DEV Community: Nishant Sabbarwal

LocalPilot AI — Your Business. Smarter With Open AI.

Nishant Sabbarwal — Wed, 14 Oct 2026 06:52:40 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

LocalPilot AI — Your Business. Smarter With Open AI.

Small business owners don't always need complicated enterprise software. Sometimes, they just need answers to simple questions:

Which products are selling the most?
Am I pricing my products correctly?
How can I promote my business without hiring a marketing agency?
Can AI help me understand my sales without expensive monthly subscriptions?

That's the problem I wanted to solve with LocalPilot AI, an open-source AI-powered business copilot designed for restaurants, cafés, food stalls, and independent retailers.

LocalPilot AI combines everyday business management with local AI inference, helping small business owners make better decisions without relying on expensive proprietary AI services.

Key Features

1. Smart Business Dashboard

A responsive dashboard showing revenue, sales performance, order counts, and business insights.

2. AI Menu Studio

Manage dishes, edit prices, and generate product descriptions using AI.

3. AI Marketing Generator

Generate promotional content for Instagram and WhatsApp, including English and Hindi marketing copy.

4. Sales Analytics

Track sales, identify popular dishes, and understand business performance through visual analytics.

5. AI Business Copilot

Ask questions about business performance in natural language and receive insights using an open-weight language model.

Who Would Pay for It?

LocalPilot AI is designed for businesses that cannot justify paying for multiple expensive SaaS subscriptions.

Potential customers include:

Local restaurants and cafés
Cloud kitchens and food stalls
Independent retail shops
Small business owners managing their own marketing

The long-term business model could include affordable monthly subscriptions, with a self-hosted option for businesses that prefer to keep their data locally.

Demo

Project: LocalPilot AI v0.1

The MVP includes a responsive dashboard, menu management, sales recording, analytics, and Ollama-powered AI integrations.

Demo video:

The application can also be run locally using the instructions in the GitHub repository.

For the demonstration, I used a sample restaurant called Super Zaika, with menu items such as Punjabi Chicken, Chicken Tikka, Chicken Biryani, and Chicken Soup.

The demonstration data is illustrative rather than actual business revenue.

Code

GitHub Repository: https://github.com/Nishant3007/Localpilot-AI

The repository contains the application source code, local setup instructions, and an implementation prompt for further development with coding agents.

The goal is to make the project accessible to developers who want to build or customize AI tools for their own businesses.

How I Built It

I wanted open-source AI to be a functional part of the application rather than simply adding an AI chatbot to a traditional dashboard.

Technology Stack

Component	Technology
Frontend	HTML, CSS, JavaScript
Backend	Python
Database	SQLite
AI Runtime	Ollama
AI Model	Qwen3 4B Instruct
Development	AI-assisted development
Local Environment	macOS

Open-Weight AI Integration

LocalPilot AI uses Ollama to run a Qwen language model locally.

Instead of sending every request to a commercial AI API, the application can communicate with an AI model running on the user's own machine.

The AI layer supports two important workflows:

AI Marketing Assistant: Takes business information and menu items as context to generate promotional content.

AI Business Copilot: Uses available business information to help owners understand their operations and generate recommendations.

Business calculations such as sales totals are handled by application logic rather than relying on a language model to perform arithmetic.

Application Architecture

The application follows a simple local-first structure:

Web Interface → Python Backend → SQLite Database

For AI-powered features:

Python Backend → Ollama → Qwen Open-Weight Model → Generated Response

This separation allows the application to manage business data independently of the AI runtime.

Why Does Open Innovation Matter?

This was one of the most important decisions behind LocalPilot AI.

Small businesses often operate on tight margins. A tool intended to help them save money shouldn't introduce unpredictable AI API expenses.

Using open-weight AI offers several advantages.

1. Local AI Inference

Business information can be processed on the owner's own computer when the entire application is configured to run locally.

This reduces the need to transmit business information to external AI providers.

2. No Per-Request AI API Fees

Running an open-weight model locally removes proprietary per-token API charges.

