do.

Governance with AI Control Tower

AI Control Tower governs AI agents, models, and workflows. ServiceNow’s site lists it as the Super Intelligence Control Tower, which connects strategy, governance, management, and performance for AI.

What you need before the first build

Before the first build, the instance and the people building on it need four things in place:

• Your subscription must carry a Now Assist entitlement, and the licensing section below explains how the tiers are named.
• Now Assist for Creator must be active on the instance.
• Each admin or developer who will build needs the now_assist_creator role (some documentation writes it now.assist.creator).
• The tables your flows will touch need clean, related records, including CMDB data mapped to CSDM.

How to generate a flow with Now Assist

1. Open Workflow Studio, select New, then select Flow.
2. Choose the Build with Now Assist tab.
3. Describe the trigger, data, actions, and logic in plain language, and add a hashtag to name a specific table.
4. Select Generate flow preview and read the outline step by step.
5. If a step is missing, update the Now Assist directions and select Regenerate preview.
6. Select Save and edit flow, then finish, test, and activate the flow in the designer.

ServiceNow AI features at a glance

ServiceNow’s AI features fall into two families: generative AI under Now Assist, and classic machine learning such as Predictive Intelligence. The catalog below lists the features in each family.

• Now LLM: ServiceNow’s large language model is trained and tuned on business workflows, and it powers the generative skills.
• Generative text: Now Assist writes case summaries, drafts knowledge articles, and briefs a swarm team on an open issue through Proactive Swarming.
• Text-to-code: Developers describe a script in plain language, receive a suggestion, and review it before use.
• AI Search: Search returns answers based on what the user means as well as the keywords typed.
• Predictive Intelligence: Machine learning trained on historical records fills in category, subcategory, and assignment group, then routes the record.
• Virtual Agent: The conversational front end, also available in Microsoft Teams and Slack, moves past rigid decision trees to natural language understanding.
• AI Agent Studio: ServiceNow’s product pages describe it as the place where teams build and customize AI agents.
• Context Engine: ServiceNow’s April 2026 newsroom release describes it as connecting the relationships, policy, and decision history behind every AI agent decision.

Every generative feature on this list requires Now Assist licensing, an activated plugin, and assigned roles before users see it.

How ServiceNow Otto reorganizes the same features

ServiceNow’s newer materials lead with ServiceNow Otto, which its 2026 product pages call the AI assistant for work. Per those pages, Otto brings Now Assist and Moveworks under one assistant. In that Otto-led order, the assistant comes first. Agents built in AI Agent Studio come next.

How Now Assist is licensed and what consumes assists

Now Assist is a premium uplift, and its usage is metered in units called assists. ServiceNow publishes no prices. Its pricing page routes buyers to a quote (as of 2 November 2026).

ServiceNow tiers that include Now Assist

ServiceNow introduced new AI-native tiers in 2026, so check which tier your quote names.

• According to Rezolve, ServiceNow has shipped three AI-native tiers since April 2026, called Foundation, Advanced, and Prime, with Now Assist bundled. Rezolve also reports that legacy SKUs reached the end of sale on 1 July 2026.

Get the tier confirmed in writing on your quote.

Activities that consume assists

The mechanism matters more than any list price. Data quality also becomes a budget line.

Pros and cons of Now Assist for workflow automation

The strengths come from Now Assist being part of the platform. The weaknesses come from what it needs around it: clean data, licensing, setup effort, and reviewers.

Pros:

• Native integration: Now Assist reads the tables, roles, and workflows the instance already runs, so generated flows land inside existing processes.
• Speed: Developers start from a generated draft, which shortens low-code builds.
• Coverage across ITSM, HRSD, and CSM: The same generative skills apply to IT, HR, and customer service workflows on one platform.

Cons:

• Cost and licensing: Metered assists and a 2026 tier change make the AI line harder to forecast.
• Data readiness: Generated flows and agents are only as reliable as the CMDB and CSDM data beneath them.
• Implementation complexity: Integration Hub connections to outside systems add their own setup. Plan that effort alongside the license.
• Human oversight: Every generated flow, playbook, and script needs a reviewer, which moves effort from building to checking.

Verdict: who Now Assist is for

Buy Now Assist for workflow automation if ServiceNow is where your IT organization already builds every week. Skip it if the data or the budget model is not ready.

A fit for:

• IT teams running ITSM on ServiceNow with developers who already work in Workflow Studio.
• Organizations that maintain a CMDB mapped to CSDM.
• Buyers moving onto Foundation, Advanced, or Prime anyway, where Rezolve reports Now Assist arrives bundled.

Not ideal for:

• Smaller teams with simple departmental workflows, which may find the platform heavier than they need.
• Teams with an incomplete CMDB, where agents burn assists on retries.
• Approval work owned by business teams whose builders will never hold a ServiceNow creator role.

Where business-owned workflows fit alongside ServiceNow

ServiceNow stays in the system for IT service work. Some business-owned approvals and requests sit outside it.

Take a marketing statement-of-work approval that needs legal, finance, and the budget owner to sign off before a purchase order reaches SAP. That work is owned by a business team, crosses three functions, and rarely justifies a ServiceNow developer’s sprint. Procurement approvals, onboarding that spans HR, IT, and finance, IT service requests, and finance approvals around Oracle, Salesforce, and Workday sit in the same gap.

Kissflow is an AI-powered enterprise application platform where business teams and IT build, automate, and govern their core work through business logic, not code. Its AI Builder is generally available as of May 2026. Give it a prompt or upload a requirements document, and it returns a working app that the owner then refines through conversation. What it generates are blueprints, structured metadata that describes the business logic, and a developer can open each one as a standard component.

Every action is timestamped and attributed in one audit log, whether a person or the AI made the change. AI features are included in the subscription at no extra cost. Kissflow’s compliance list covers SOC 1 Type II, SOC 2 Type II, SOC 3, ISO/IEC 27001, HIPAA, GDPR, and CCPA. For a fuller side-by-side, see Kissflow vs ServiceNow for how IT keeps control of what business teams build.

Frequently asked questions

Can you activate a flow that Now Assist generates without editing it?

Not always. When a request cannot be matched to an available action or subflow, Now Assist adds a placeholder step, and ServiceNow does not let you activate the flow until you replace or delete each placeholder. Review the generated outline either way.

Is ServiceNow a workflow tool or an ITSM tool?

It is both. ITSM is the core that ServiceNow keeps, and Now Assist adds prompt-driven flow building on top of it. Buyers should evaluate Now Assist as a workflow capability on the Now Platform that extends past the IT help desk.

Do test and cloned instances draw from the same assist pool?

Do not assume test and cloned instances are free. Ask your ServiceNow account team to confirm in writing whether they draw from the same assist allocation as production before you plan a large testing program.

Book a demo to see how Kissflow generates governed apps for the approval work around your core systems.