What do the JUL'26 Make.com updates really mean for your team?
We break down what matters, what to act on, and where it impacts your operations.<\/b>
<\/p>\n
https:\/\/help.make.com\/databricks-app-is-now-available-enterprise-plans<\/a><\/b> https:\/\/help.make.com\/claude-sonnet-5-is-now-available<\/a><\/b> https:\/\/help.make.com\/new-model-support-for-opus-5-and-gemini-36-flash-greenhouse-recruiting-app-update-and-lower-credit-costs-for-open-ai-models<\/a><\/b> https:\/\/help.make.com\/openai-gpt-55-is-now-available-in-make<\/a><\/b> https:\/\/help.make.com\/new-version-of-api-endpoint-for-credential-requests-scenario-usage-tracking-connection-filtering-and-app-updates<\/a><\/b> https:\/\/help.make.com\/anthropic-claude-model-deprecations-on-june-15-2026<\/a><\/b> https:\/\/help.make.com\/claude-fable-5-is-now-available<\/a><\/b> https:\/\/help.make.com\/team-edit-permission-removed-from-team-member-role<\/a><\/b> https:\/\/help.make.com\/2026<\/a><\/b> https:\/\/help.make.com\/unified-canvas-navigation<\/a><\/b> https:\/\/help.make.com\/trash-for-deleting-scenarios-module-renaming-ai-error-explanations-new-ai-models-and-more<\/a><\/b> https:\/\/help.make.com\/new-mms-options-app-updates<\/a><\/b> https:\/\/help.make.com\/openai-gpt-54-nano-and-mini-now-available-in-make<\/a><\/b> https:\/\/help.make.com\/introducing-make-skills<\/a><\/b> https:\/\/help.make.com\/dynamic-labels-for-http-and-sleep-modules<\/a><\/b> https:\/\/help.make.com\/introducing-private-spaces<\/a><\/b> https:\/\/help.make.com\/bitbucket-modules-deprecation-new-ai-models-and-more-app-updates<\/a><\/b> https:\/\/help.make.com\/new-mondaycom-ai-modules-mcp-client-update-and-more<\/a><\/b> https:\/\/help.make.com\/claude-opus-47-now-in-make<\/a><\/b> https:\/\/help.make.com\/make-ai-agents-new-mcp-tools-are-now-available<\/a><\/b> https:\/\/help.make.com\/introducing-mcp-toolboxes<\/a><\/b> https:\/\/help.make.com\/connect-your-third-party-automation-and-ai-agents<\/a><\/b> https:\/\/help.make.com\/content-search-in-make-grid<\/a><\/b> https:\/\/help.make.com\/unified-help-search-and-deprecation-of-filters<\/a><\/b> https:\/\/help.make.com\/action-required-for-stripe-users-new-gemini-module-and-more-app-updates<\/a><\/b> https:\/\/help.make.com\/introducing-user-roles-visibility<\/a><\/b> https:\/\/help.make.com\/functions-standalone-module-setup-scenario-note-cloning-and-more<\/a><\/b> https:\/\/help.make.com\/the-make-cli-is-now-live<\/a><\/b> https:\/\/help.make.com\/google-chrome-app-and-browser-extension-deprecation-on-august-31-2026<\/a><\/b> https:\/\/help.make.com\/sub-agents-and-tool-output-filtering-are-now-available-in-the-make-ai-agent-new-app<\/a><\/b>
With the Databricks app on enterprise plans, you can automate SQL execution, job and pipeline monitoring, and file movement into Unity Catalog volumes directly from business events, schedules, or triggers, then route results to Sheets, email, Slack, or BI tools. This improves operational responsiveness and governance if you use a dedicated service account and align job\/pipeline access with Unity Catalog permissions. Plan for service principal security, environment separation, and error handling so automations stay maintainable.<\/p>
You can now use Claude Sonnet 5 in Make with a 1,000,000 token context, letting your automations process large PDFs\/exports without chunking, and handle multi-step routing decisions more reliably. Built-in output caps improve control for scheduled or agentic runs. Sonnet-tier cost control makes high-volume runs more predictable, but your team should review credit usage and output truncation risk, since the new tokenizer may change token counts.<\/p>
You can now use Claude Opus 5 and Gemini 3.6 Flash for deeper, multi-format extraction and \u201call-in-one\u201d analysis without splitting prompts, which should reduce workflow complexity and improve data quality for routing, CRM updates, and finance record keeping. Claude\u2019s agent coordination and configurable thinking effort can improve accuracy, but increase cost and require tighter governance. Greenhouse recruiting switches to OAuth with harvest v3: migrate before Aug 31. OpenAI credit costs for some models drop, lowering run cost if you use those modules.<\/p>
You can now upgrade agent-style automations with GPT-5.5 for planning across many steps, chaining tool calls, and handling very large inputs, which improves run completion and output consistency in workflows that process documents, spreadsheets, or code. Use this to reduce manual intervention and variance in tool-driven processes. Watch for higher context and cost tradeoffs, tighter governance needs for tool selection, and regression testing since behavior may change. Best value is in complex, multi-stage tasks, less in simple field updates.<\/p>
