Building Aidoo

SHIPATON 2026 · BUILD JOURNAL

From scattered thoughts to a shipped AI productivity assistant.

Aidoo is an AI-powered productivity and planning app built by Sineix Labs. It transforms natural-language instructions, quick notes, brain dumps, and realtime voice conversations into structured, actionable plans.

Status: Shipped

Product: Aidoo

Hackathon: RevenueCat Shipaton 2026

Published as a retrospective Shipaton 2026 build journal documenting the journey from concept to shipped product.

The problem we started with

We kept running into the same problem: our thoughts were everywhere, but our tasks were supposed to be somewhere.

Productivity apps work well once a plan already exists. The problem is that real life rarely arrives as a perfectly organized plan.

Most days start with scattered thoughts, unfinished ideas, reminders, obligations, and conversations. Before a productivity app can help, users are often expected to manually translate all of that into tasks, categories, priorities, dates, and reminders.

We wondered whether that translation step should exist at all.

People do not naturally think in productivity-app fields. They think:

I need to go to the gym tomorrow morning at 7, finish my report by Friday, and remind me tonight to prepare my clothes.

That single thought already contains multiple actions, times, priorities, and context.

Why should the user have to translate that thought into a productivity system before the system can help?

That question became Aidoo.

01 · Start with the thought

Aidoo begins with unstructured human input instead of forcing people to manually construct every task.

Users can:

Type natural-language instructions
Capture quick notes
Write brain dumps
Speak naturally through realtime voice

Aidoo interprets that input and turns it into structured task information.

The goal is not to make users learn another system.

The goal is to let them begin with the way they already think.

02 · AI that proposes, not silently changes

One of our most important product decisions was a draft-first workflow.

Aidoo can understand, organize, and propose actions.

It does not silently modify a user's real task system.

The workflow is simple:

Thought → AI Understanding → Structured Draft → Human Review → Action

Before meaningful AI actions are saved, the user can:

Review
Edit
Accept
Reject

Productivity data is personal.

Giving an AI too much silent control can make the experience feel unpredictable, even when the suggested action is technically correct.

Keeping a person in the loop made the system easier to understand, easier to trust, and safer to use.

03 · Teaching messy language to become structured plans

Human instructions are rarely perfect.

A user may omit dates, mention multiple tasks in one sentence, use conversational language, or describe something incompletely.

Aidoo uses Gemini 2.5 Flash with structured JSON output and schema validation.

The model does not directly operate the database.

Instead, every request follows a controlled workflow:

User Input
    ↓
Gemini Understanding
    ↓
Structured Output
    ↓
Validation
    ↓
Editable Draft
    ↓
User Confirmation
    ↓
Firestore

This separation helped us make AI actions more predictable.

It also gave us a clear place to handle incomplete or invalid model output before it could affect real user data.

04 · Giving Aidoo a voice

Aidoo supports realtime voice conversations using Gemini Live and low-latency WebSocket streaming.

The voice assistant can:

Retrieve tasks
Work with task drafts
Create or adjust task drafts
Complete or delete tasks after verbal confirmation

This became a very different engineering challenge from adding a text prompt.

We had to think about:

Realtime audio streaming
PCM playback
Interruptions
Conversation state
Tool calling
Latency
Error handling
Reliable app state during active conversations

Voice also changed the interaction model itself.

Conversations move quickly, and users do not always phrase requests like commands.

The assistant therefore needed to remain useful without becoming overconfident about what the user meant.

05 · Turning an AI feature into an actual product

Aidoo is not simply a chatbot placed on top of a todo list.

The product had to connect the AI experience to the ordinary details that make a planning app useful:

Tasks and subtasks
Priorities and categories
Recurring tasks
Due dates
Reminders
Quick notes
Brain dumps
Calendar views
Analytics
Completion rates
Category distribution
Daily and weekly streaks
Progress tracking
Authentication
Push notifications
Subscriptions

One of the biggest challenges was connecting all of these pieces into one coherent experience.

A feature could work perfectly in isolation and still feel wrong if it did not fit naturally into the overall planning workflow.

The technology behind it

Aidoo was built using:

Flutter & Dart for the mobile experience
Riverpod for state management
Firebase Authentication for account management
Cloud Firestore for tasks, notes, and user data
Firebase Cloud Functions for backend logic
Gemini 2.5 Flash for structured AI planning
Gemini Live for realtime voice interactions
RevenueCat for subscriptions and entitlements
OneSignal and Apple APNs for notifications

06 · Building the business layer

We integrated RevenueCat so subscriptions, entitlements, paywalls, and customer subscription management would not require us to build an entire billing infrastructure ourselves.

