Pricing AI features: what 11 episodes of Pricing Friends show about markups, credits and pricing metrics
- Published
If you're responsible for revenue or product at a software company, you have to decide whether AI features sit inside a plan, cost extra or are billed by usage. This page collects what founders, investors and pricing experts said in 11 episodes about AI markups, billing units and AI in price setting itself. You can listen to each episode directly.
What the episodes show
- AI features often aren't paid for. About 50% of the SaaS and AI companies surveyed bundle their AI features into the core product, and another 20% don't monetize AI at all yet (#143).
- The AI margin is often unknown. 65% of respondents can't put a number on the gross margin of their AI offerings (#143).
- Per-user pricing is losing ground. Only 7% still plan to rely primarily on per-seat pricing, while usage-based models such as credits and tokens grow from 6% to over 30% relevance (#142).
- The shift is underway. 80% of the experts surveyed plan to rework their pricing within the next 12 months (#144).
AI changes the billing unit
For years, seat-based pricing was the standard contract between software companies and customers: per user, per month, predictable for both sides. AI changes that logic because companies increasingly judge software by outcomes (#126). Demodesk charges €49 per user per month for its core product. CEO Veronika Wax thinks self-built AI agents need different pricing (#119).
Malte Kosub, co-founder and CEO of Parloa, explains why SaaS pricing models with fixed per-user prices no longer work for many AI applications. For AI agents on the phone, per-minute pricing, platform models and enterprise APIs come into play (#089).
“I do believe that at a time when we see companies wanting to deploy more and more AI agents, seat-based pricing doesn’t quite work anymore, and the focus is much, much more on outcomes, real ROI or revenue impact.”
Gülsah Wilke, partner at DN Capital, likes hybrid models that combine a platform fee with a clear share of revenue or ROI (#126).
Credits and minutes make tokens understandable
Credits bridge a technical token model and a billing unit customers understand (#142). ElevenLabs uses credits and per-minute pricing that work like an “in-game currency” (#120). On its agents platform, the most expensive unit is about 8 cents per minute (#126). Melanie Riebel, chief revenue officer at Chaos, explains how their credit model spreads AI costs fairly across customers without overwhelming them with full pay-as-you-go (#144).
The downside: A credit balance alone doesn't create predictability, because many users only learn at month's end how many credits they've used (#142). Buyers demand predictability and still prefer flat-fee models (#126). Despite all the debate, Felix Rahlmeyer of the German software review platform OMR Reviews says per-seat pricing remains the largest pricing group in the market (#144).
44% of SaaS companies charge for AI-powered features (Maxio, 2025). Among companies that monetize AI, 53% use a subscription, 31% a hybrid model and 11% pure usage billing (Stripo Research, 2026).
Free, commodity or markup
At awork, AI features are currently still free, and Tobias Hagenau explains why that could stay that way (#090).
“AI will become a commodity feature, just like PowerPoint, that customers won’t pay extra for in the long run. That means you make yourself dependent on the pricing power of the AI providers.”
That's Arne Wolter, CEO of Appinio (#143). According to Wolter, a flexible model gateway keeps you independent.
Jamie, an AI notetaker, deliberately chose neither classic cost-based nor classic value-based pricing (#140). Before the product was finished, the team spent a week testing a real price on its website to see whether anyone would pay. Within 24 hours, several hundred sign-ups came in (#140). Later, a tightly tiered seat price became a generous allowance plus a usage component for workflows (#144).
Value-based pricing rarely works
Dr. Oliver Pabst, general partner at Redalpine, asks why only 5% of software vendors succeed with true value-based pricing. Few AI applications deliver impact measurable enough to make value-based or outcome-based pricing stick (#084). He argues for a pricing logic that looks beyond tokens and usage units (#084).
Many see outcome-based pricing as the ultimate form of AI monetization, but it isn't always easy to implement. Agreeing on a shared measure of success is a challenge (#143). A model like “pay only for resolved cases” sounds fair, but on closer look it often creates consulting work or distorted incentives (#089). Veronika Wax of Demodesk doesn't see outcome-based pricing as relevant (#119). DN Capital sees it differently: It views investments without an ROI component in the pricing model critically (#126).
AI in price setting: pricing software
AI also sits on the other side of the price, in the software companies use to set their prices. Dr. Markus Husemann-Kopetzky, senior research fellow at Paderborn University, shows how AI and data analysis are changing pricing, which pricing software fits which business model and which mistakes often happen during rollout (#034).
In the same episode, three founders present their solutions: Martin Wricke of pricefx; Felix Hoffmann of 7Learnings, which, according to the episode, helps online retailers grow revenue with data-based pricing strategies; and Sebastian Baier of buynomics, which builds data-driven tools to predict customer behavior (#034).
What you can do on Monday
- Calculate the gross margin of your AI features. 65% of respondents can't (#143). Only with this number will you know whether AI should be included or charged extra.