You need to migrate to API v2 for credential requests before June 10, 2026 or risk breaking integrations that create connections programmatically. Expanded scenario usage tracking helps you map data stores, structures, and custom variables to scenarios, improving governance and troubleshooting. Connection filtering by \u201cmy connections\u201d speeds operational cleanup. App updates improve AI context accuracy with auto timestamps and expose token-based credit usage, but require builders to re-check prompt length and cost controls.<\/p>
If you use the Anthropic Claude Sonnet 4 or Opus 4 models inside Make scenarios, your automations will start failing after June 15, 2026. Review and replace those model selections with the supported replacements, and re-check pricing because cost may change. In practice, update modules proactively, test in a non-production scenario, and add monitoring so you catch model errors before they impact lead qualification, support triage, or proposal generation workflows.<\/p>
You can now build Make scenarios that push far more content through one step, including long-context document, image, and code analysis, with large outputs mapped directly into downstream modules. That reduces brittle \u201cmulti-step\u201d workflows and improves resilience by returning refusals instead of hard errors for high-risk inputs. Plan for governance: code-generation accuracy, prompt versioning, and Anthropic\u2019s 30-day data retention, plus limits still needed to keep retries and adoption stable.<\/p>
Your team members will lose the ability to rename teams in public Make.com environments after Aug 2, 2026; only the \u201cedit teams\u201d permission will remain, so they cannot make broader team edits. Private environments are unaffected. Ensure the users who need this control are assigned the admin role. Operationally, this improves governance and reduces accidental configuration changes, but it may slow collaboration if roles aren\u2019t updated.<\/p>
You get better control and scale levers for AI-driven automations: Make AI Agents now support sub-agents and tool output filtering, with newer Opus 5 and Gemini 3.6 Flash models plus lower OpenAI credit costs. Operationally, you should use Grid\u2019s faster dependency\/content search and scenario recovery, and adopt private spaces for sensitive workflows. Watch for breaking changes: Stripe action items, Chrome app deprecation, and model\/module updates requiring revalidation.<\/p>
If your team builds complex automations in Make, improved unified canvas navigation reduces friction across devices. Auto-detected mouse versus touch controls and saved preferences help builders switch between laptops, tablets, and mixed setups without reconfiguring, which can lower errors and speed up scenario maintenance. The main risk is inconsistency in collaboration if builders assume different navigation behaviors; set clear team defaults and standardize builder guidance for shared projects.<\/p>
You get safer operations for scenario lifecycle with a recoverable \u201ctrash\u201d flow, plus clearer maintenance with AI-generated error explanations inside the scenario. New OpenAI GPT 5.6 and Claude model support lets you route steps to the right model for cost, speed, and context, improving throughput for classification, logic, and document-heavy work. Session timeout controls improve governance for long-running automations. Rename-by-double-click helps team handover. Main risk: AI explanations must be reviewed to avoid masking root causes.<\/p>
You get more control over Make Managed Services with API-managed child organizations and visibility into credit transfer and consumption, which helps governance, cost allocation, and scaling across teams or regions. Tooling updates also matter operationally: new AI model availability, renamed Google app naming, corrected Fathom limits, and multiple connection-type and OAuth refresh improvements (notably Zendesk) reduce fragile auth failures. Migration deadlines for legacy Zendesk connections add a planning and testing need.<\/p>
