.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.
Note: I have not used the attachment. The article below is based on independent research and current market developments.
AI agents may not simply automate procurement; they could fundamentally change where pricing power comes from.
For decades, B2B pricing has depended on a simple reality: neither side sees the full market. A buyer may know its incumbent supplier, last year’s price and a few alternative quotes. A seller may understand its own cost-to-serve, a customer’s historic spend and perhaps the competitive positioning of a handful of rivals. But neither party has had the time, data access or analytical capacity to continuously compare every relevant alternative across price, contract terms, reliability, integration effort, service levels, risk and expected business outcomes.
That limitation has supported much of the traditional B2B commercial model. Price books establish an anchor; sales representatives create differentiated offers; procurement teams run periodic sourcing events; and the final deal reflects negotiation skill, internal politics, switching costs and imperfect information as much as it reflects intrinsic product value.
AI agents challenge that model because they can turn procurement from an episodic process into a continuous comparison engine. A capable buyer-side agent can inspect internal contracts, historical purchases, supplier performance, third-party risk data, usage data and current market alternatives. It can identify price drift, normalize quotes that use different units or commercial assumptions, invite suppliers to bid, and escalate only the exceptions that require human judgment. Seller-side agents will increasingly do the reverse: model willingness to pay, identify cost-to-serve, generate compliant offers, recommend terms and defend margin in real time.
The consequence is not necessarily a universal race to the lowest price. In many B2B categories, the lowest sticker price is a poor proxy for the best commercial choice. The more important shift is subtler: pricing power will move away from opacity, sales-process control and discretionary discounting, and toward demonstrable, machine-verifiable economic value.
That is a profound change. When both sides can model the deal, sellers will need to make their value legible not only to a human buyer but also to the buyer’s procurement system. Pricing will become more transparent in some categories, more dynamic in others, and more tightly linked to measurable outcomes almost everywhere.
1. Why traditional B2B pricing works the way it does
B2B pricing is often portrayed as rational, but it has never been perfectly efficient. It is constrained by fragmented data and organizational bandwidth.
A large enterprise may buy the same category through multiple business units, countries and purchasing channels. Contracts may be buried in PDFs, product configurations may differ across suppliers, rebates may sit outside the invoice price, and performance data may live in separate systems. Even where a company has negotiated favorable terms, employees may buy outside the agreement because the approved vendor is difficult to find or use.
This fragmentation creates a large “information tax.” McKinsey has noted that long-tail indirect purchases are often characterized by scarce product and price information; in one common spend pattern, roughly 20% of indirect spend is spread across 80% of the supplier base. That makes disciplined comparison uneconomic for human procurement teams, especially for low-value or infrequent purchases.
Suppliers benefit from a related set of asymmetries. They may know which customers are price-sensitive, where a buyer is dependent on a particular integration, which usage commitments are underutilized, or how difficult a replacement project would be. They can use complicated packaging, nonstandard metrics, renewal timing and bespoke bundles to make comparison difficult. In enterprise software and cloud services, a buyer may be choosing among thousands of SKUs, pricing tiers, usage thresholds, support packages, implementation requirements and commitment discounts.
The point is not that complexity is always artificial. Many enterprise offerings genuinely are complex because the buyer’s requirements are complex. But commercial complexity also creates negotiating room. The seller that can frame the deal, control the information flow and anchor the discussion around a list price or feature bundle often has an advantage.
Traditional procurement does not eliminate this imbalance because it is expensive to scale. A sourcing professional can run a thoughtful competitive process for a strategic category. They cannot personally renegotiate every small contract, validate every supplier price increase, inspect every invoice against every clause or solicit alternatives for every requisition.
That capacity constraint is precisely where agents matter. Oracle now offers an Autonomous Sourcing Assistant that can convert eligible requisition lines into sourcing events, invite suppliers using historical data, collect bids and generate purchasing documents after approvals. Its Supplier Negotiation Award Assistant can apply awards according to policy, including minimum-bid requirements and spend thresholds. The important commercial implication is not the workflow automation itself. It is that competitive tension can be applied to a much larger share of spend.
2. What changes when AI agents compare every option
The immediate effect of procurement agents will be to lower the cost of comparison.