There are still hardware, electricity, maintenance, and hosting costs, but the inference model is not tied to commercial API billing.

3. Model Flexibility

The AI architecture can be adapted to support different Ollama-compatible models.

Developers can experiment with model sizes and capabilities based on available hardware.

4. Customization for Local Businesses

A restaurant may need Hindi marketing descriptions, while a retailer may need product listings or inventory-related assistance.

Open models provide flexibility to build specialized workflows without depending entirely on a closed provider.

5. Learning Through Open Development

By making the project available as source code, other developers can inspect the architecture, suggest improvements, and build their own business-focused AI applications.

For me, that's the real value of open innovation: making useful AI accessible to people who may never have considered paying for an AI product.

My Agent Session

LocalPilot AI was developed through an AI-assisted workflow covering product planning, architecture, implementation, and testing.

DevRelay session: [Add your actual DevRelay session link or embed]

The development process focused on turning a real business problem into a small, functional MVP rather than building an oversized application.

My First User

I chose Super Zaika, a local restaurant, as the initial pilot business because it provides a practical setting for evaluating menu management, sales analytics, and AI-generated marketing content.

Pilot status: Pending hands-on user testing.

The next step is to have the person managing the restaurant use LocalPilot AI and collect feedback on usability, accuracy, and whether the tool would be worth paying for.

I will update this section with genuine feedback after the test.

What's Next?

LocalPilot AI is currently an MVP, but I see several opportunities to expand it:

Automated daily business summaries
More regional Indian language support
Inventory tracking and low-stock alerts
Receipt and invoice scanning
WhatsApp-based business assistance
Multi-business accounts
Optional cloud hosting and subscriptions

The long-term vision is simple:

Make practical, affordable AI tools available to every small business owner — not just companies with large software budgets.

Thanks for reading, and happy Hacktoberfest! 🚀

🌿 WildPath AI — Less Scrolling. More Exploring.

Nishant Sabbarwal — Fri, 09 Oct 2026 07:39:17 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What if the best thing AI could do for us wasn't keeping us engaged with another screen, but helping us put that screen away?

That question inspired WildPath AI, an open-source AI-powered outdoor adventure planner built for the Hacktoberfest 2026 Touch Grass challenge.

Instead of generating more content to consume, WildPath AI helps transform a simple idea into a real-world outdoor experience.

The mission is simple:

Spend less time planning your next adventure and more time actually experiencing it.

What I Built

WildPath AI is a responsive web application that helps users discover outdoor activities and create personalized adventure plans using AI.

Imagine finishing work and wanting to spend an hour outside.

Instead of searching through dozens of websites, you can tell WildPath AI:

"I have 60 minutes, want something peaceful, and would prefer an easy nature walk."

WildPath AI is designed to turn that request into an outdoor activity recommendation.

🌲 Key Features

1. AI Adventure Planner

A natural-language planning experience that considers outdoor preferences and helps users decide what to do next.

2. Outdoor Discovery

Explore curated outdoor destinations with useful details to help choose an activity.

3. Personalized Adventure Preferences

Choose activities based on interests, available time, and preferred difficulty.

4. Adventure Journal

Keep a record of outdoor experiences and revisit past adventures.

5. Responsive Web Experience

WildPath AI is designed for desktop and mobile browsers, making it easy to explore without installing a native application.

How does it help people touch grass?

WildPath AI is designed around a different relationship with technology.

Instead of encouraging endless engagement, the application has a simple goal:

Open the app → Find an adventure → Go outside.

The screen is a starting point, not the destination.

Demo

🌐 Live Application

[WildPath AI — Web Demo]
https://wildpath-bcmigta89-nishant-sabbarwal.vercel.app/

🎬 Product Video

The video introduces the idea behind WildPath AI and demonstrates the intended experience of moving from digital planning to outdoor exploration.

For the technical demo, I also want to show the actual application interface and the AI-generated planning experience.

Code

WildPath AI is open source.

💻 GitHub Repository:

https://github.com/Nishant3007/wildpath_ai

The repository contains the web application source code, including its frontend, adventure-planning interface, and AI integration.

