Make AI Agents + Claude Sonnet 5 and Fable 5: What the 2026 Updates Mean for Your Small Business

Every month, Make publishes its list of updates. Most of the time, an SMB owner skims the email and moves on: too much jargon, not enough "what does this mean for me?". The 2026 release deserves attention, though, because it touches the core of what makes automation useful: the intelligence of your scenarios.

Make has integrated Anthropic's latest Claude models, Claude Sonnet 5 and the creative model Claude Fable 5, into its AI Agents, alongside a set of improvements that make the tool easier to pick up. Here is what changes, translated into plain business language.

Quick Refresher: What Make Does and Why It Matters

Make is a no-code automation platform. You visually connect your applications (inbox, CRM, calendar, invoicing, social media, spreadsheets) and create "scenarios" that execute tasks on your behalf: send a follow-up, create a contact record, publish a post, route an attachment to the right folder.

Until now, those scenarios were mostly rigid: if this condition, then that action. Effective for predictable tasks, but powerless when the situation requires judgment. That is exactly the wall that AI Agents break down. If you are discovering Make, start with our Make vs n8n vs Zapier comparison to see how it stacks up against competitors.

The Real News: Claude Sonnet 5 and Fable 5 in AI Agents

This is the headline announcement. Two new Anthropic models arrive in Make:

• Claude Sonnet 5: Anthropic's latest reasoning model. It handles tasks that require analysis, sorting, and decision-making better than its predecessors. Exactly what an automation agent needs when the situation is not binary.
• Claude Fable 5: a creative-class model ("Mythos") geared toward content and narrative generation. Useful whenever you need to write, rephrase, or storytell: product descriptions, article drafts, personalized responses.

You can use them in two ways: inside a Make AI Agent, or via the Anthropic Claude app in Make. The difference compared to a simple ChatGPT call in a scenario? An agent decides and chains actions: it reads, understands intent, retrieves information from your other connected apps, then acts. We move from automation that executes to automation that reasons.

This is the same shift toward agentic AI we observe elsewhere, from Dust for structured enterprises to AI agents for SMB customer support. Make has the advantage of putting it within reach of a micro-business, without a developer.

Three Concrete Use Cases for a Small Business

Sorting and responding to customer requests. An agent reads each incoming email, classifies the request (quote, support, common question, complaint), drafts an appropriate response with Fable 5, and assigns it to the right person. You only handle what truly needs your attention.

Qualifying prospects. A contact form triggers an agent that analyzes the request with Sonnet 5, enriches the record, calculates a priority score, and creates the opportunity in your CRM. Paired with a quote automation workflow, the hot prospect receives a proposal the same day.

Recurring content production. Fable 5 generates descriptions for your new products, drafts for your posts, or responses to customer reviews, in your tone, from a simple trigger. Combine with our guide on automating Google reviews.

Other Updates That Matter (and Those That Matter Less)

Make shipped a batch of improvements. Not all carry the same weight for a small business. Let's be honest.

Useful in daily operations:

• AI in the MCP Client app. You can run AI steps (via OpenAI or Claude) directly in your MCP Client workflows. MCP (Model Context Protocol) is the standard that lets AI models connect cleanly to your tools; to understand why it is structurally important, read our explainer on the MCP protocol for SMBs.
• Extended content search in Make Grid. Find a module, a field, a webhook URL, or a note across all your scenarios in seconds. Trivial on the surface, invaluable once your automation library grows.
• Disconnected module configuration. You can configure and test a module before connecting it to the scenario. In practice: fewer errors, more isolated tests, a calmer building experience.
• Favorite apps and modules in search, and unified help search. Comfort features that save minutes on every scenario.
• User roles page: administrators can see all system roles and their permissions. Useful as soon as multiple people touch the automations.

Mostly for large organizations:

• Databricks in Make (execute SQL, trigger jobs, manage pipelines, load files into Unity Catalog). Powerful, but reserved for companies with real data infrastructure. A typical SMB will not need it, and that is fine.

The thread running through this release: making intelligent automation accessible to non-technical users. That is the right direction for our reader.

Limitations to Keep in Mind

You do not switch your entire company to AI agents overnight.

An AI agent costs operations. Each model call consumes Make "operations" and, depending on configuration, tokens on the AI provider side. At high volumes, the bill adds up. Measure cost per task before scaling.

AI makes mistakes sometimes. An agent that drafts a customer reply should, initially, remain supervised (draft approved by a human) before any automatic sending. Automate the preparation, keep control over the final decision until trust is established.

The learning curve remains real for advanced workflows. The updates lower the barrier to entry, but building a reliable agent takes method. Start with a single use case, measure, then expand.

Make vs. Its Alternatives

Make vs Zapier. Zapier is simpler to start with, but Make offers more power and much better value for money on complex scenarios. Our detailed comparison helps you decide based on your profile.

Make vs n8n. n8n attracts technical profiles (self-hosting, total control), while Make prioritizes accessibility and a massive library of ready-to-use apps. For an SMB without a technical team, Make is often the pragmatic choice.

Make vs a generalist AI alone. An assistant like Mistral or ChatGPT helps you on a case-by-case basis; Make industrializes AI by connecting it to your tools and running it without you. The two are complementary.

Make's Batemark Score

We give Make a Batemark Score of 82/100 (Excellent).

• Value for money: excellent. Free plan, usage-based pricing, rare power at this price point.
• Ease of use: good and improving. The 2026 updates lower the barrier; advanced workflows still require method.
• Power and features: exceptional. AI Agents, MCP, thousands of modules: the ceiling is very high.
• Integrations and ecosystem: exceptional. One of the broadest app libraries on the market.
• SMB relevance / France: very good. No-code, accessible, GDPR manageable; interface partially in English.
• Support and community: good, varies by subscription tier, active community.

Make is one of the best choices for automating a small business without hiring a developer. Find its full profile, tutorials, and alternatives in our AI tools library, Make page.

Should You Start Now?

If you already use Make: yes, test an AI Agent with Sonnet 5 on one recurring and variable task (sorting your emails is an excellent starting point). If you do not use Make yet: these updates make the timing ideal to begin, as long as you aim for one specific use case rather than trying to automate everything at once.

Intelligent automation is no longer reserved for tech companies. With AI Agents and the new Claude models, it fits in the hands of a motivated small business owner. Start small, measure the time saved, and let the system grow with you.

To go further, compare the major platforms in our Make vs n8n vs Zapier 2026 guide.