Functionality

PRO-dela AI features: how to set tasks faster and not lose details

A complete catalog of PRO-dela text AI features: smart task creation, splitting a project into stages and tasks into subtasks, splitting and merging tasks, improving the technical specification — and how it is enabled by the administrator and LLM provider.

Updated August 17, 2026
PRO-dela AI features: how to set tasks faster and not lose details

AI in PRO-dela is not a separate assistant on the side, but several targeted functions within already familiar screens: the task creation form, the project page, and the task card. Each of them takes on a specific piece of routine — coming up with a short title from free text, breaking down a raw technical specification into stages or subtasks, splitting an overgrown task into several, merging several small ones into one, and polishing a draft technical specification into a decent form. None of them work "out of the box": each is enabled separately by the PRO-dela administrator, and only when a working access key is configured for the selected LLM provider.

It is important not to confuse this with voice input — dictating text into fields and the editor: that is a separate function with its own logic (see the question at the end of the article). Below is a catalog of the platform's text AI features: what each one can do, what data it works on, what happens if the AI is disabled or unavailable, and where the line is drawn between "a draft you confirm" and "a result that is applied immediately".

What "optional" means: the PRO-dela administrator decides, not the company

Each of the six functions described in this article has its own independent toggle, and by default they are all disabled. They are enabled not by the company owner or an administrator within the project, but by the administrator of the PRO-dela platform itself — in a closed management section inaccessible to regular users and company owners. There, for each function, an LLM provider is selected separately (currently OpenRouter and DeepSeek are supported) and a working access key is specified.

Practical consequence: until at least one of the two conditions is met — the function is enabled and the provider is configured — the corresponding button in the interface is not displayed at all. It does not become gray and inactive with a "coming soon" hint: the control element is simply absent from the screen, as if the function does not exist. So if you do not see, for example, the button for splitting a project into stages, it is not a bug or a tariff restriction, but a matter of the current platform configuration.

The second consequence is more important for those who have already seen one of the AI buttons in the interface: this does not mean that all the others are enabled as well. Each function is a separate flag and, if necessary, a separate model. Further in the article, this dependency is clarified for each function separately.

Intelligent task creation from free text

A task in PRO-dela is created from the same form — it is the same on the task list page, the project page, and the stage page. If the AI task creation function is disabled or the provider is not configured, the form works as usual: the "Title" field and, optionally, a description. As soon as the function is enabled by the administrator and the provider is ready, a free text mode appears in the same form.

In this mode, you simply describe what needs to be done — the way you would formulate a task for a colleague. PRO-dela sends this text to the server along with the other form fields (project, deadline, priority), the LLM generates a short and clear task title, and the entire original text is saved without abbreviations in the task description — it is not "compressed" or lost, only the title becomes shorter.

Separately, and also at the administrator's discretion, you can enable recognition of the assignee and status directly from the command text — these are two independent toggles. In this case, the LLM does not choose a specific user or status on its own: it only returns a text request ("to Maria", "in progress"), and the matching with real data is done by PRO-dela — and only among participants and statuses of the current project, not across the entire platform. If the match is ambiguous — for example, there are two participants with similar names in the project — the assignee is not assigned, and instead a warning is returned; the same for the status. The right to assign an assignee or change the status is checked separately from the right to create a task — one does not replace the other.

An AI error never blocks task creation. If the function is disabled, the provider is unavailable, or the model's response comes in an unexpected form, PRO-dela uses a fallback scenario: the title becomes the first non-empty line of your text, truncated to the field length, and the full text is still saved as the description. The task will be created in any case — in the worst case, just with a less neat title. How statuses, participant rights, and the normal task lifecycle work in PRO-dela without AI is a separate topic, covered in detail in the article "Tasks in PRO-dela".

AI splitting: project into stages, task into subtasks

Here there are two similar but different functions — each has its own toggle at the administrator, its own text source, and its own place in the interface.

Splitting a project into stages

The source is the project's technical specification. If it is empty, PRO-dela will immediately ask you to fill in the technical specification first and will not contact the LLM. The split button is shown only when the function is enabled by the administrator and the provider is configured, the user has the right to manage project stages, and — an important condition — the project does not yet have a single stage: this is a project start tool, not a permanent button on top of an already built structure.

The result comes as a draft: a window with a numbered list of proposed stages, each with a title and a brief description. While the request is in progress, the interface shows the waiting time — generating a meaningful plan can take noticeable time. Nothing is created automatically: stages appear in the project only after you explicitly confirm the list with the create button. If files are attached to the project's technical specification, the draft will come with a warning — their contents are not analyzed, data from files must be transferred to the technical specification text manually. What a project stage is and how it lives on without AI is in the article "Projects and stages in PRO-dela".

Splitting a task into subtasks

The same principle, but the source is the task's description (which is also its technical specification), not the entire project. The button appears on the task card only when the function is enabled by the administrator and the provider is configured, you have the right to edit it, and the task does not yet have a single subtask. As with stages, the proposed list of subtasks is a draft with titles and descriptions that must be explicitly confirmed; attachments to the task do not participate in the splitting, only the saved description text is taken into account.

Splitting one task into several

If a task has grown so much that it actually contains several independent pieces of work, PRO-dela — with the task splitting function enabled by the administrator and the provider configured — can offer a draft of several new tasks based on the text of the current description. As in other cases, nothing happens without your confirmation: first you see the proposed titles and descriptions, and only an explicit button press starts the creation.

After confirmation, PRO-dela creates each proposed task in the same project, transfers the same stage links, the same assignee, and the same participants as the original task had, copies attachments — and archives the original task. This should be understood in advance: splitting does not "add" tasks next to the old one, but replaces one task with several, and the original task is no longer displayed in the active list. The right to create new tasks in the project and the right to archive the original task are checked separately from each other.