A buyer agent can evaluate alternatives in parallel rather than sequentially. It can parse an incumbent’s proposed renewal, identify price increases by SKU, compare those increases with historic rates and external benchmarks, search approved alternatives, and create a negotiation brief. It can also consider data a human buyer might overlook: on-time delivery, defect rates, security findings, support-ticket resolution time, emissions data, payment terms, contract-renewal risk and integration dependencies.
This does not require a fully autonomous “machine buyer” on day one. The first commercial impact comes from agents making every buyer more prepared. Gartner reported in 2025 that 74% of procurement leaders said their data was not AI-ready—an important constraint—but the direction of travel is clear: procurement leaders are being pushed to make data accessible, governed and usable by more autonomous systems.
The market is already moving beyond simple chat interfaces. SAP said its Sourcing Agent, available in beta in 2025, accelerated sourcing-event creation, while SAP Ariba’s AI summarizer reduced document-review time by 50%. Zip has introduced more than 50 procurement-specific agents, including agents for vendor risk assessment, invoice-to-contract compliance and price negotiation. Ivalua, GEP, Oracle and other source-to-pay platforms are similarly positioning agents as systems that connect sourcing, contract, supplier and purchasing data rather than merely answer questions.
The economic shift is straightforward. If it becomes cheap to evaluate a credible alternative, the incumbent loses some of the protection previously afforded by buyer inertia. If it becomes cheap to challenge every unjustified price increase, suppliers will find it harder to rely on the fact that a small buyer-side team cannot review everything.
Autonomous negotiation platforms offer an early indication of the scale effect. Pactum says Honeywell has put nearly $500 million of spend through its AI system and completed more than 2,500 autonomous negotiations; Honeywell’s chief procurement officer, Matt Duffy, described the goal as accelerating deal cycles while aligning negotiations with real-time market data. These are vendor-reported results, not independently audited market averages, but they demonstrate that autonomous negotiation is already being used for real procurement workflows rather than remaining purely conceptual.
The most disruptive capability is not that an agent can negotiate one deal well. It is that it can pursue thousands of “too-small-to-manage” opportunities simultaneously. That turns unmanaged spend, contract leakage and price drift into addressable commercial territory.
3. Will more transparency mean commoditization?
In standardized categories, probably yes.
Where products are comparable, switching costs are low and suppliers can present normalized data, agents will intensify price competition. Think office supplies, maintenance parts, freight lanes, basic cloud infrastructure, contingent labor categories with clear rate cards, or many marketplace purchases. These are markets in which a buyer agent can rapidly ask: Which qualified supplier meets the specification at the lowest landed cost, within the required delivery window and risk threshold?
Digital marketplaces have already shown the direction of travel. They improve visibility into product availability, supplier options, pricing and purchasing terms, particularly for indirect and long-tail spend. Agents will make those marketplaces more powerful by converting visibility into ongoing action: continuously checking prices, reordering from compliant sources, aggregating demand and running tactical sourcing events when thresholds are crossed.
Yet transparency is not the same as commoditization.
First, prices become meaningful only when products are comparable. A $100,000 software proposal may look cheaper than a $130,000 proposal until the buyer agent models implementation labor, required third-party tools, security reviews, downtime risk, usage overages, data-migration costs and support. A lower freight quote may prove more expensive after claims rates, late-delivery penalties and capacity reliability are included. A lower consulting day rate can be irrelevant if the project takes twice as long or requires more senior intervention.
Second, greater transparency can expose poor comparability. An agent can make clear that two “similar” products are not functionally equivalent. That may strengthen the pricing power of a differentiated vendor, provided it can prove the difference.
Third, transparency itself can create strategic risks. If every seller uses the same market signals and pricing software in a concentrated market for homogeneous products, rapid observation and response can dampen incentives to undercut. The OECD warns that algorithmic pricing can facilitate explicit collusion, hub-and-spoke coordination or tacit coordination, and that automated price matching may stabilize prices in some markets rather than reduce them.
Therefore, the likely outcome is not “AI causes lower prices.” It is a sharper sorting of markets:
- Highly comparable, low-switching-cost categories face greater price pressure.
