AI Agents Could Change Who Buys Software
For decades, enterprise software has been designed around human users. Companies have relied on sales teams, subscription plans, graphical interfaces and procurement processes to sell software to employees and business departments.
However, the growing capabilities of autonomous AI agents could challenge this model.
In his latest analysis, Paul Veradittakit, Managing Partner at Pantera Capital, argues that software is moving beyond its traditional role as a productivity tool. AI agents could increasingly become independent economic participants, selecting services, purchasing computing resources and completing transactions on behalf of businesses.
This shift could change how software companies build products, charge customers and manage payments.
From Human Users to Machine Buyers
Traditional SaaS platforms are generally designed for people who interact with dashboards, websites and applications. Their commercial models typically rely on monthly or annual subscriptions based on the number of users.
AI agents operate differently.
An autonomous agent performing market research, for example, may need to purchase financial data, access GPU computing resources and use third-party data-processing services. These transactions could happen automatically without requiring an employee to approve every individual purchase.
According to Veradittakit's analysis, this creates a new category of software customers: machines that can discover services, evaluate options and pay for resources independently within predefined permissions.
As this model develops, software providers may need to make their products easier for AI agents to discover, access and purchase.
SaaS Pricing Could Move Beyond Per-Seat Subscriptions
One of the biggest potential changes concerns how software companies charge customers.
The traditional per-seat subscription model assumes that each user is a person who needs access to a platform. However, businesses using hundreds or thousands of AI agents may not find this pricing structure suitable.
Instead, Veradittakit expects pricing models to become more granular, potentially based on:
API calls and service requests.
Computing resources consumed.
Individual tasks completed.
Processing time and usage levels.
Business outcomes delivered.
This could create more flexible pricing arrangements in which companies pay for the specific services their agents consume.
For SaaS providers, the shift may require changes to product architecture, billing systems and customer acquisition strategies.
Why Crypto Payments Could Become Important
Autonomous software agents need payment infrastructure that operates continuously and supports small, frequent transactions.
Traditional payment systems often involve account verification, banking hours, card networks and settlement delays. These processes may create friction when software agents need to make numerous low-value payments across different service providers.
Veradittakit identifies cryptocurrency infrastructure, particularly stablecoins and on-chain settlement systems, as potential solutions.
Stablecoins such as USDC could support machine-to-machine payments by providing digital currency that can be transferred programmatically. Micropayment streaming protocols could also allow services to charge based on actual usage rather than requiring conventional monthly billing.
The analysis highlights several companies operating across this emerging infrastructure, including Circle, Coinflow and OpenFX.
However, widespread adoption will depend on regulatory requirements, security, payment reliability and integration with existing financial systems.
Identity and Security Become Essential
Allowing AI agents to spend money introduces another important challenge: determining who controls an agent and what it is authorised to do.
Businesses cannot simply give autonomous systems unrestricted access to financial accounts.
Agents will need verifiable identities, defined spending limits, permissions and mechanisms for revoking access.
Veradittakit highlights several blockchain-related infrastructure categories that could support this ecosystem:
Identity and credentials: Systems that allow agents to demonstrate authorised access without exposing unnecessary personal or business information.
Programmable wallets: Financial accounts with spending limits, restrictions and temporary permissions.
Machine settlement: Payment systems that allow software agents to send and receive funds automatically.
Computing infrastructure: Services that provide access to computing resources while maintaining control over data and execution environments.
These capabilities could help establish trust between autonomous systems operated by different organisations.
Fission and the Possibility of Machine-to-Machine Commerce
The analysis also discusses Fission as an example of an emerging ecosystem where AI agents can publish requirements, provide services and exchange payments through decentralised infrastructure.
In this model, software agents developed by different teams could interact commercially without requiring traditional human-led negotiations for every transaction.
Such systems could operate continuously, allowing services to be discovered and purchased whenever required.
This represents a possible direction for digital commerce, although the broader market still needs to address interoperability, security, accountability and regulatory compliance.
What Startups Should Consider
For technology startups, the emergence of AI-driven procurement could require a rethink of how products are designed and delivered.
Veradittakit highlights three important considerations:
Reconsider pricing and delivery: Build systems that can support usage-based billing, API consumption and automated payments.
Prioritise machine readability: Ensure products can be discovered, evaluated and accessed by AI agents through reliable interfaces.
Prepare payment infrastructure: Monitor developments in programmable wallets, stablecoins and cryptographic settlement systems.
These changes may become increasingly relevant as businesses introduce autonomous agents into procurement, finance, software development and other operational workflows.
The Future of SaaS and Machine Commerce
AI agents could eventually become a significant category of software buyers, creating new opportunities for SaaS companies and digital payment providers.
The transition will not happen overnight. Human oversight, enterprise security, financial regulations and accountability will remain important considerations when agents are allowed to make purchasing decisions.
Nevertheless, the direction identified by Pantera Capital points towards a different software economy: one in which applications do not simply serve human users but also purchase services, allocate resources and transact with other software systems.
For SaaS businesses, adapting to this possibility could mean designing products not only for people but also for the autonomous systems that increasingly perform work on their behalf.
