Cook Along

An AI cooking assistant that listens, talks and sees .Helping you cook in real time, hands-free. Import any recipe, plan your meals, shop smarter, and guide you from first step to finished dish.

Comment

Awards Entered

Cook Along is submitted for:

  • Grand Prize Award
  • HAMM Award
  • RevenueCat Peace Prize
  • RevenueCat Design Award
  • Influencer Award — Nutrition & Healthy Eating (Abbey's Kitchen)

Inspiration

Recipes are everywhere. Actually cooking them is still surprisingly inconvenient.

Once cooking starts, your hands are busy, ingredients are moving, timers are running, and you keep reaching for your phone just to find the next step. And when something looks wrong, a static recipe cannot help.

We started with a broader question: what happens when AI moves beyond the chat box and the computer screen, and begins helping people in the middle of real, physical activities?

Cooking felt like the perfect place to start. It is one of the most universal and frequent activities in everyday life — deeply connected to health, family, culture, and the way knowledge is passed from one generation to another.

We didn't want AI to replace that knowledge by randomly generating recipes. We wanted it to help people understand, preserve, and actually use the recipes created and shared by people.

That became Cook Along.

What it does

Cook Along is an AI cooking assistant that helps from inspiration to dinner — and stays with you while you actually cook.

Share a recipe from TikTok, Instagram, YouTube, or the web, and Cook Along turns it into a structured, cookable recipe with ingredients, steps, timing, and visuals — even when a cooking video has no narration or captions.

But extraction is only the beginning. Coco prepares each imported recipe for real cooking with useful cooking tips and things to watch for, helping users understand the details that often make the difference between simply following instructions and getting the dish right.

When cooking begins, Coco becomes a real-time, hands-free assistant.

Users can ask what comes next, repeat a step, ask questions about the recipe, or manage timers by voice without touching their phone.

And when words aren't enough, just ask Coco to look. Coco can activate vision hands-free, see what's happening with the food, and respond in the context of the recipe and the current cooking step.

No tapping through menus. No stopping to open the camera. The interaction is designed for the reality of cooking: busy hands, changing food, and questions that happen in the moment.

Beyond cooking, Cook Along connects the whole journey:

  • Search across 150,000+ indexed recipes
  • Explore 2,000+ curated recipes from 19 cuisines
  • Import and organize recipes from across the web and social media
  • Get AI-generated cooking tips and things to watch for
  • See nutrition information before choosing or planning a meal
  • Plan meals and adjust servings
  • Turn plans into organized grocery lists
  • Cook with real-time, hands-free voice and visual AI guidance

From discovering something delicious to putting it on the table, the experience stays connected.

How we built it

Cook Along is a cross-platform mobile product built around a structured recipe system that acts as the shared source of truth across discovery, importing, planning, shopping, nutrition, and cooking.

Different sources — webpages, social posts, narrated videos, and even silent cooking videos — are transformed into the same canonical recipe structure so the rest of the product can reason from consistent data.

Coco builds on that context.

The real-time cooking experience combines the recipe, current cooking step, conversation context, timers, voice interaction, and on-demand vision. This means Coco isn't answering cooking questions in isolation — it can understand what the user is cooking, where they are in the process, and what is happening now.

RevenueCat powers Cook Along Plus and subscription entitlements across platforms, including premium unlimited recipe imports and unlimited cooking with Coco.

Challenges we ran into

The hardest problem wasn't generating AI answers.

It was making AI genuinely useful while someone is actually cooking — taking AI beyond the chat box and the computer screen, and making it work through real-time interaction in a dynamic physical environment.

A kitchen is noisy. Hands are occupied. Multiple people may be talking. Food changes visually over time. The user may suddenly ask a question that wasn't anticipated by the recipe. Responses need to be fast, contextual, short, and useful enough to act on immediately.

Voice interaction that feels fine at a desk can fail around an extractor fan, sizzling food, music, or conversation. Building for the kitchen meant thinking differently about when AI should listen, when it should respond, how much it should say, and how to make the interaction feel natural without requiring the user to stop what they're doing.