Aidoo uses a tiered subscription model:

Free

$0/month
20,000 AI tokens per month
Interactive calendar and tasks
Quick idea-note capture
Firebase cloud backup

Starter

$4.99/month
1,000,000 AI tokens per month
Interactive voice and chat assistant
OneSignal push notifications
Subscription management through the App Store or Play Store

Pro

$9.99/month
2,000,000 AI tokens per month
Unlimited realtime voice sessions
Prioritized Gemini processing
VIP support
RevenueCat Customer Center

One lesson became clear during development:

Monetization is part of the product experience.

It needs to be understandable, aligned with user value, and sustainable, while still leaving a meaningful free experience available to everyone.

07 · Bringing users back at the right moment

Aidoo's Smart Push Notifications are powered by OneSignal and Apple APNs.

The goal is not notification volume.

The goal is relevance.

When users intentionally create tasks and reminders, Aidoo can bring that information back to their attention at the moment it matters.

This turns tasks from passive information stored in the app into timely actions that help users return to their plans and follow through on what they intended to do.

08 · The privacy decision

Planning data is personal.

Because of that, privacy had to be part of the product design rather than a final checklist.

Aidoo includes:

Sign in with Apple
Google OAuth2
Email and password authentication
Password recovery
Account deletion
Deletion of profiles, notes, tasks, and notifications
Considerations for GDPR and Apple Guideline 5.1.1(v)

Personal todo data and voice streams are not used to train public models.

09 · What was harder than expected

The difficult part was not making one AI interaction work in a demo.

The difficult part was making the complete product reliable enough to ship.

We had to solve for:

Turning unpredictable human language into reliable structured actions
Keeping AI output constrained and validated
Handling realtime voice streaming and tool calls
Managing state between AI conversations and the user's actual task system
Connecting multiple services without making the product feel fragmented
Building monetization while keeping a meaningful free experience
Preparing for an actual mobile release instead of stopping at a prototype

The final constraint turned out to be the most valuable one:

We had to ship.

From idea to shipped app

Aidoo's journey looked like this:

Idea → Product Design → AI Architecture → Voice → Data → Notifications → Monetization → Privacy → Testing → App Store → Shipped Product

Aidoo was built to be more than a hackathon AI demo.

We wanted to take the full journey from an idea to a product that could exist beyond a demo environment.

What building Aidoo taught us

Building Aidoo taught us that an AI product is much more than a model and a prompt.

Structured outputs, validation layers, confirmation flows, and state management become critical when AI can interact with real user data.

A powerful model can generate impressive results, but reliability comes from everything built around it.

We also learned that voice creates an entirely different interaction model.

Latency, interruptions, conversation state, and audio feedback become part of the user experience in ways that do not exist in traditional interfaces.

Most importantly, we learned that trust is a product feature.

Many AI systems try to maximize automation.

Building Aidoo pushed us toward a different philosophy: give the AI enough capability to do the difficult work, but keep the user in control of meaningful decisions.

AI does not always need more control. Sometimes it needs better boundaries.

Where Aidoo goes from here

Aidoo already includes natural-language planning, structured AI drafts, realtime voice, tasks and subtasks, notes and brain dumps, a calendar, reminders, analytics, notifications, and subscriptions.

The next direction is to explore:

More proactive daily and weekly planning recommendations
Intelligent workload balancing
Context-aware prioritization
Deeper calendar ecosystem integrations
Shared and collaborative planning
More personalized productivity insights
Better continuity between voice and text interactions

Aidoo should not just remember what you need to do.

It should help you figure out how to actually get it done.

We're just getting started

Shipaton gave us a reason to move quickly, make difficult decisions, and actually ship.

Aidoo is still at the beginning of its journey.

But the product now exists beyond the idea.

It can listen.

It can understand.

It can organize.

It can suggest.

But unlike many AI systems, it does not assume it should make every decision.

The goal was never to replace human judgment.

The goal was to remove the friction between a thought and a plan.

That is the future we are continuing to build toward with Aidoo.

Aidoo The AI Assistant That Plans for You.

Built by Sineix Labs

SHIPATON 2026