- Differentiated categories face greater scrutiny but can preserve—or even expand—pricing power if they prove superior economics.
- Concentrated markets require governance to ensure algorithmic pricing does not replace human opacity with machine-enabled coordination.
4. Why the lowest price may not win
The strongest vendors in an agentic procurement market will not try to hide from comparison. They will define the comparison.
A capable procurement agent should optimize for a buyer’s objective function, not for unit price alone. That function may include total cost of ownership, revenue impact, service continuity, compliance, implementation time, supplier financial health, interoperability, data portability, sustainability requirements and the cost of failure.
This matters because many B2B purchases are consequential. A manufacturing supplier that offers a part at a 3% lower unit cost may be a poor choice if its late deliveries halt a production line. A cloud platform that appears cheaper per compute unit may be costlier if it requires substantial re-architecture or lacks the availability commitments needed for a critical application. A professional-services firm may command a premium because it has scarce expertise, better execution history or accountability for an outcome rather than merely supplying hours.
The seller’s opportunity is to convert these advantages into data. Rather than saying, “We offer premium service,” it should provide measurable evidence:
- verified uptime and incident-resolution metrics;
- on-time, in-full delivery performance;
- integration coverage and implementation time;
- support response and resolution commitments;
- retention, adoption and realized-ROI data;
- cost-to-switch estimates;
- service credits and remedies;
- references relevant to the buyer’s industry and use case.
In other words, the procurement agent changes the burden of proof. Commercial claims that once worked in a sales presentation will increasingly need structured evidence behind them.
This is why the cheapest vendor will often lose. But the vendor with the clearest proof of superior expected value may win more consistently—and at a healthier price—than a vendor relying on salesmanship, brand familiarity or opaque bundling.
BCG’s research on agentic software pricing reaches a related conclusion: buyers show more willingness to pay where an AI capability provides proprietary expertise or completes a meaningful job, while isolated and easily replicated workflow features are harder to monetize separately. That distinction will apply beyond AI products. If an offer is easy for an agent to normalize and substitute, margin will be vulnerable. If it produces an outcome that competitors cannot credibly match, transparency becomes an asset rather than a threat.
5. The rise of machine-to-machine negotiation
Machine-to-machine negotiation is plausible, but it will arrive first in bounded commercial environments rather than in complex strategic deals.
The prerequisites already exist. Procurement systems can exchange catalogs, orders and invoices through established structured formats. cXML, for example, supports supplier catalog content, item descriptions, supplier identifiers and prices; SAP notes that separate pricing files can be used to manage customer-specific contract pricing. Google’s Agent2Agent protocol, announced in April 2025, was designed to let agents communicate, exchange information and coordinate actions across enterprise platforms and applications.
Combine those foundations with authorized commercial guardrails and a basic machine-to-machine negotiation becomes possible:
- A buyer agent issues a request with specification, volume, location, acceptable alternatives, service requirements and authority limits.
- Seller agents return structured offers: price, availability, lead time, implementation plan, service level, payment terms, volume tiers and exceptions.
- The buyer agent scores offers against a pre-agreed decision model.
- The seller agent adjusts within its approved margin, capacity and commercial-policy boundaries.
- The system either creates an agreement or escalates the unresolved trade-off to humans.
This will likely begin with categories that are repetitive, low-risk and governed by clear rules: tail-spend purchases, catalog goods, rate-card services, supplier onboarding, payment-term harmonization, post-RFP optimization and contract renewals within defined thresholds.
Pactum’s work with Walmart is an important early example. Harvard Business Review documented Walmart’s use of AI-supported supplier negotiations, and Pactum reports that Walmart achieved an average 3% gain while extending payment terms by 35 days; it says 68% of suppliers that engaged reached agreement. Those results should be treated as case-specific, but they reveal a critical point: automated negotiation need not be limited to price. It can trade across payment timing, rebates, service commitments and other terms.
The near-term model will not be unsupervised agents haggling over every strategic enterprise agreement. Complex contracts involve ambiguity, legal interpretation, relationship history and business strategy. Instead, agents will negotiate the structured portions of deals and escalate the judgment-heavy portions.
