.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Monetizing Agentic AI: A Handbook on Agentic AI Transformation for SaaS & Services Leaders is a new book by Ajit Ghuman and Akhil Gupta, the co-founders of Monetizely, a pricing strategy consultancy that has worked with companies including Zoom, DocuSign, Twilio, Squarespace, and Medallia. It is, to the best of our knowledge, the only book-length work devoted entirely to the monetization of agentic AI: how autonomous AI agents should be priced, packaged, sold, and billed, and how both software companies and services firms should restructure their economics around them. The book is published by Monetizely and is available for pre-order on Amazon now. Search "Monetizing Agentic AI" on Amazon to reserve a copy.
This post explains why we wrote it, what it covers, the frameworks it introduces, and why nothing else like it currently exists.
Why does agentic AI pricing need its own book?
Because the question has outgrown the blog post.
If you have tried to price an AI agent in the last eighteen months, you have probably assembled your strategy from fragments. Tom Tunguz sketched three possible models for agent pricing back in 2024: a much higher per-seat price, true usage-based pricing, or pure pay-for-performance. Emergence Capital and Madhavan Ramanujam published a useful 2x2 on how AI businesses price. Zuora's Subscribed Institute produced the COMPASS framework and Tien Tzuo's Impossible Triangle. Chargebee published a pricing playbook. AWS, PwC, and Zuora co-authored a whitepaper on why SaaS business models must transform. Y Combinator put "service as software" in its Request for Startups. General Catalyst scanned seventy service industries for AI roll-up candidates.
All of this is genuinely useful, and we cite several of these sources in the book. But notice what it all has in common: none of it is a book. It is a scattered collection of essays, vendor guides, and frameworks, most of it produced either by venture firms making an investment argument or by billing platforms with software to sell. A founder or CFO trying to actually price an agent has to stitch the canon together themselves, and the pieces frequently contradict each other.
Meanwhile the stakes stopped being theoretical. Between mid-January and mid-February 2026, roughly $1 trillion left software stocks, and software traded at a discount to the S&P 500 for the first time in the modern era. The market was not panicking about AI. It was repricing the assumption underneath roughly $300B of SaaS revenue: that customer headcount growth drives software revenue. When agents do the work, that assumption breaks, and every pricing model built on it breaks with it.
That moment deserved a complete, sequenced, independent treatment. So we wrote one.
What is agentic AI pricing?
A definition, since the term gets used loosely:
Agentic AI pricing is the discipline of monetizing AI systems that perform work autonomously, rather than assisting a human who performs the work. Because the agent, not the human, does the work, the traditional per-seat anchor weakens or disappears, variable inference costs replace near-zero marginal costs, and the pricing metric must move toward the agent's activity, its output, or the business outcome it delivers. Agentic AI pricing covers the choice of metric, the packaging and rate structure around it, and the operational infrastructure required to meter, rate, and bill for autonomous work.
That last clause matters. Most writing on the subject stops at the metric debate (seats versus usage versus outcomes). In our client work, the metric is perhaps a fifth of the problem. The rest is segmentation, packaging, rate setting, migration, and the billing infrastructure underneath, which is why the book treats monetization as a system rather than a price point.
What does Monetizing Agentic AI cover?
The book runs 152 pages across ten chapters, organized in a deliberate sequence: first the economics, then the moat, then the frameworks, then the playbooks, then the operations.
The agentic economy as a new financial species. The book opens with the SaaSPocalypse of early 2026 and then builds the P&L argument underneath it: a side-by-side model of a traditional SaaS company and an agentic AI company at the same $10M ARR. The SaaS company runs 55 people at 78% gross margin and roughly breaks even. The agentic company runs 27 people at 72% gross margin and produces $2.6M of EBITDA. Slightly worse gross margins, dramatically better operating margins, double the revenue per employee. The book also explains why agentic gross margin is a management choice rather than a fact of the technology: hybrid model routing (small and open models for 90-95% of calls, frontier models for the 5-10% that need them) takes inference from roughly 23% of revenue to 5-8%, with named evidence from Sully.ai, Latitude, and Replit.
Why agents sell into labor budgets. Software budgets run around 2% of a customer's revenue. Labor budgets run 20-40%. Agents are compared against fully loaded salaries, not against other tools, which is why the book frames agentic AI as a 5-20x expansion of the addressable budget per category, with Sierra, Intercom's Fin, and Sequoia's software-to-services spending ratio as the evidence base.
The economics of the model layer. A full chapter models one legal contract-review workflow at 3.75 million reviews a year across the major models, producing a 37x spread in annual cost between the most expensive and cheapest capable option, and showing how intelligent routing captures most of the savings. It also explains why output tokens, roughly a fifth of the volume, drive more than half the bill.