Your team can now use OpenAI GPT-5.4 mini and nano in Make, with faster, cheaper options for high-volume, multi-step and agentic workflows. Practically, you can route subtasks to nano for repetitive extraction\/classification and use mini for coding or multimodal reasoning, while keeping larger models for orchestration. This improves cost and run-time for triage, intent, ranking, and routing. Keep governance by fixing prompts, validating outputs, and monitoring drift in routing accuracy.<\/p>
Make Skills lets your team give AI a file-based, reusable \u201chow-to\u201d for building and editing Make scenarios, including configuring modules, mapping data flows, and working with the Make MCP\/API. This can speed up scenario creation and standardize best practices, but you must control what it can change, review generated logic, and test connections and data structures before production to avoid governance and maintainability risk. Biggest value: faster scaling with consistent patterns. Smaller value: isolated tweaks.<\/p>
Dynamic labels on HTTP and Sleep modules make your automation canvas self-documenting: you can see request method plus endpoint path, and the configured delay time, without opening settings. That improves review speed, reduces errors when refactoring complex flows, and helps governance by making intent visible during audits. It is minor functionality-wise but high leverage for maintainability. Just ensure endpoint\/delay labeling stays aligned with how you expect operations to behave.<\/p>
Private Spaces let you isolate specific scenarios and connections so builders and teams can\u2019t see each other\u2019s sensitive automation settings, improving operational security and reducing accidental interference. Credit limits per space add governance for cost control. Enterprise admins can view limited info in read-only mode, balancing oversight with privacy. Use this for confidential workflows (pricing, lead routing, customer data) and high-risk experiments. Main consideration: document dependencies and standardize access so troubleshooting and scaling remain manageable.<\/p>
You should expect more platform churn than usual: Bitbucket trigger modules are largely deprecated due to API changes, so validate scenarios and reselect repositories and required event fields to avoid silent failures. Several LLM modules gained new models and reasoning controls, enabling better cost\/token control but increasing governance needs for parameter drift. Finance apps added document modules and date\/email fields, improving correctness. Review agent betas for unattended risk.<\/p>
Your automation stack gets more flexible AI building blocks, including monday.com conversational agents, an MCP AI flow path, and an \u201cAI provider\u201d option that can run without external API keys via Make credits or with your own for spend control. HubSpot auth is more secure and ready for v1 retirement, while triggers and bug fixes reduce brittle edge cases (email in shared folders, Sheets headers). New modules across Reddit\/Discord\/GitHub\/Pipedrive expand orchestration, but you\u2019ll need governance for model choice, credentials, and prompt\/memory consistency.<\/p>
You can now run Claude Opus 4.7 inside Make for tougher multi-step automations where vision, instruction fidelity, and dependable tool use matter. For B2B workflows, this can improve parsing of screenshots, PDFs, invoices, and dashboards, reduce human rework, and support larger knowledge contexts for agents without splitting. Use xhigh only where quality justifies cost and latency. Governance still matters: validate outputs, monitor tool calls, and keep prompts and model settings versioned for maintainability.<\/p>
You can now connect MCP tools directly into Make AI Agents, expanding beyond standard apps without building custom HTTP scenarios. For B2B teams this enables more complete lead research, enrichment, and ticket handling across systems, using the same MCP servers your team already trusts in AI assistants. Strategically, this improves resilience and reduces duplication, but you must govern tool permissions, keep MCP endpoints reliable, and maintain clear decision boundaries so agents stay controlled and debuggable.<\/p>
MCP toolboxes give you controlled, shareable bundles of AI-accessible \u201ctool\u201d scenarios inside Make, with per-toolbox keys and unique server URLs. You can restrict what each AI agent or user can do, see exactly which tools ran, and standardize cost-effective, reliable AI decisioning. For RevOps, this enables governed AI workflows across CRM, data, forms, and support systems, but you must invest in toolbox design, naming clarity, and access management to avoid sprawl and accidental privilege creep.<\/p>
