MAYA

MAYA is an AI business growth agent that finds opportunities, prioritizes customers, creates marketing campaigns, automates follow-ups, and recommends the next best action for small businesses.

Comment

Inspiration

Every small business owner or homepreneur we spoke to had a product, but not a growth system. They didn't know where their next customer was coming from, what content to post, who to follow up with, or which offer would actually convert. This isn't a tools problem — they already have Instagram, WhatsApp, and spreadsheets. It's a decision-making problem: too many daily micro-decisions, too little time, no dedicated marketing or sales person.

We wanted to build something that acts less like a "content generator" and more like an actual AI employee — one that can look at a business, understand its customers, and decide the next best action on its own.

That's how MAYA — your AI employee for getting customers, creating content, and growing your business — was born.

What it does

A business owner just describes their business in plain language, for example:

"I sell handmade jewellery. My target customers are women aged 18–35."

From there, MAYA's agents take over:

  • Customer Discovery Agent — builds a target customer profile
  • Market Agent — analyzes market and competitor signals
  • Content Agent — generates Instagram/Reels/WhatsApp content
  • Sales Agent — prioritizes leads by purchase intent
  • Follow-up Agent — drafts and triggers personalized follow-up messages
  • Pricing Agent — suggests pricing and offers
  • Growth Agent — reviews performance and decides the next best action

The core "wow" moment: a business owner asks "I have 20 leads — which customers should I follow up with today?" MAYA analyzes engagement and intent, ranks the leads by priority with a recommended action for each, and then runs a full Analyze → Decide → Generate → Execute → Measure growth cycle when the owner clicks "Run today's growth plan."

How we built it

MAYA is structured as a central AI Business Manager coordinating specialized sub-agents (Customer, Marketing, Sales, Growth) that each handle one part of the workflow — lead scoring, content generation, follow-up drafting — and feed their output into a single "next best action" recommendation for the business owner.

The Gemini API powers the reasoning and generation layer across these agents — customer profiling, content creation, and lead prioritization all run through live Gemini calls in the deployed app.

Challenges we ran into

  • Designing agent boundaries so each one (Customer, Marketing, Sales, Growth) had a clear, non-overlapping responsibility while still feeding into one coherent recommendation.
  • Making the "next best action" output feel genuinely useful and specific rather than generic AI advice.
  • Balancing scope for a hackathon timeline — building real, working workflows instead of just describing them.
  • Handling Gemini API rate limits and model availability on the free tier by adding automatic retry logic.
  • Getting authentic user feedback: rather than projecting a large addressable market, we focused on getting a small number of real small-business/homepreneur testers to try MAYA and give honest feedback.

What we learned

Judges for this track weigh business viability, AI-native operations, and category impact equally — so a working, demonstrable workflow with real (even small-scale) user feedback matters more than a big idea alone. Building MAYA reinforced that the most valuable AI agent isn't the one that generates the most content, but the one that makes the best decision about what to do next with limited time and leads.

What's next for MAYA

  • Expand the Follow-up Agent to support direct WhatsApp/Instagram DM automation
  • Add more granular analytics for the Growth Agent
  • Introduce multi-product and multi-channel support for the Agency tier
  • Onboard more small businesses as active testers to validate the subscription model

Built With

Share this project:

Updates