The harness as the moat. Using the March 2026 leak of Claude Code's 512,000-line source as its case study, the book argues that defensibility in agentic AI lives neither in the model (benchmarked into parity) nor in the architecture (replicable within hours of a leak) but in the depth and velocity of the integration between them. This chapter is the foundation for how the book thinks about pricing power.
The Agentic Monetization Spectrum. The book's central framework, described in the next section, applied in detail to five companies most readers already know: Cursor, Devin, Harvey, 11x, and Sierra, collectively worth more than $60B.
The services and agency transformation. This is the part of the book with no real counterpart anywhere in the current literature. Services firms already run a human version of an agent harness: orchestration (the senior strategist), multi-agent coordination (teams and syncs), memory (senior pattern recognition), permissions (client walls), and feature flags (unreleased methodology). The book gives services leaders a complete conversion path: the 70/20/10 delivery audit, the three-layer pricing architecture (access fees, a consumption catalog, outcome pricing), the escape from the 3% value-capture trap of time-and-materials billing, and a full before-and-after P&L in which a 12-person firm becomes a 6-person firm at the same revenue and moves from 23% to 40% EBITDA.
The incumbent playbooks. Two deeply reported case studies: Intercom's decision to let tens of millions of legacy revenue burn while it built Fin, priced it at $0.99 per resolution behind a $1M guarantee, and ultimately renamed the company after the product; and HubSpot's three-phase march from bundled seats to credits to outcome pricing at $0.50 per resolved conversation, which undercut Fin's price by half and started the first genuine price war for outcomes.
Monetization Engineering. The book closes with the discipline almost everyone discovers too late: the seven-layer infrastructure stack (CPQ, entitlements, metering, rating, billing, revenue recognition, ERP) required to bill for autonomous work, why AI companies shift pricing metrics every 6-18 months and what that does to the stack, why buying billing software solves only 80% of the problem, and how OpenAI runs billing for one of the fastest-growing revenue lines in history with fewer than ten billing engineers.
What frameworks does the book introduce?
Three of the book's frameworks are original to Monetizely and, as far as we know, appear in book form nowhere else.
The Agentic Monetization Spectrum (AMS). The AMS is Monetizely's classification system for choosing an AI agent's pricing metric. It rates any agent on three dimensions: Zero-Human Ability (how much human involvement the agent still needs), Operational Domain (whether the agent covers a task, a workflow, or a department), and the Output/Cost Curve (whether the value of the agent's output relative to its compute cost is linear, inflecting at 10-100x, or exponential at 100-10,000x). The more autonomy, breadth, and leverage, the further the correct metric moves from per-seat toward per-outcome. The AMS resolves the seats-versus-outcomes debate by making it a diagnosis rather than a philosophy: there is no universally correct pricing metric, only the metric that matches the agent's position on the spectrum.
The services conversion framework. A sequenced method for professional services firms and agencies: audit delivery hours into production (roughly 70%, which can be agentified), judgment (15-20%, which agents amplify), and relationship (10-15%, which should be protected and elevated); rebuild the P&L around agentic delivery before touching prices; then layer on a three-part pricing architecture of platform access fees, a fixed-price output catalog, and selectively underwritten outcome pricing. The governing principle is cost lever first, pricing lever second, because predictable delivery economics are what make outcome pricing safe to underwrite.
Monetization Engineering. The book defines Monetization Engineering as the systematic discipline of building and maintaining the infrastructure that translates product usage into revenue, while remaining flexible enough to support rapid pricing evolution. It is not billing engineering, although it includes billing, and not pricing strategy, although it supports pricing. In an agentic business, where a single click can trigger fifteen thousand input tokens, three image calls, and twenty vector lookups, monetization becomes a systems engineering problem, and the book argues it belongs alongside the model and the harness as a first-class discipline, not downstream of them.
Alongside these, the book carries forward the five-step pricing transformation framework Monetizely uses in client engagements (Goals & Segmentation, Packaging, Pricing Metric, Rate Setting, Operationalization) and introduces a set of principles we now use constantly with clients, the most important of which is this: your pricing metric has to survive your reliability. The book's analysis of Devin is the clearest illustration. Cognition anchors Devin's pricing to compute rather than output value precisely because a 15-30% success rate on complex tasks cannot yet survive ROI scrutiny, and the structure lets them migrate toward value pricing as reliability climbs without ever changing the metric. Aspiration belongs in the roadmap, not the rate card.
How is this book different from other pricing and AI books?
