**Steven Forth** is CEO of Ibbaka. Connect on LinkedIn

An emergent pattern in the agent economy is for companies to provide customers with the ability to build custom agents that sit on their platforms.

If you have a software application that collects data, has a complex UI, can support many different workflows this is a compelling tactic.

There is a growing move toward replacing conventional user interfaces with agents. Some people, like user experience and design thought leader Jakob Nielsen believe that agents will more or less replace conventional user interfaces (see Hello AI Agents: Goodbye UI Design, RIP Accessibility). Others, like Hubspot CTO Dharmesh Shaw think we will see a more hybrid approach. See Beyond Chat: Blending UI For An AI World.

Standard Agent Innovation Patterns and Pricing

Agents are becoming one of the standard ways that AI is being packaged and taken to market. There are five basic patterns for adding agents to an existing application, each with there own pricing implications.

Ibbaka Standard Agent Innovation Patterns

Extend: Use agents to extend the existing functionality, often by layering in AI based summaries and insights, ore recommendation engines.→ Extensions are generally NOT priced separately from the underlying application. They are included in the price. In some cases they are used to justify an overall price increase.

Add: Add new functionality that is has synergies with existing functionality.→ These additions are generally priced. The best practice is to align pricing metrics with that of the existing functionality to simplify the buying process (and the sales process).

Embed: AI functionality is embedded within the application and the user may not even be aware that an AI is engaged. More and more applications use some form of AI at some point in the tech stack.→ Pricing will depend on how and why the AI is being used and its impact on both operating costs and on value to customer (V2C).

Replace: Existing functionality is replaced by an agent, in many cases the existing functionality can only be accessed through the agent which replaces the legacy application. → This is a great opportunity to rethink pricing and introduce new pricing metrics that do a better job of aligning price and value.

Replatform: This is the most radical approach in which an existing application is decomposed into its core functionality and taken to market. → Pricing will follow best practices in workflow and agent system pricing, an evolving area. See How to price AI agents.

The move to user generated agents

Quite often users will have better insights into what agents are needed than the product team, or will have niche needs that do not make the cut a overwhelmed product teams try to prioritize. The solution is to give users the ability to construct their own agents.

This has several advantages for the vendor:

Have users identify and address pain points (if a user takes the trouble to build an agent it likely solves a real pain point, at least for that users)
Increase loyalty and stickiness (users that have gone to the effort of building something on a platform are likely to want to continue to use that agent)
Give product teams a deeper understanding of the potential of agents (many. product teams are struggling to understand what agents are and how they can be used, having users show them is a powerful learning opport
An internal agent marketplace can be developed (in some cases, vendors may want to give users the opportunity to exchange and monetize agents, creating an external market place)

The key patterns for user generated agents are Extend and Add. A user could take an existing application and extend it with various AI functionality, adding summarization, insights generated by custom frameworks and prompt sequences, blends from two or more different parts of the application. User created agents could add completely new functionality that leverages data or functionality of the application.

One of the tricky bits here is that users may want to combine (or mashup as we used to say) data from two different applications in their agents. Whether vendors allow, or even encourage, this is a critical decision that needs to be made early in design.

Pricing user generated agents

Ibbaka uses it agent AI pricing layer cake to fram pricing of AI agents.

Let’s see how this could be applied to user generator agents.

1. Role Layer: Defining the Agent’s Job-to-be-Done

User-generated agents vary widely in purpose, from niche workflow automation to cross-platform data mashups.
Pricing metrics could include:

Agent complexity tiers (basic automation vs. advanced multi-tool workflows)
Specialization (industry-specific templates vs. general-purpose agents)
Marketplace positioning (free vs. premium agent listings)

For example:

A simple email summarizer agent might fall under a "Basic" tier.
A complex CRM-analytics agent combining Salesforce and HubSpot data could be "Enterprise."

Value alignment: Role-based pricing ensures users pay for the intended value of the agent, whether it’s saving time, reducing errors, or generating insights.

2. Access Layer: Enabling Creation and Deployment

User-generated agents require infrastructure access (APIs, compute resources, tool integrations).
Pricing strategies:

Retainer models for guaranteed platform access (e.g., $50/month for API credits).
Marketplace listing fees (e.g., 10% of agent revenue for featured placement).
Cross-platform access tiers (e.g., $100/month to integrate external data sources).

This layer ensures creators can build and deploy agents while vendors monetize platform access.

