AI Fill Tasks turns a short project brief into a task tree you can edit before saving. It uses OpenAI's gpt-4o-mini model via a server function, so generation takes a few seconds and counts against your monthly AI quota.

Goal: by the end of this you'll have generated, reviewed and saved a set of tasks for one project.

AI Fill Tasks and the AI quote draft are the two features in FreelancerFlow that call a large language model. Zeka's suggestions and the Money tab's prioritization are rule-based — not LLM.

Prerequisites

A project to attach tasks to. See Creating your first project.
You're on the trial or Pro. Free-tier users can read but can't create tasks.

Step 1: Open AI Fill Tasks from the project

From the project detail screen, tap AI Fill Tasks. The first screen asks for a project brief.

Step 2: Write the brief

Describe what the project is in two or three sentences. The model uses three things to tailor the output:

The brief text you write.
The project's name, billing type, and budget.
The persona you picked during onboarding (Designer, Developer, Marketer, etc.). It affects task naming and estimates.

The brief detector also looks for project-type hints (web, mobile, interior, design, API) and adjusts depth. An interior-design brief, for example, gets longer task trees than a single-page website.

Tap Generate.

Step 3: Review the generated output

After a few seconds the screen shows:

Project analysis — total estimated hours, a complexity rating (Low / Medium / High), a risk level, and a list of suggested phases.
Tasks — usually six to ten, each with a name, an hour estimate, and a one-line "why this matters" note.
Next actions — three to five suggested follow-ups (write the contract, send the kickoff email, etc.).
Cost summary — how many AI calls you have left this month.

Each task can be edited inline. Tap a task to rename it, change its estimated hours, or remove it. The generation flow does not save anything until you tap Save tasks.

Step 4: Save or regenerate

Save tasks — adds the visible tasks to the project. Order is preserved. Status is To do. Priority defaults to Normal.
Regenerate — re-runs the model on the same brief. Costs another AI call.

If you regenerate, the previous draft is discarded. Anything you'd already edited is lost; if you want to keep some, save first then regenerate from scratch for a separate pass.

What the model is good at

Splitting a typical service project into sensible phases.
Giving rough hour estimates you can sanity-check.
Suggesting follow-up actions you might miss.

What it isn't good at

Knowing your specific client's preferences.
Estimating budget — that comes from you.
Anything that depends on private context (NDA scope, third-party blockers).

Treat the output as a fast first draft, not a finished plan.

Troubleshooting

"Quota exceeded." You've used your monthly allotment. See AI quota explained.
"Authentication required." Sign in. AI Fill Tasks requires an account; anonymous sessions can't call the server.
AI replied in the wrong language. The model detects language from the brief, but you can force the output language in Settings → AI Preferences.
Tasks look generic. Add specifics to the brief — names, deliverables, deadlines. The output mirrors the input's specificity.

AI Fill Tasks walkthrough