Quick Summary: Make.com AI Agents is a platform feature available on all paid plans (including the Free plan) that enables users to build intelligent automation workflows with adaptive decision-making capabilities. Released in 2025 and updated in February 2026, it allows creation of reusable AI agents that work across multiple scenarios, with customizable system prompts, flexible model selection including OpenAI-compatible LLMs, and the ability to add tools and knowledge files for context-aware responses.

Make.com AI Agents represent a significant evolution in workflow automation, bringing intelligent decision-making directly into the platform's visual builder. Available now on all paid plans, this feature lets teams create sophisticated AI-powered workflows without writing code.

The platform introduced AI Agents in 2025, then released an updated version—Make AI Agents (New)—on February 2nd, 2026. Both versions remain in open beta, which means functionality and pricing may still change.

What Makes These Agents Different

Unlike simple automation rules, Make AI Agents handle complex business processes with adaptive logic. They analyze context, make decisions, and adjust their behavior based on the information they receive.

The centralized management approach stands out. Build an agent once, then deploy it across multiple workflows. This cuts redundancy and keeps your automation strategy consistent.

Build AI Automations Around Your Stack With OSKI

OSKI develops custom software and integrations for companies that need automation to fit existing systems. Their team can work with APIs, backend services, cloud infrastructure, AI components, and third-party platforms.

Need More Than a Basic Automation?

OSKI can help with:

• building custom AI automation logic

• connecting tools, APIs, and databases

• adding LLM features to workflows

• preparing AI systems for real use

👉 Contact OSKI to discuss your project.

Setting Up an Agent

The AI Agents configuration tab sits in the left sidebar of Make's builder. Here's where agents come to life.

Start with the system prompt—the agent's core instructions. This prompt defines purpose, behavior, constraints, and guidelines. The agent follows these instructions across every workflow where it's deployed.

An optional "improve" button applies AI enhancement to existing prompts, helping maintain consistent outcomes. Leave the field empty for an automatic improvement, or provide specific guidance for targeted refinement.

Model Selection and Providers

Make AI Agents supports various large language models. All plans can use Make's AI provider. Paid plans unlock custom AI provider connections—letting teams connect their own OpenAI, Anthropic, or other compatible services.

But here's the catch: some users report limitations when trying to switch providers during agent creation. Community discussions from September 2025 highlight friction around provider selection on certain plan tiers.

Working with the Modules

The Make AI Agents app includes specialized modules for building and running agents within scenarios.

The "Run an Agent" module handles most operations: creating agents, providing tools, sending messages as instructions or context. It's the workhorse of the system.

Adding Knowledge and Tools

Knowledge files give agents additional context. Supported file types include txt, pdf, docx, csv, md, and json files, with a 20 MB maximum per file. Upload documents, data sets, or reference materials so the agent tailors responses to specific business needs.

Tools extend what agents can do. Connect them to other Make modules, external APIs, or data sources. The agent decides when to use each tool based on the task at hand.

The newer Make AI Agents (New) app, released February 2nd, 2026, includes a built-in chat interface for testing. Configure tools and knowledge, then validate agent behavior before deploying to production workflows.

Real-World Impact

Academic research on AI agents in production environments reveals substantial operational gains. Telecommunications company Vodafone deployed an agent-based customer support system that handles 70% of inquiries without human intervention while reducing average resolution time by 47%.

In logistics, AI agents cut shipment documentation processing time by 62% and reduced error rates by 83%. An investment firm's agent-based decision support system improved risk-adjusted returns by 12%.

Pricing and Availability

Make AI Agents is available on all paid plans. The platform uses a credit-based billing model for AI features, with consumption aligned to actual resources used.

Community discussions from February 2025 reveal pricing isn't fully transparent yet. Unlike traditional modules where one operation equals one credit, AI features consume credits based on model usage, context size, and processing complexity.

The beta status means both functionality and pricing may shift. Check Make's official documentation for current credit costs before building production workflows at scale.

Key Considerations

• The open beta label matters. Features may change, and early adopters should plan for potential adjustments to their implementations.

• Provider flexibility varies by plan. All plans can use Make's AI provider, while paid plans allow custom connections. For teams with existing AI infrastructure or specific model requirements, this distinction impacts architecture decisions.

• Credit usage tracking becomes crucial. Unlike predictable per-operation costs, AI agent consumption varies with prompt length, model choice, and response complexity. Monitoring consumption patterns helps avoid billing surprises.

FAQ

What plans include Make AI Agents?

All plans can access the feature using Make's AI provider, while paid plans unlock the ability to connect custom AI providers like OpenAI or Anthropic.

Can I use my own OpenAI API key?

Yes, but only on paid plans. Custom AI provider connections let teams bring their own API credentials for supported language models, though some users report interface limitations when selecting providers during agent setup.

How much do AI Agents cost?

Make uses a credit-based system where AI feature costs align with actual resource consumption. Unlike standard modules (1 operation = 1 credit), AI agents consume credits based on model type, prompt size, and processing complexity. Check the official documentation for current credit rates.

What's the difference between the old and new AI Agents app?

Make AI Agents (New), released February 2nd, 2026, adds a built-in chat interface for testing agents before deployment. It also refines the configuration interface and tool integration process. Both versions remain in open beta.

Can I reuse agents across multiple workflows?

Yes. Centralized management lets teams create an agent once and deploy it across multiple scenarios. This reduces redundancy and maintains consistency in how agents behave across different automation workflows.

What kind of files can I upload as knowledge?

Agents accept context files in the following formats: txt, pdf, docx, csv, md, and json files. Maximum file size is 20 MB per file. File limitations include a maximum of 50 files per team (250 for enterprise plans) and 100 files per organization (500 for enterprise plans), with a 20 file maximum per individual AI agent.

Is Make AI Agents production-ready?

The feature remains in open beta as of May 2026. While functional and available on all paid plans, both product functionality and pricing may change. Teams should factor beta status into production planning and monitor official announcements for updates.

Getting Started

For teams ready to experiment, the barrier to entry is low. Any paid Make plan provides access. Start small with a single-use case, test thoroughly using the chat interface, then scale as confidence builds.

The system prompt deserves careful attention—it's the foundation of agent behavior. Clear, specific instructions yield better results than vague directives.

Check the official Make documentation for tutorials, credit usage details, and updates on feature evolution. The platform's help center includes module references and configuration guides that clarify setup details.