That division of labor is commercially significant. If an agent handles routine concessions, price claims, benchmark comparisons and alternative scenarios, human negotiators can focus on the terms that genuinely require executive judgment: exclusivity, intellectual property, risk-sharing, strategic capacity, joint product roadmaps and long-term partnership economics.
6. Dynamic pricing when both sides use AI
As buyer agents become more capable, seller-side dynamic pricing will become more sophisticated too.
Sellers will be able to adjust offers based on utilization, capacity, input costs, customer segment, implementation burden, competitive alternatives, strategic account value and probability of renewal. Buyers will respond with agents that model whether the offer is genuinely attractive after commitments, consumption patterns and switching costs are considered.
This will not look like consumer-style surge pricing applied indiscriminately to enterprise accounts. Procurement contracts often require predictability, governance and auditability. But it will accelerate the move from static list prices toward flexible commercial architectures: usage bands, capacity commitments, outcome-linked fees, pre-commit discounts, performance credits and real-time rate adjustments within contractually defined corridors.
Cloud computing offers a useful precedent. AWS combines on-demand pricing with one- and three-year Savings Plans, reservations and discounted Spot capacity; AWS says Spot can provide discounts of up to 90% from on-demand pricing, while reservation and commitment structures exchange flexibility for lower rates. This is already a machine-readable, multi-variable pricing environment. An agent can continuously optimize workloads and commitment coverage in ways that are difficult for a human FinOps team to replicate at transaction level.
Enterprise software is moving in the same direction. Salesforce now offers Agentforce through multiple commercial models: $2 per conversation, consumption-based Flex Credits, and user licenses. Its Flex Credits are priced at $500 per 100,000 credits, with standard Agentforce actions consuming 20 credits; customers can choose pay-as-you-go, pre-commitment or pre-purchase structures. This is more than a pricing update. It illustrates a wider shift from charging for access to charging for measurable units of work.
The challenge is that dynamic pricing can easily become opaque pricing. A buyer agent may be able to inspect usage, but only if vendors expose the relevant data and contract logic. Vendors that hide pricing behind unexplained credits, proprietary scoring or unpredictable overages may create short-term revenue opportunities while becoming disfavored by procurement systems that prioritize forecastability and auditability.
7. New sources of pricing power
In an agentic economy, the durable sources of pricing power are likely to be:
1. Verifiable outcomes.
Can the supplier prove that it reduces downtime, accelerates implementation, improves conversion, lowers defects or resolves issues without human intervention? Outcome proof will matter more than broad feature claims.
2. Proprietary data and domain expertise.
A generic capability is easier to substitute. A vendor with unique data, specialized workflow knowledge or demonstrated industry expertise is harder to replace.
3. Embeddedness without lock-in.
Deep integrations can create real value, but agents will expose punitive switching costs and opaque data barriers. The winning vendor will be embedded because it improves the customer’s economics—not simply because exit is painful.
4. Reliability and risk absorption.
Vendors that accept measurable accountability through service-level agreements, warranties, performance credits or shared-risk contracts can justify a premium.
5. Ease of machine evaluation.
This is the underappreciated source of advantage. If a supplier’s product, price, availability, terms, integration requirements and performance evidence are available in structured form, it becomes easier to discover, evaluate and buy.
That last point means standardized, machine-readable pricing may become a competitive asset. Not necessarily fully public pricing—enterprise pricing will remain account-specific—but structured pricing. A supplier should be able to provide an authorized agent with a clear description of the offer: unit metric, inclusions, exclusions, minimums, tiers, overage logic, renewal terms, implementation scope, service levels and contract dependencies.
The vendor that says “contact sales” may still win strategic deals. But for an increasing share of spend, it may not even make the shortlist if an agent cannot evaluate it efficiently.
8. What pricing teams should change now
Pricing leaders should not respond by publishing a simplistic price list and hoping transparency works in their favor. They should redesign the commercial system around comparability, evidence and controlled flexibility.
First, simplify the offer architecture.
Agents will expose redundant SKUs, inconsistent discount rules and arbitrary packaging. Rationalize products into understandable packages with clear value metrics. If a product is priced by seats, usage, transactions, outcomes and implementation tiers simultaneously, ensure each metric has an economic rationale.