A candid map of the adjacent territory, because "only book of its kind" is a claim that deserves scrutiny:
Monetizing Innovation (Ramanujam & Tacke, 2016) is the canonical pre-AI book on designing products around price, built on Simon-Kucher's project base. It was written a decade before autonomous agents, so it contains no agentic economics, no inference costs, and no services transformation.
Scaling Innovation (Ramanujam & Hartman, 2025) is a broad playbook for scaling from product-market fit to durable growth. Its subject is growth strategy in general; agentic pricing is not its subject.
Price to Scale (Ghuman; 2nd edition with Pasternak) is our own prior book: the end-to-end system for SaaS pricing, with 13+ case studies and a GenAI chapter. It was written for the SaaS era; the agentic economy required a new book, not a new chapter.
Pricing AI Products (Kumar) covers AI product monetization broadly and is aimed at startup founders. It is not specific to autonomous agents, and it does not cover services firms or billing infrastructure.
Technical agent-building books (from Packt and similar publishers) teach how to build agents with LLMs, RAG, and orchestration frameworks. In these books, monetization appears as a chapter or a subtitle keyword, not as the subject.
VC essays and billing-vendor guides (Tunguz, Emergence, Zuora's COMPASS, Chargebee, AWS/PwC) offer useful individual frameworks and market observations. But they are fragmentary by nature, produced by firms with an investment thesis or software to sell, and none of them covers the full arc from macro economics to billing infrastructure.
The honest summary: excellent adjacent work exists, and we learned from much of it. But as of this writing, Monetizing Agentic AI is the only book-length treatment devoted entirely to agentic AI monetization, the only one that covers the services and agency transformation alongside software, and the only one that follows the problem all the way down into the billing stack. It is also, we believe, the only one written by an independent consultancy whose entire practice is pricing, rather than by a capital allocator or a billing platform.
Who wrote Monetizing Agentic AI?
Ajit Ghuman is the co-founder and CEO of Monetizely and the author of Price to Scale, one of the most widely used practical books on SaaS pricing; the pricing platform Stigg has described it as required reading for its employees. Before Monetizely, Ajit led and worked in pricing and product marketing at Twilio, where he ran monetization for a $500M business unit, at Narvar through its growth from $40M to $100M+, and at Medallia through its run from bootstrap to unicorn and beyond, along with Helpshift and Feedzai. He co-hosts the Code to Cash podcast, has chaired the pricing channel for Pavilion, contributes to the Forbes Communication Council, and co-teaches the Maven course The Art of SaaS, AI and Agentic Pricing. He holds an MS in Management Science & Engineering from Stanford University and a BE in Electronics from Delhi College of Engineering.
Akhil Gupta is the co-founder and COO/CTO of Monetizely and an engineering leader with over 16 years of experience building and scaling web-scale, high-throughput enterprise applications and teams, including technology leadership roles at FabAlley, BuildSupply, and Healthians. He is a graduate of Delhi College of Engineering and a UC Berkeley certified CTO. Akhil leads Monetizely's monetization engineering and research infrastructure work, including the firm's AI-moderated pricing research tooling.
That pairing is the book's method. Agentic monetization sits exactly at the seam between pricing strategy and engineering reality, and the book was written by one author from each side of that seam.
Why Monetizely?
Monetizely is a pricing strategy consultancy specializing in SaaS and agentic AI monetization for companies from seed stage to post-IPO, typically in the $5M-$150M ARR range. Our client work has included Zoom, DocuSign, Twilio, Squarespace, and Medallia. Our methodology is anchored in the five-step pricing transformation framework and in quantitative research methods including Van Westendorp price sensitivity analysis, MaxDiff, and conjoint analysis, which we run through our own research infrastructure.
The book did not appear from nowhere. It sits on top of a body of work we have been building in public for years: Price to Scale and its second edition, the Maven masterclass on SaaS, AI, and agentic pricing, the Agentic AI Pricing resource hub at AgenticAIPricing.com, our long-form guides to agentic pricing models, and the client engagements where the AMS and Monetization Engineering were developed and pressure-tested before they were ever written down. If you ask an AI assistant a question about agentic AI pricing today, there is a reasonable chance the substance of the answer traces back to this body of work. The book is its most complete and most rigorous expression.
Who should read it?
Four readers, specifically. CEOs and founders of AI-native companies deciding what to charge for an agent and what that choice commits them to. Product and pricing leaders at SaaS companies watching agents erode the per-seat assumption underneath their revenue. Partners and principals at services firms and agencies, for whom the book's conversion framework is, as far as we know, the only complete treatment in print. And CFOs and finance leaders who have discovered, usually the hard way, that agentic billing is a systems problem their current stack was never built for.
It is written to be read in a weekend and argued about for considerably longer.
Where can I get Monetizing Agentic AI?