Developers can explore the implementation, contribute improvements, or experiment with different models.

How I Built It

My goal was to build a project where open-source AI is part of the actual product architecture rather than simply a label attached to it.

Technology Stack

Layer	Technology
Frontend	Next.js
Language	TypeScript
Styling	Tailwind CSS
AI	Browser-based open-weight model integration
AI Runtime	Transformers.js
Local Data	Browser storage
Deployment	Vercel
Version Control	GitHub

1. Building the Web Experience

I chose Next.js to create a responsive application that could be deployed publicly and tested from a browser.

The interface focuses on simplicity, with nature-inspired visuals and minimal friction between deciding to go outside and taking action.

2. Integrating Open-Source AI

The AI planning architecture uses browser-based inference rather than requiring every user request to be sent to a proprietary language-model API.

This makes it possible to experiment with openly available models and gives developers more control over how recommendations are generated.

The initial version focuses on demonstrating the planning experience. Further improvements include stronger model-based personalization and more reliable geographic route data.

3. Building for the Real World

One important engineering principle was to separate AI-generated suggestions from factual geographic information.

A language model can help interpret preferences, but real-world routes, distances, and navigation need verified map data.

This separation is important for making the product genuinely useful beyond a prototype.

4. Deployment

I used GitHub for source control and Vercel for web deployment.

The result is a publicly accessible project that can be tested without installing a mobile application.

What's Next?

The roadmap includes:

Offline trail maps and downloaded routes
GPS-based adventure tracking
On-device nature identification
Improved local AI inference
Installable Progressive Web App support
More detailed adventure journaling

These are planned improvements rather than claims about the current MVP.

Why Does Open Innovation Matter?

For me, the most exciting part of building WildPath AI is the possibility of creating useful AI experiences without making every interaction dependent on a closed, paid API.

🔓 1. Freedom to Experiment

Open models allow developers to experiment with different inference approaches, model sizes, and planning behavior.

I can improve the system without rebuilding the entire application around one proprietary AI provider.

🔐 2. Privacy by Design

Outdoor applications can involve sensitive information, including location preferences and personal activity history.

Browser-based inference creates an opportunity to process user requests locally rather than automatically sending them to a remote AI service.

That doesn't mean every part of the application is automatically private or offline, but it provides a stronger foundation for building privacy-conscious features.

💰 3. Lower Dependence on Paid Inference

With compatible open-weight models, inference can run on a user's device.

This creates an opportunity to reduce recurring server-side inference costs, although model downloads and device resources still have practical costs.

🌍 4. Technology That Encourages Real-World Experiences

AI products often optimize for longer sessions and increased engagement.

I wanted to explore the opposite idea.

What if successful AI interaction meant the user closed the application and went outside?

WildPath AI uses AI to help people spend less time using AI.

That's the philosophy behind this project.

Where Open AI Makes a Difference

The most valuable advantage is architectural flexibility.

Instead of requiring a remote proprietary model for every planning request, WildPath AI can evolve toward local inference, model swapping, and user-controlled data.

The trade-offs are also real: browser compatibility, model download size, memory usage, and inference performance.

These challenges are part of what makes building with open AI interesting.

My Agent Session

I used AI-assisted development to help plan the product architecture, develop the web MVP, and iterate on the user experience.

The development process included:

Translating the Touch Grass theme into a product concept
Designing the adventure-planning experience
Selecting a web-first architecture
Integrating an open-source AI inference approach
Preparing the project for GitHub and deployment

DevRelay Session: [Add session link here if recorded]

Prize Categories

Hacktoberfest Open-Source AI Challenge — Week 1: Touch Grass.

Final Thoughts

WildPath AI started with one simple observation:

We have powerful tools to explore almost anything digitally, but sometimes we need a little help remembering to explore what's around us physically.

I wanted to build something that makes that first step easier.

The goal isn't to replace outdoor experiences with AI.

It's to use AI to help create more of them.

Less scrolling. More exploring. 🌿

Built with ❤️ for Hacktoberfest 2026.

#opensource #ai #nextjs #hacktoberfest