By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

ServiceNow President & CPO on Why AI Kills the Companies That Don't Transform | Amit Zavery | E305

In this episode of The Product Podcast by Product School, Carlos González de Villaumbrosia sits down with Amit Zavery, President and Chief Product Officer at ServiceNow.

Amit ZaveryguestCarlos González de Villaumbrosiahost

Jul 22, 202654mWatch on YouTube

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

CHAPTERS

Jump to publisher timestamps

Chapter summaries

Why the “end-to-end AI orchestrator” pitch fails—and why companies must transform fast
Amit Zavery argues that any company claiming it will be the single end-to-end orchestrator for the enterprise is headed for failure. He frames AI transformation as urgent, emphasizing that delaying adoption while “putting tools together” creates existential risk. He contrasts superficial AI efforts with real “action” workflows that require heavy lifting behind the scenes.
- •Says the idea of one company being the enterprise’s end-to-end orchestrator is a fallacy and predicts failure for those who claim it
- •Warns that lack of clear ROI and a “spare part approach” leads many AI initiatives to fail
- •States ServiceNow’s AI business is tracking to $1.5B and that products have become AI-native
- •Argues companies die when they delay AI adoption for months while they attempt to assemble toolchains
- •Explains the difference between retrieving information and taking real action via workflows (example: onboarding a new employee end-to-end)
Meet Amit Zavery and what this conversation will cover
The host introduces Amit Zavery and positions ServiceNow as a workflow platform operating at massive scale. He previews the discussion topics: how markets misunderstand AI winners, why orchestration isn’t a winner-take-all game, what causes enterprise AI projects to fail, and how agents change interaction from UI to action.
- •Introduces Amit Zavery as President and Chief Product Officer at ServiceNow
- •Host describes ServiceNow’s scale and notes stock performance versus AI business growth
- •Previews discussion: AI winners vs losers, orchestration fallacy, and enterprise AI failure modes
- •Sets up themes of agents shifting interaction away from traditional UI toward execution
- •Transitions into the interview with a focus on product strategy and transformation
Combining President, CPO, and COO: aligning strategy, build, and execution
Amit explains how combining product leadership with operational responsibility reduces fragmentation and speeds decisions. He describes the CPO scope as product strategy and product engineering, while COO scope is operationalizing delivery and value at scale. He says the unified role improves alignment across engineering, experience, and operations.
- •Defines CPO responsibilities: product strategy, building products, and customer capability delivery
- •Defines COO responsibilities: operationalizing value, scaling efficiently, and delivering faster
- •Says end-to-end oversight reduces handoffs and enables faster decision-making
- •Notes the company’s CEO structured the role to increase efficiency and alignment
- •Emphasizes moving fast while delivering well for customers
ServiceNow at scale: revenue, margins, headcount, and R&D emphasis
Amit outlines ServiceNow’s business size, growth, profitability profile, and employee count. He emphasizes that the company invests heavily in R&D, calling it a product-led and engineering-driven culture with a customer-first mindset. The host uses this as context for later questions about market perception and AI investment.
- •Shares revenue level and growth rate and describes the business as scaled and fast-growing
- •Gives profitability indicators (free cash flow margin and operating margin)
- •Estimates total employees and the size of the product/engineering R&D organization
- •Says over half of investment goes into building products for customers
- •Describes the culture as product-led, engineering-driven, and customer-first
Why the market struggles to identify AI winners—and how ServiceNow thinks about it
Amit says investors are still sorting out whether AI is a tailwind or headwind for established software companies, and many are lumped together. He argues ServiceNow is differentiated as a platform company and highlights AI traction and product changes. He stresses focusing on controllables and long-term strategy rather than short-term stock moves.
- •Explains investor uncertainty during technology shifts leads to “same bucket” treatment for software companies
- •Claims ServiceNow is a different kind of platform versus more vertical-stack providers
- •Highlights AI traction and that products have become AI-native
- •Says companies that don’t transform will get killed, but incumbents with domain and customer base can win if they transform
- •Emphasizes controlling execution and not overreacting to short-term stock price changes
Balancing investor expectations with long-term platform conviction
