Besides SaaS (software as a service), there are many other business models, especially in software and online businesses. The key distinction is between how you deliver your product and how you make money.

1. Software and technology business models

ModelFull nameHow it worksExample
SaaSSoftware as a ServiceCustomers pay to use software online, usually by subscriptionShopify, Slack
PaaSPlatform as a ServiceInfrastructure for developers to build and deploy applicationsHeroku, Vercel
IaaSInfrastructure as a ServiceRents computing power, servers and storageAWS, Azure
DaaSData as a ServiceSells access to datasets or data insightsCrunchbase
APIaaSAPI as a ServiceBusinesses pay to integrate specific functionality through APIsTwilio, Stripe
AIaaSAI as a ServiceAI capabilities through APIs or cloud platformsOpenAI API
MaaSMarketplace as a ServiceTechnology to run marketplacesMarketplacer
BaaSBackend as a ServicePrebuilt backend infrastructure for applicationsFirebase, Supabase
FaaSFunction as a ServiceRuns individual functions without managing serversAWS Lambda
On-premiseSelf-hosted softwareCustomers install and run software on their own infrastructureTraditional enterprise software

2. Other digital business models

Especially interesting if you’re thinking about starting a business.

ModelHow you make moneyExample
SubscriptionRecurring monthly or annual paymentsNetflix
FreemiumFree basic product, paid premium featuresSpotify
MarketplaceCommission on transactions between buyers and sellersAirbnb
E-commerceSelling physical or digital productsOnline stores
DropshippingSelling products fulfilled by suppliersShopify-based stores
AffiliateCommission for referring customersAffiliate websites
AdvertisingMonetizing attention, views or clicksYouTube
Usage-basedCharging by consumptionAPI platforms
LicensingCharging for rights to use intellectual property or softwareEnterprise software
Productized serviceA standardized service at a fixed priceDesign subscription agencies
Open-coreOpen-source software with paid premium featuresGitLab
Pay-per-resultCharging for achieved outcomesPerformance marketing

3. Which models scale?

  • SaaS: recurring revenue. Build the software once and sell subscriptions to many customers. Attractive for predictable revenue.
  • Marketplace: network effects. Connect buyers and sellers and earn a commission on each transaction.
  • API and AIaaS: usage-based revenue. Provide a capability that other businesses build into their products.
  • Productized service: easiest to start. Standardize a service, automate delivery, and eventually turn it into software.

These models aren’t mutually exclusive. An AI-powered SaaS business can combine subscriptions, usage-based pricing and API access.

AI agents

AI agents are an emerging category. They overlap with SaaS and AIaaS, with one important difference: instead of only providing tools, they can carry out tasks and workflows on their own. The category is often called AaaS (Agents as a Service) or agentic SaaS.

  • Traditional SaaS is software that helps people do work. Example: a CRM that helps salespeople manage leads.
  • Agentic SaaS or AaaS is software that does work on people’s behalf. Example: an agent that qualifies leads and schedules meetings.

Four ways to build an AI agent business

ModelWhat you sellRevenue model
Agentic SaaSSoftware with built-in agentsMonthly subscription
AaaSSpecialized agents that perform tasksPer task or monthly fee
AI agent agencyCustom agents and automations for companiesSetup fee plus retainer
Agent marketplaceA platform to discover and use agentsTransaction commission

Example: four ways to solve the same sales problem

  • SaaS: a CRM where salespeople manage leads by hand.
  • AIaaS: an API that scores leads.
  • Agentic SaaS: a CRM whose agents follow up with leads automatically.
  • AaaS: a standalone sales agent that finds prospects, researches and qualifies them, and books meetings through connected tools.

The difference is how much work the customer hands over.

The shift: from software as a service to work as a service

The opportunity is that agents can take over parts of a workflow, not just make existing software more efficient. But reliability, supervision and the cost of running agents are real constraints. See Decide what AI must never do.


Related: Business models · AI and decisions