Second, make discounts less discretionary and more conditional.
The future of discounting is not “no discounts.” It is discounts that buy something valuable: longer commitments, higher volumes, lower support burden, reference rights, faster payment, standardized deployment or broader adoption. An agent will be better at identifying whether a discount has a legitimate give-get exchange or merely reflects sales pressure.
Third, build a commercial evidence layer.
Pricing teams need telemetry, benchmark data, customer-success metrics, cost-to-serve data and contract-performance data. BCG argues that companies moving toward agentic pricing need capabilities such as usage forecasting, telemetry, quoting and billing; these are no longer back-office concerns but prerequisites for monetizing value. (bcg.com)
Fourth, create machine-readable offer APIs or data feeds.
Do not wait for a universal procurement-agent standard. Start with structured catalogs, product metadata, service-level definitions, pricing logic and approved quote-generation interfaces. Existing procurement standards already support structured catalog and pricing exchange; the commercial opportunity is to extend that discipline to complex enterprise offers. (xml.cxml.org)
Fifth, prepare negotiation guardrails for seller-side agents.
Define floors, walk-away conditions, approved trade-offs, risk thresholds and escalation rules. A seller agent should be able to make a faster offer without giving away margin or accepting contractual risk accidentally.
Sixth, measure total deal quality, not booked price.
As agents improve comparison, pricing teams should track realized usage, margin after cost-to-serve, renewal economics, implementation performance, customer outcomes and leakage. The agentic market will punish a commercial model that looks attractive at signature but fails in operation.
9. Which businesses are most exposed?
SaaS and enterprise software are highly exposed because their pricing is often complex, negotiated and increasingly challenged by seat compression. BCG found that 40% of IT buyers identified seat reduction as their main lever for reducing software spend, while 68% of vendors in an ICONIQ survey charged separately for AI enhancements or reserved them for premium tiers. (bcg.com) Vendors selling undifferentiated workflow features may face strong pressure; vendors that can prove task completion or operational outcomes may gain pricing power.
Cloud infrastructure will see rapid change because usage is already metered, machine-readable and operationally measurable. Buyer agents can optimize commitments, workloads and cross-cloud alternatives continuously. The competitive battleground will extend beyond unit price to egress costs, reliability, ecosystem integration, compliance and engineering productivity.
Professional services will face a different challenge. AI agents can compare rate cards, staffing profiles, project histories and delivery terms more efficiently, weakening the pricing power of generic labor arbitrage. But elite expertise, accountability, speed and outcome-linked risk sharing can become more valuable. Professional-services firms should expect pressure on billable-hour models and greater demand for fixed-fee, milestone-based or shared-value contracts.
Marketplaces and indirect-procurement suppliers may shift first because standardization is high and transaction sizes are often too small for intensive human negotiation. Here, agents can rapidly increase competition and channel volume toward suppliers that provide clean data, reliable fulfillment and frictionless purchasing.
Industrial and direct-material suppliers will move more cautiously. Product qualification, supply continuity, regulatory requirements and engineering dependencies limit easy substitution. Still, agents can challenge price increases with should-cost models, track commodity indices, compare delivery performance and expose hidden variation across plants or regions. The result may be less commoditization than better-informed, more frequent negotiation.
Conclusion: Pricing power in an agentic economy
AI agents will make B2B pricing more competitive—but not merely by pushing prices down.
They will reduce the economic value of information asymmetry. They will make it cheaper to compare vendors, challenge price increases, enforce contracts and negotiate the long tail of spend that human teams cannot systematically manage. They will increase pressure on opaque packaging, arbitrary discounting and products whose claimed differentiation cannot survive structured comparison.
But they will also create a countervailing opportunity. When buyer agents can evaluate total cost of ownership and operational outcomes more rigorously, vendors with real differentiation should be able to defend premium pricing more effectively. The winner will not be the company with the lowest list price. It will be the company that makes its superior economics easiest to verify.
That is the central strategic implication: in an agentic economy, pricing power shifts from what a seller can conceal to what it can prove.