You can now view third-party automations and AI agents alongside your Make scenarios in a single grid map. That\u2019s a big governance and scalability win: you can trace dependencies, spot where AI and PII flow, and audit usage spikes back to the exact providers driving cost. You also get faster impact analysis when URLs, IDs, or apps are referenced across systems. The risk is a false sense of coverage if external dependencies aren\u2019t correctly mapped.<\/p>
You can now search your entire Make Grid landscape by the content that matters, not just scenario names. This lets your team quickly find where a module, field ID, webhook URL, scenario notes, or even AI agent prompts are used before editing, reducing breakage risk and speeding governance reviews. It also improves maintainability and onboarding by making dependencies and context discoverable. Watch for sensitive prompt or URL exposure when broad-access users use global search.<\/p>
You get one consistent help search across the scenario overview and builder, reducing time lost to scattered docs and supporting faster, safer scenario changes. Separately, four data store \u201cpattern\u201d filters are deprecated for new builds. Existing scenarios keep running, but you should update any templates and future automations now, using available filters or moving equivalent pattern logic into a follow-up module to protect maintainability and prevent silent gaps in new scenario behavior.<\/p>
You need to reauthorize Stripe to regain hosted invoice and payment link creation, and switch GoHighLevel connections from retired v1 API keys to OAuth 2.0 to prevent sudden module failures. Lexoffice voucher category IDs must be updated by Aug 31, 2026. Gemini is rebranded with a new video generation module enabling chained, conversational edits. Several connectors were hardened for reliability (Sheets\/Excel\/Zoho\/Workday\/ClickMeeting\/WooCommerce). The main risk is breakage from permission, auth, and schema changes.<\/p>
You now get a dedicated Org > User roles page that lets your administrators see all default system roles for each team and drill into full role details. This improves governance and reduces permission mistakes when onboarding or changing access for automation builders, ops users, and reviewers. Operationally, it helps you standardize least-privilege, document who can build or edit workflows, and audit risk faster. The main consideration is keeping internal processes aligned with Make\u2019s role definitions over time.<\/p>
You can now do more logic directly in mapping fields with arraydiff, arrayintersect, set, and escapejson, reducing module sprawl, latency, and credit use while improving consistency of data transformations. Module pre-configuration and scenario preview lets you test outputs before wiring routes, boosting reliability and governance. Scenario cloning now copies notes, improving handoffs and maintainability. The main risk is more complex mappings, so enforce naming and validation to prevent hidden logic drift.<\/p>
The Make CLI gives your developers and AI agents programmatic control over scenarios, connections, data stores, and webhooks from the terminal, making automation changes fit naturally into Git and CI\/CD. This can improve resilience and governance by enabling peer review, versioned deployments, and repeatable environments, but you must protect credentials and test changes outside production. It\u2019s high value for teams that manage many scenarios; it\u2019s less relevant if you only do occasional manual edits.<\/p>
If any of your scenarios rely on the native Chrome app or Chrome browser extension, they will stop working after August 31, 2026, and Make doesn\u2019t provide a drop-in replacement. You should inventory every scenario that uses browser capture\/extension actions, replace with supported alternatives (for example, API-based or form\/email based triggers), and test end to end before the deadline to avoid workflow downtime and adoption risk.<\/p>
You can now build richer multi-agent workflows in the Make AI Agent app by using sub-agents as tools, delegating reasoning to specialists, and keeping context via conversation IDs across runs. Tool output filtering lets you exclude irrelevant fields to reduce token\/credit spend and improve focus, with sensible defaults. Plan-wide availability helps scaling. Watch one-level nesting limits, and note filtering is unavailable in scenarios, MCP, and knowledge tools, so design around those constraints.<\/p>