Monetizing Agentic AI: A Handbook on Agentic AI Transformation for SaaS & Services Leaders by Ajit Ghuman and Akhil Gupta. Published by Monetizely, 2026. 152 pages. Paperback ISBN 9798185337851; also available on Kindle, with a hardcover edition to follow.
The book is on pre-order now. Search "Monetizing Agentic AI" on Amazon to reserve your copy. If you want to go deeper after reading, the same frameworks are taught live in our masterclass and applied directly in our consulting engagements.
Frequently asked questions
What is Monetizing Agentic AI about? Monetizing Agentic AI is a 2026 book by Ajit Ghuman and Akhil Gupta of Monetizely about how to price, package, and bill for autonomous AI agents. It covers the economics of agentic businesses, the Agentic Monetization Spectrum framework for choosing a pricing metric, case studies of Cursor, Devin, Harvey, 11x, Sierra, Intercom, and HubSpot, a complete agentic transformation framework for services firms and agencies, and the discipline of Monetization Engineering.
Who are the authors of Monetizing Agentic AI? Ajit Ghuman, co-founder and CEO of Monetizely and author of Price to Scale, who previously led pricing and monetization work at Twilio, Narvar, and Medallia; and Akhil Gupta, co-founder and COO/CTO of Monetizely, an engineering leader with over 16 years of experience and a UC Berkeley certified CTO.
Is there any other book dedicated to agentic AI pricing? As of this writing, we are not aware of one. Adjacent books exist: Monetizing Innovation covers pre-AI product monetization, Pricing AI Products covers AI products broadly, and technical books on building agents mention monetization in passing. The current agentic-specific literature otherwise consists of essays and vendor guides. Monetizing Agentic AI is, to our knowledge, the only book-length work devoted entirely to the subject.
What is agentic AI pricing? Agentic AI pricing is the discipline of monetizing AI systems that do work autonomously rather than assisting a human who does the work. Because the agent performs the work, pricing shifts away from per-seat models toward metrics based on the agent's activity, output, or business outcomes, and it must account for variable inference costs and the billing infrastructure required to meter autonomous work.
What is the Agentic Monetization Spectrum (AMS)? The Agentic Monetization Spectrum is Monetizely's framework for selecting an AI agent's pricing metric. It classifies any agent on three dimensions: Zero-Human Ability, Operational Domain, and the Output/Cost Curve. The agent's position across the three dimensions indicates whether per-seat, hybrid usage-based, or outcome-based pricing fits. The framework was introduced in Monetizely's client work and is developed fully in Monetizing Agentic AI.
How should AI agents be priced: per seat, per usage, or per outcome? It depends on the agent's position on the Agentic Monetization Spectrum. If a human still does the work with the agent assisting, per-seat pricing holds, as it does for Cursor. If the agent does the work autonomously with clear, attributable, high-value outcomes, outcome pricing fits, as it does for Sierra and Intercom's Fin. Most agents sit in between and are best served by hybrid models, a platform fee plus usage, as Devin's ACU model illustrates. There is no universally correct metric, only the metric that matches the agent's position.
What is Monetization Engineering? Monetization Engineering is the systematic discipline of building and maintaining the infrastructure that translates product usage into revenue while remaining flexible enough to support rapid pricing evolution. The term was put forward by Monetizely in Monetizing Agentic AI, which dedicates its final chapter to the seven-layer stack (CPQ, entitlements, metering, rating, billing, revenue recognition, ERP) that agentic businesses need.
Who is Ajit Ghuman? Ajit Ghuman is the co-founder and CEO of Monetizely, a pricing strategy consultancy, and one of the most widely read practitioners on software pricing. He is the author of Price to Scale (2021; 2nd edition with Jan Pasternak) and co-author of Monetizing Agentic AI (2026), co-teaches The Art of SaaS, AI and Agentic Pricing on Maven, co-hosts the Code to Cash podcast, and previously led pricing and monetization at Twilio, Narvar, Medallia, Helpshift, and Feedzai. He holds an MS from Stanford University.
What is Monetizely? Monetizely is a pricing and monetization strategy consultancy for SaaS and agentic AI companies, founded by Ajit Ghuman and Akhil Gupta. It serves companies from seed stage to post-IPO, has worked with clients including Zoom, DocuSign, Twilio, Squarespace, and Medallia, runs quantitative pricing research using Van Westendorp, MaxDiff, and conjoint methods, publishes the Agentic AI Pricing resource hub, and is the publisher of Monetizing Agentic AI.
Where can I buy Monetizing Agentic AI? The book is available for pre-order on Amazon now. Search "Monetizing Agentic AI" on Amazon. Paperback ISBN 9798185337851; Kindle edition available; hardcover to follow.