Amit describes how he thinks about investor stakeholder needs without letting short-term sentiment override long-term bets. He emphasizes sustainable growth and building the right products over quick fixes that require constant rework. He argues product quality depends on choosing what to build and delivering it in a usable way, with constant customer feedback and strong judgment.
- •Says market cap doesn’t directly dictate plans, but investor communication and efficiency matter
- •Connects new roadmap investments to customer value and long-term investor value
- •Advocates sustainable growth over short-term moves that create rework and dissatisfaction
- •States real product quality is driven by understanding what to build, not just engineering speed
- •Stresses continuous feedback loops plus “gut feel” for where to invest and monetize long term
AI transformation vs past shifts: tech stack, commercial models, and faster cycles
Amit compares AI’s transformation dynamics with earlier transitions like client-server to web, and web to cloud. He argues each shift changes technology, but AI also accelerates change frequency and pushes pricing toward consumption. He frames AI as requiring both technical adaptation (models and reasoning integrated into workflows) and commercial adaptation (usage-based expectations).
- •Says fundamentals stay the same: build great products customers love
- •Compares past tech shifts: client-server to browser-based, then to cloud hosting and subscriptions
- •Describes AI as changing architecture (LLMs, reasoning) and integrating into deterministic workflows
- •Notes pricing pressure shifts toward usage/consumption-based models
- •Highlights time compression: more frequent delivery and less time for long planning cycles
From scattered pilots to “pacesetters”: platform-first vs spare-part AI
Amit explains what he sees in enterprise AI maturity: many companies encourage “do something with AI” without coherent guidance, leading to scattered efforts and unclear ROI. He labels this as a spare-part approach—cobbling pieces together while the underlying tech changes too fast to keep up. He says pacesetters take a holistic, structured approach, pick high-value projects, and treat AI as both a cultural and technology shift.
- •Describes common failure mode: scattered projects without coherent expertise or ROI clarity
- •Calls out the “spare part approach” of stitching tools together as a frequent cause of failure
- •Says pacesetters apply platform-first thinking and provide structure and expertise
- •Emphasizes selecting projects with clear value and staffing them with the right people
- •Notes successful companies regularly inspect what’s working and fix what isn’t quickly
Closing the data gap without pretending data will live in one place
Amit argues enterprise data will remain distributed, making “everything in one place” unrealistic. He says the key is understanding the data landscape and creating a strong integration layer that avoids painful, repeated data migration. He describes ServiceNow’s approach as connectivity plus a semantic layer, augmented by knowledge/context graphs to enable insight-to-action while data stays federated.
- •States enterprise data is fragmented and will remain distributed
- •Says the first step is mapping the data landscape: where it is, ownership, and systems
- •Advocates integration/federation approaches rather than moving and re-rationalizing data constantly
- •Describes a semantic layer enabling a virtual view across silos
- •Links semantic/graph layers to enabling “insight to action” without heavy data relocation
Access shifts beyond UI: enabling agents, copilots, and action execution
Amit says relying on a single user interface is a fallacy; systems must support multiple access patterns for humans and agents. He describes an interface abstraction approach and emphasizes the real unlock: moving from answering questions to executing complex multi-step actions. He uses employee onboarding as an example of action requiring orchestration across many systems.
- •Says access must work for humans and for agents/coplots, not only through UI
- •Describes the need for multiple interfaces and integrations (including agent-to-agent patterns)
- •Mentions an interface abstraction layer to make connecting into products and data easier
- •Distinguishes information retrieval from action execution as the key leap
- •Gives an example of end-to-end onboarding as a complex action requiring back-end orchestration
The scalability gap: why agent volume changes platform requirements
Amit argues scalability becomes critical as interaction shifts from human-driven usage to 24/7 agent-driven execution. He says building from scratch won’t solve spawning and running many agents securely and reliably. He highlights the need for platform-level considerations like SLAs, authorization, distributed architecture, failover, and performance evolution.
- •Says DIY agent building doesn’t address real scalability needs at enterprise level
- •Notes usage patterns shift: agents operate continuously, increasing volume and load
- •Emphasizes platform needs: authorization, volume controls, SLAs, and security
- •Calls for distributed architecture across environments and zones for resiliency
- •Frames scalability as an evolving architectural requirement as agent usage grows