Vision creates another challenge: simply describing an image isn't useful. Coco needs to understand what the user is cooking, which step they are on, what the food should look like at that point, and what they should do next.

That requires more than visual recognition. It requires enough cooking knowledge to interpret what it sees in context and provide practical guidance — more like a knowledgeable sous-chef beside you than a model describing a photograph.

We also had to turn wildly different recipe sources into one reliable structure. A webpage with perfect schema, a narrated YouTube tutorial, and a silent short-form cooking video may all contain the same kind of knowledge in completely different forms.

Making all of that work together — recipe knowledge, cooking state, voice, vision, timing, and real-world context — was the real engineering challenge.

Accomplishments that we're proud of

We're proud that Cook Along is exploring a different role for AI in everyday life: not replacing human knowledge, but helping people put it into practice.

Recipes carry generations of human knowledge, culture, creativity, and experience. Instead of asking AI to randomly generate what people should cook, Cook Along starts with recipes created and shared by people — then uses AI to understand and structure them, surface useful cooking tips and things to watch for, and stay alongside the cook with real-time voice and visual help.

In just over a month, we turned that idea into a production mobile app and tested Coco in real kitchens — with busy hands, kitchen noise, changing food, unexpected questions, and all the messiness that comes with actually cooking.

We built a system that can understand recipe webpages and cooking videos, including videos without narration or captions; connect that information to planning, grocery lists and nutrition; and carry the same context into a live cooking session.

But what we're most proud of isn't the number of features or how quickly we built them.

It's that Cook Along represents the beginning of what we want to explore:

AI that participates in human activities and provides useful service in the physical world — while keeping human knowledge, culture, and experience at the center.

What we learned

The biggest insight was that AI doesn't need to replace recipes. It needs to help at the moment recipes stop being enough.

Finding inspiration is easy. Saving recipes is easy.

The difficult part begins when someone actually starts cooking.

That's when instructions meet reality: your pan is hotter, your ingredients look different, you don't have exactly the right equipment, your hands are occupied, or you simply aren't sure whether what you're seeing is right.

That changed how we think about Cook Along.

It isn't simply an AI recipe app, and hands-free interaction isn't just a more convenient way to turn pages.

We're exploring what happens when multimodal AI has enough context to participate in an activity as it unfolds — listening when needed, looking when asked, understanding what stage you're at, and helping you decide what to do next.

The most useful AI may not always be the AI that generates more content.

Sometimes it's the AI that helps people do something in the real world.

What's next for Cook Along

Our immediate goal is to learn from real cooks: where Coco genuinely removes friction, where it gets in the way, and which moments benefit most from voice or vision.

Next, we want to make Cook Along multilingual and deepen its understanding of food, nutrition, and individual needs.

Nutrition information is already part of the experience. Over time, we want to help people make more informed choices about what they cook, build meal plans around their preferences and goals, and make healthier home cooking easier to understand, plan, and actually follow through on.

We also want Cook Along to become a place where human food knowledge continues to grow.

We plan to bring creators into the platform, let people follow cooks they trust, discover their recipes and collections, and eventually build communities around cuisines, interests, dietary preferences, and different ways of eating.

And Coco will keep evolving beyond step-by-step cooking.

We want to explore where AI can provide genuinely useful service throughout everyday food life — understanding what you have, helping decide what to eat, adapting recipes to your kitchen, helping you understand nutrition, planning and shopping, cooking alongside you, and learning what works for you over time.

Our long-term idea is bigger than recipes:

AI shouldn't only help us inside a chat box. It should participate thoughtfully in the things we do in the real world — while keeping human knowledge, culture, and experience at the center.

Cooking is where we're starting.

Built With

Share this project:

Updates

posted an update —

Update (Sept 30): Our iOS app was approved by Apple on September 30, the final day of the Shipaton submission period. Android was already live, but iOS only became available at the very end, so there has been very limited time to acquire users or collect iOS usage data before judging.

Log in or sign up for Devpost to join the conversation.

Submission history