3. Usage Layer: Scaling with Adoption

Usage metrics track how often agents are deployed or consumed.
Approaches:

Consumption-based pricing (e.g., $0.01 per API call or task execution).
Tiered credit bundles (e.g., 1,000 tasks/month for $99).
Shared revenue models (e.g., 20% of cost savings from an agent’s automation).

For user-generated agents, usage pricing aligns costs with value delivery while allowing scalability.

4. Outcomes Layer: Rewarding Value Creation

Outcome-based pricing ties fees to measurable results, such as:

Revenue share (e.g., 5% of sales influenced by a marketing agent).
Performance bonuses (e.g., $500/month if an agent reduces support tickets by 30%).
Cost-saving splits (e.g., 15% of operational savings from a workflow agent).

This layer incentivizes creators to build high-impact agents and aligns vendor revenue with customer success.

Credit Systems for User Generated Agents

Given the wide range of use cases, cost profiles and value generation that come with user generated agents a credit system will be the best approach to pricing in many cases. See Why tokens and credits are becoming a standard approach to pricing AI solutions.

Here are some examples to get your design thinking going.

Credit Allocation for Agent Creation

Tiered creation credits: Offer users base credit pools for building agents (e.g., 500 credits/month for basic agents, 5,000 for advanced workflows, one could base this on the number of steps in the work flow, the number or prompts in a prompt sequence, the number of tokens consumed).
Cross-platform integration fees: Charge credits for API calls to external systems (e.g., 10 credits per Salesforce/HubSpot data mashup).
Complexity-based pricing: Assign credit costs proportional to an agent’s computational intensity (e.g., 50 credits for simple chatbots vs. 500 for multi-step analytics agents).

Outcome-Based Incentives

Success rebates: Refund 20% of credits if an agent achieves defined outcomes (e.g., 30% ticket reduction). This may seem counterintuitive but rewarding a customer for success is a powerful way to encourage use.
Performance multipliers: Award 2x credits for agents adopted by >100 users, creating viral growth loops.
Cost-saving splits: Share 10% of operational savings from user-created agents as bonus credits.

Tiered Architectures

Basic Tier

$99/month + 10,000 credits
Up to five agents
Ability to buy additional agents or credits as needed
Credits based on number of tokens consumed

Standard Tier

$199/month + 30,000 credits
Up to fifteen agents
Ability to buy additional agents or credits as needed
Credits based on number of tokens consumed

Pro Tier

$499/month (unlimited basic agents + 50,000 credits)
Priority compute access

This combines Ibbaka’s Role + Usage layer with Box’s Enterprise Plus packaging strategy.

Ecosystem Design

Credit liquidity: Allow trading unused credits between users (5% platform fee).
Staking mechanisms: Let creators stake credits to boost agent visibility, aligning with blockchain-inspired token economies. "Stake credits" refers to a mechanism where users commit (or "lock") a portion of their credit balance as collateral to access specific platform benefits, incentivizing quality and accountability in ecosystems like AI agent marketplaces.
Transparency tools: Real-time dashboards showing credit burn rate per agent, similar to Box AI Units.

By applying these credit-based strategies, vendors can monetize user innovation while maintaining platform stickiness – a balance exemplified by Intercom’s Fin AI (outcome pricing) and Box’s hybrid models. The system rewards high-value agents through market-driven credit flows while containing costs through granular consumption metrics.

Step by Step Approach to Pricing Agent Building for a Leagacy Application

Establish clear business goals for this functionality
Make the legacy application stickier
Identify product innovation opportunities
Increase value to customer (V2C)
Defence against emerging agent first disruptors
Create a marketplace for user built agents
Set policies. Can user built agents …
Include third party data?
Connect multiple applications?
Be orchestrated to support workflows
Can users provide agents to each other?
Will the vendor clone and possibly monetize the most successful agents?
Understand the costs of operating user built agents
Choose which types of pricing metric to engage (use the AI Agent Pricing Layer Cake)
Develop a model to estimate the value of different types of agent and uses of the agent
Decide whether there should be a credit system based on the value model
Make sure that billing systems can support your pricing model for user built agents

Conclusion

The rise of user-generated agents is fundamentally reshaping the agent economy, turning every software platform into a canvas for customer-driven innovation. By empowering users to build, extend, and add their own AI agents, vendors not only unlock new sources of value but also foster deeper engagement and loyalty.

The key is to support this creativity with flexible, credit-based pricing models that scale with complexity, usage, and outcomes-ensuring both customers and vendors benefit as agents evolve. As platforms open up and agent marketplaces emerge, the companies that embrace user agency and smart monetization will define the next era of software.

Are you ready to let your users build the future?

contact ibbaka to get a value model you can use to win in a generative AI world