Operations platform · 2025–2026
Pipa Colab
The operations app behind an animation studio's freelance network. It handles task planning and assignment, collaborator payouts in two currencies, invoicing compliance and an AI forecast of the cash each month will need.
Demo with fictional data.
The situation
The monthly payout to a large freelance network was assembled by hand across spreadsheets and chats, with two currencies, fees, exchange rates and local invoicing rules.
What changed
Tasks, payouts and invoices live in one system. Collaborators see exactly what they'll be paid, admins close the month from a single table, and an AI forecast says how much cash each currency will need.
Built for an animation studio that works with a network of freelance animators, illustrators and editors across several productions at once.
“Closing the month used to mean spreadsheets and endless messages. Now every collaborator knows what they'll be paid, and we close payouts from a single screen.”
Goals
- Replace a manual, error-prone monthly payout process for a large group of freelancers.
- Handle mixed USD and local-currency payments, exchange rates, fees and balances carried from month to month.
- Guide collaborators through local electronic-invoicing requirements.
- Move scattered conversations about tasks into one place, with notifications.
What it does
- A planning board with list and kanban views, grouping by status, week, payout period or work area, and nested AND/OR filters with saved views.
- A collaborator area with itemised payouts, payment method setup, payment history and a step-by-step invoicing guide with invoice upload.
- An admin payday table per period covering available and paid amounts, fees, exchange rate and final balance, plus CSV imports and month rollover.
- Comment threads with @mentions and a status history on every task.
- Exports to PDF, Excel and Google Sheets.
- Email, Slack and in-app notifications, with daily reminders for tasks starting or due.
- Operations monitoring for the email queue and the health of background functions.
AI and automation
- An AI funds forecast estimates how much cash each currency will need next period from each collaborator's history. It returns a recommended amount with a safety margin, a confidence level, risk factors and per-person insights, and answers follow-up questions in a chat.
- Smart assignment ranks the best candidates for a task by workload, specialisation and client continuity. The AI then picks and explains the top three, choosing only from that shortlist so it can never invent a person.
- Every AI suggestion is logged, and the forecast falls back to a deterministic estimate if the model does not answer.
- A queued email pipeline handles rate limits, retries and backoff, with daily reminders and Slack notifications running unattended.
- Studio documentation syncs from a git repository into the app, and edits made in the app go back as commits.
Engineering notes
- One shared Postgres project serves several studio apps, with table prefixes and per-app roles. A module incubated inside Colab was later extracted into its own app without a rewrite.
- Sales email and payout email use separate senders, so a deliverability problem in one can never affect the other.
- Task statuses, labels and colours are data, not code, and the reminder job reuses the same filter logic as the board.