Make launched AI assistant Maia: assembling scripts with text
The Make platform introduced Maia, an AI assistant for designing and debugging integrations with text. The tool is available in Uzbekistan without third-party API keys.
On October 1, 2026, the Make platform launched public beta testing of the built-in AI assistant Maia, which allows you to design, configure and test automation chains using text commands. For businesses in Uzbekistan and Central Asia, this means a noticeable reduction in the technical threshold when creating integrations: now complex connections between external services, Telegram bots and CRM systems can be formulated in natural language directly inside the script editor, without setting up the mapping of each field manually from scratch.
What exactly has changed in the Make interface
Before the implementation of Maia, setting up any process required detailed manual labor from a specialist. It was necessary to independently select the necessary modules, determine the order of their execution, configure webhooks, manually register routers, compare variables from incoming data arrays and test the behavior of each step on test data packets. If the service returned an error due to a changed JSON structure or an incorrect variable type, the engineer had to open the execution history and verify the parameters manually.
The Maia tool is built directly into the Scenario Builder visual canvas. The assistant performs three main functions:
- Generation of script architecture from a text description:The user enters a task in natural language, and the system independently selects the necessary modules, arranges them in a logical sequence and creates basic branches through routers.
- Configuration of parameters and filters within modules:Maia determines the context of the task and fills in technical fields, formulas and data filtering rules, based on the structure connected services.
- Diagnostics and error correction:the assistant analyzes faulty scenarios, indicates the cause of the error (for example, an incorrect date format or a missing required field) and makes adjustments to the module configuration.
The tool is fully available to users in Uzbekistan and other countries in the region without any geographical restrictions. To work, you do not need to connect your own API keys from OpenAI or Anthropic: all request processing occurs within the Make infrastructure.
Who is this useful for: the practice of companies in Central Asia
In projects to automate business processes in Tashkent and other cities in the region, Telegram remains the key communication and sales channel, working in conjunction with CRM systems, warehouse databases and financial tables. The appearance of Maia speeds up the solution of typical application problems faced by sales departments, customer support and operational services.
In real practice, this opens up opportunities for quickly assembling several popular processes:
- Routing requests from Telegram: processing requests from corporate bots with the distribution of contacts among responsible managers in CRM depending on the selected region, transaction budget or type of request.
- Synchronization of accounting records systems:linking incoming payments from external payment gateways with status recording in Google Sheets or databases with automatic notification sent to the client and accountant.
- Control of failures in existing chains:automatic audit of integrations that have stopped transmitting data due to updates to third-party APIs or changes in required fields in lead collection forms.
Instead of involving a developer at each stage architecture design, the operations manager can formulate a task for Maia, receive a working integration framework and transfer it to technical specialists for final validation and connection of combat accounts.
Step-by-step guide: how to apply Maia in current processes
To test the tool and apply it to the company's workflows, just follow the following sequence of steps:
- Login to your Make account:open an existing script or create a new project on Scenario Builder canvas.
- Activate the Maia interface:Select the Assistant panel in the Scenario Editor. A starting promotional package of 70 requests is available for paid plans; a 30-day trial period with 45 requests is available for free accounts.
- Formulate the business logic of the task:describe the incoming event, participating services and the required result. For example: “Receive a contact from a Telegram bot, check the availability of a client in CRM by phone number, create a deal in the sales funnel and send a message to the manager with the details of the lead.”
- Check the created module structure:evaluate the correctness of the filters and field comparisons that Maia filled in automatically.
- Carry out a test run on real data:pass a test package of information through a script, track execution logs and, in case of errors, request Maia to analyze the faulty module.
- Fix the scenario in production mode:enable regular integration after checking access rights and the security of the transmitted information.
Tariffs, technical limitations and data security
Before introducing the tool into permanent operation of the company, it is necessary to take into account the financial model, limitations of the current beta version and security policy.
Pricing and quotas:Maia is available on all paid plans of the platform (starting with the Core tier starting at $9 per month, as well as Pro, Teams and Enterprise plans). For the first 30 days, users of paid subscriptions will receive 70 free promotional messages. New free accounts receive a 30-day trial period with 45 hits. After the promotional limits are exhausted, the use of the assistant is charged directly from the general pool of operations (Make operations / credits) of the organization. Loan write-off depends on the complexity of the prompt and the volume of the task completed.
Security of corporate data:For corporate clients, the issue of confidentiality is critical. Make applies zero-data-retention safeguards in relation to corporate compliance policies: script configurations, business logic diagrams and user data passing through scripts are not used by platform developers to further train language models.
Beta version limitations:At the current stage, Maia operates strictly within the confines of an active script editor. The assistant cannot autonomously analyze external child sub-scenarios and does not have direct access to isolated, stand-alone webhooks without first including them in the context of the current canvas.
Maia's place in the overall enterprise automation strategy
The emergence of conversational AI assistants in no-code environments is changing the speed of prototyping, but does not eliminate the need for systems data engineering. In practice, we see that the main difficulties in digitizing a business arise not at the stage of adding modules to the screen, but in the area of regulations: the purity of client databases, contact deduplication standards and control of employee access rights.
The Maia level tool removes the routine of initial assembly, allowing you to reduce the time from the idea of integration to its initial testing from several hours to several minutes. However, control over architectural integrity, contingency handling, and end-to-end analytics still requires skilled project oversight.
Frequently Asked Questions
Do you need to purchase OpenAI or Anthropic API keys to use Maia?
No, the tool is completely self-contained and does not require third-party API keys. Requests to Maia are billed directly through Make's billing system using internal subscription transactions.
Is our company and customer data used to train AI?
No. According to the official Make specifications, user data, transferred content and script configurations are protected by a zero retention policy and are not transferred to train underlying machine learning models.
Can Maia automatically fix an error in a script that is already running?
Maia can diagnose the cause of a failure in the active editor and suggest the correct configuration of the problematic module. However, the assistant only works within an open script and cannot yet check external webhooks or isolated sub-scripts without the participation of an engineer.
The NDC (Digital Business Solutions) team helps companies in Uzbekistan design reliable automation architectures and implement integration solutions taking into account real business processes - contact us to analyze your automation tasks.