Governance that enables adoption: built-in visibility, control, and fast response
Amit calls governance the biggest barrier to AI adoption but argues it must be embedded in the platform rather than bolted on. He describes governance as visibility into what’s happening, where, and with what risks, including lifecycle and access reporting for human and non-human actors. He warns against governance as a stalling tactic, advocating lightweight structure that lets teams move while staying able to stop issues quickly.
- •Calls governance (and security) the biggest barrier to AI adoption in large enterprises
- •Argues governance must be a core platform feature: visibility, reporting, and lifecycle tracking
- •Describes needs: control, risk management, and observability across AI usage
- •Rejects policy-for-policy’s-sake; says guardrails should not block innovation
- •Warns that delaying adoption for months “until tools are put together” is where companies die
Centralized vs federated AI operating models—and ServiceNow’s COE approach
Amit says there’s no single right governance structure; it depends on company culture, but clarity and coordination are essential. He describes ServiceNow’s model: every department needs AI capability, supported by departmental centers of excellence, with IT acting as a central clearing house when needed. He emphasizes having enough structure to manage risk without over-optimizing the org chart.
- •Says both centralized and federated models can work depending on culture
- •Warns against mixed structures that don’t coordinate, leading to chaos
- •Describes ServiceNow’s approach: AI capability in every group rather than only central ownership
- •Explains departmental COEs with senior leadership, collaborating in a federated way
- •Notes IT can provide central support as needed to keep projects on track and moving fast
Proving ROI: choosing outcome metrics that match what the business values
Amit argues ROI measurement depends on what each company values—efficiency, speed, or financial impact. He lists outcome-oriented metrics such as time to value, time to resolve, automation hours saved, and process simplification. He cautions against fixating on a single input metric like token spend and says leadership ultimately wants to see top-line and bottom-line improvements.
- •Says ROI is company-specific and tied to what the business values
- •Mentions metrics used with customers: time to value, time to resolve, and time to fix issues
- •Highlights efficiency measures like hours saved through automation and process simplification
- •Acknowledges input metrics like spend and token usage but urges holistic evaluation
- •Says credibility comes from visible improvements that flow into revenue and efficiency outcomes
Getting customers started: AI Control Tower, forward-deployed help, and autonomous workers
Amit explains two big blockers to agentic adoption: governance/security concerns and knowing where to start. He describes AI Control Tower as a way to centralize visibility into AI systems, cost, lifecycle, access, and observability. He also explains how ServiceNow helps customers adopt through forward-deployed engineers, out-of-the-box agentic playbooks, and a marketplace concept of autonomous workers that perform complete job roles end-to-end.
- •Identifies two adoption blockers: governance/security anxiety and uncertainty about where to start
- •Describes AI Control Tower as providing visibility into cost, lifecycle, versioning, access/security, and observability
- •Explains use-case starter help via solution consultants and forward-deployed engineers implementing early workflows
- •Introduces autonomous workers as digital equivalents of job roles that execute end-to-end tasks faster and continuously
- •Positions the goal as solutions and outcomes, not customers having to build agents from scratch
Platform expansion bets: security/OT exposure management and customer-service-led CRM
Amit explains why ServiceNow is expanding into cybersecurity and CRM areas that align with its workflow orchestration strengths. In security, he describes serving CISOs through post-breach workflows and expanding into identity governance for non-human identities, plus OT/IoT exposure management to protect operational environments. In CRM, he clarifies the focus is not “CRM equals sales,” but customer service and other complex orchestration areas like CPQ and order management, with migration tooling and partner support to reduce switching friction.
- •Rejects the ‘single orchestrator’ dream while positioning ServiceNow as an orchestration layer among many
- •Says ServiceNow already serves CISOs through security workflows integrated with many tools and has a sizable security business
- •Explains expansion into non-human identity governance and access graph needs as agent volume grows
- •Describes OT/IoT device exposure management as critical due to operational disruption risk
- •Defines CRM focus as customer service and complex orchestration (CPQ, order management), not broad marketing; describes migration and partner support to help customers switch

Publisher timestamps

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.