Merging several tasks into one

The reverse scenario — when several small and similar tasks are more convenient to manage as one item with subtasks inside. In the project task list, you can select from two to five tasks of the same project, each of which has a description filled in, and — if task merging is enabled by the administrator and the provider is configured — request a draft of the merged task from the AI.

The draft contains the proposed title and description of the new task and a list of subtasks — one for each original task, with text taken from its description. After confirmation, PRO-dela creates one new task with these subtasks, combines participants from all original tasks, copies attachments, and links the new task only to those stages that were common to all merged tasks — if there are no common stages, the task is created without a stage link, and a separate warning is issued. All original tasks are then archived. If some of them already had their own subtasks, they are not transferred automatically and remain only with the archived original tasks. The right to modify and archive is checked for each selected task individually, not once for the entire list.

Improving the technical specification — for a task and for a project

The same function works in two places: on the task card — for its description, and on the project tab — for the project's technical specification as a whole. In both cases, the logic is the same, and the function is enabled by the administrator separately from the other five and also requires a configured provider.

The improve technical specification button is available only when the source text is filled in — on an empty description, the function will not run; instead, a request will appear to first add at least some text. The AI result is inserted into the same editor as a draft: you see the proposed variant, can compare it with the original (within the editing session, there is a return to the previous text), and, just as with regular manual editing, save the changes with a separate button. The request to the AI itself does not write anything — until you press "Save", the previous text remains in the database. As with splitting into stages, files attached to the task or to the project's technical specification are not read by the model — only the main text in the editor.

What the administrator configures for each function

All six functions are managed from one section of the PRO-dela admin panel, but are configured independently of each other:

  • a separate "enabled / disabled" toggle for each function;
  • its own provider and model per function — for example, you can use one model for task creation and another for improving the technical specification;
  • the access key is set once per provider and reused by all functions configured to it — you do not need to enter it separately for each of the six functions;
  • for each function, you can set your own system prompt — optional; if left empty, the built-in default is used;
  • before saving settings, the administrator can check the provider's operability directly from the panel, without going into the task and project interface.

From this follows a practical point: the specific model that processes your text is not a fixed fact of the platform, but the current choice of the administrator. It can change over time, differ for different functions, and does not guarantee a certain percentage of result accuracy — the draft should be reviewed before confirming it, for any function from this article.

What PRO-dela AI features do not do

  • Not enabled by default. All six functions start disabled and remain unavailable until the PRO-dela administrator enables the needed one and configures the provider — this is a platform decision, not a company one.
  • Do not work on an empty source. Without task text, project technical specification, or task description, each function returns a clear error instead of inventing something out of nothing.
  • Do not read attachments. Files attached to a task, project, or its technical specification are not analyzed in any of the described functions — only the main text is taken into account, and a separate warning is issued when attachments are present.
  • Do not guarantee a specific model or result accuracy. The provider and model are the administrator's choice, which can change; the platform does not promise a percentage of accuracy or result quality — drafts should be checked, not applied blindly.
  • Do not override normal access rights. Assigning an assignee, changing status, creating and archiving tasks are still checked against your rights in the project — AI does not act around them.
  • Splitting and merging tasks archive the original tasks. This happens only after explicit confirmation of the draft, not at the moment of generation; the archived task does not physically disappear and is subject to the general rules of the PRO-dela task archive, not a separate AI-specific restoration.

How to start working with AI features

  1. Check with the PRO-dela administrator which of the six functions are currently enabled and whether the provider is configured — you cannot enable them for a single company yourself.
  2. Start with the most common scenario — intelligent task creation: it does not require a preliminary technical specification and immediately saves time on formulating the title.
  3. For a new project, first fill in the technical specification, and only then try splitting into stages — before creating the first stage manually.
  4. For one overgrown task, use splitting into several, and for several small and similar ones — merging; for a new task with an empty subtask list — splitting into subtasks.
  5. Any draft technical specification can be preliminarily run through the improve technical specification function and compared with the original before saving the final text.

Frequently asked questions

Why do I not see any AI buttons in PRO-dela?

Most likely, the needed function has not yet been enabled by the PRO-dela administrator or the LLM provider is not configured for it — for each of the six functions, this is checked separately. Check the status with the person who administers your platform.

Can AI create or change something without my confirmation?

No. Task creation always happens by your explicit action — AI only helps with the title and, if separately allowed, tries to recognize the assignee and status. Splitting into stages and subtasks, splitting, merging, and improving the technical specification — all of them first show a draft and do not apply anything until you press the confirmation button.

What happens if the AI provider is unavailable at the moment of task creation?

The task will still be created — a fallback scenario will work: the title will be the first line of your text, and the full text will be saved in the description. Task creation is not blocked by errors or AI unavailability.

Where do the original tasks go after splitting or merging?

They are archived — just like when archiving a task normally in PRO-dela — and not permanently deleted. This happens only after you have confirmed the draft with the create button, not when the AI merely proposed a variant.

Does AI take into account files attached to a task, project, or technical specification?

No, in none of the functions described in this article. The model works only with the main text — the title, task description, or technical specification text; if attachments are present, the system separately warns that their contents need to be transferred to the text manually.

Is this the same as voice input?

No. Voice input — converting speech to text in the active field — is a separate function with its own logic, described in detail in the article "Voice control of tasks". This article is only about text AI features: creating, splitting, merging, and improving tasks and projects.

Try AI features in your project

If you already have access to PRO-dela, the best way to understand what of the described here is actually enabled for you is to open the task creation form or the project page and see what controls are there. If the platform is not yet familiar to you — register in PRO-dela and create your first project, and then together with the administrator decide which AI features are worth enabling for your team.

The full list of tariffs and how limits are distributed among them is on the tariffs page.