- Review your billing unit. Set an alternative of allowance plus usage, as Jamie did (#144), next to your seat price, and run the numbers for three existing customers.
- Decide how your AI features enter the price. About 50% bundle them into the core product, and 20% don't monetize them at all (#143). Choose deliberately between plan, add-on price and credits.
Episodes on this topic
- #034 · Dr. Markus Husemann-Kopetzky, senior research fellow at Paderborn University: how to choose the right pricing software with AI and data analysis.
- #084 · Dr. Oliver Pabst, general partner at Redalpine: when AI creates measurable value and value-based pricing holds up.
- #089 · Malte Kosub, co-founder and CEO of Parloa: per-minute pricing, platform models and the limits of outcome-based pricing for AI agents.
- #090 · Nicola Breyer, Tobias Hagenau (awork), Philipp Schrader (OMR Reviews) and Jörg Rheinboldt (heartfelt capital): a live panel on free AI features, hybrid models and the limits of usage- and outcome-based pricing.
- #119 · Veronika Wax, CEO of Demodesk: what happens to the seat price when AI agents do the work.
- #120 · Yanek Brinkmann, go-to-market director for Germany, Austria and Switzerland at ElevenLabs: credits and per-minute pricing in audio AI.
- #126 · Gülsah Wilke (DN Capital), Dr. Toni Schmidt (Flex Capital), Gordian Braun (ElevenLabs) and Felix Rahlmeyer (OMR Reviews): a panel on whether AI will replace user-based pricing.
- #140 · Benedikt Böringer, co-founder and CEO of Jamie: how an AI notetaker found its own path between cost- and value-based pricing.
- #142 · Mariella Knospe, senior consultant on hy's pricing team, and Felix Rahlmeyer, VP strategy at OMR Reviews: market data on pricing metrics, credits and software search via AI assistants.
- #143 · Arne Wolter, CEO of Appinio: whether AI is becoming a commodity and what that means for monetization.
- #144 · Arne Wolter, Benedikt Böringer, Felix Rahlmeyer, Melanie Riebel (Chaos) and Norman Nielsen (idealo): five answers on the dominant pricing metric.
Next step
The full numbers from 4,400 analyzed software profiles, a survey of more than 150 SaaS and AI companies and 23 expert interviews are in the free SaaS & AI Pricing Report 2027, available for download (#144). If you want an outside check of your pricing model, start with the free pricing audit.
Bottom line: AI features need their own pricing logic, and predictability still matters
Per-user pricing is losing ground for AI products, and credits and hybrid models are taking over. But a balance alone doesn't create the predictability buyers keep demanding. Many vendors give AI features away or don't know their margin. True value-based pricing only works where AI delivers measurable value. And inside pricing software itself, AI helps set prices based on data.
Sources
- Episode #034: Pricing Tools mit Dr. Markus Husemann-Kopetzky: Wie entscheide ich mich für die richtige Pricingsoftware?
- Episode #084: KI und Pricing mit Dr. Oliver Pabst: Warum gelingt echtes Value-Based Pricing nur 5 % der Softwareanbieter?
- Episode #089: AI Agents und Pricing mit Malte Kosub: Welches Preismodell funktioniert wirklich für KI-Telefonassistenten?
- Episode #090: SaaS und AI Pricing 2026 mit Nicola Breyer, Tobias Hagenau, Philipp Schrader und Jörg Rheinboldt: Was treibt die nächste Preisrevolution?
- Episode #119: Sales, AI und Pricing mit Veronika Wax: Was passiert mit dem Seat-based Pricing, wenn KI-Agenten die Arbeit übernehmen?
- Episode #120: Voice AI und Pricing mit Yanek Brinkmann: Wer tippt denn noch auf einer Tastatur?
- Episode #126: SaaSpocalypse - Panel mit Gülsah Wilke, Toni Schmidt, Gordian Braun und Felix Rahlmeyer: Stirbt User-based Pricing durch KI?
- Episode #140: Meeting Intelligence und Pricing mit Benedikt Böringer: Wie Jamie seinen eigenen Weg zwischen Cost- und Value-Based Pricing gefunden hat?
- Episode #142: SaaS- und AI-Pricing mit Mariella Knospe und Felix Rahlmeyer: Wie bepreist man Software und AI im Jahr 2027?
- Episode #143: SaaS- und AI-Pricing mit Arne Wolter: Wird KI zur Commodity?
- Episode #144: SaaS & AI Pricing Report mit Arne Wolter, Benedikt Böringer, Felix Rahlmeyer, Melanie Riebel & Norman Nielsen: Wie verändert KI das Pricing von morgen?
- SaaS & AI Pricing Report 2027, 2026, https://hy.co/pricing-and-sales/saas-ai-pricing-report-2027/lp/
- Maxio, 2025, https://maxio.com/resources/2025-saas-pricing-trends-report
- Stripo Research, 2026, https://research.stripo.email/saas-pricing-trends-2026
Quotes are translated from German.