Perkora

Inspiration

Early-stage founders often miss startup perks not because they are hidden, but because eligibility information is scattered across many platforms and programs. Determining which perks apply to a specific startup requires time, context, and careful comparison, which small teams rarely have.

Perkora was inspired by the idea of turning this manual and fragmented process into a fast, guided experience that feels effortless for founders.

What it does

Perkora is an AI startup agent that interviews founders one question at a time and determines which startup perks they qualify for.

The agent understands a startup’s stage and technology stack, maintains context throughout the conversation, and returns a clear list of eligible perks along with estimated savings and direct links to claim them. The focus is on producing actionable results in seconds rather than vague recommendations.

How I built it

Perkora is built and deployed on LiquidMetal AI’s Raindrop Platform using Raindrop Smart Components.

• SmartMemory maintains founder context across multiple questions
  
• SmartSQL stores and queries structured perk and eligibility data
  
• SmartInference reasons about eligibility and generates summaries
  

Backend services are deployed through the Raindrop MCP Server.
The frontend provides a guided chat experience and a results dashboard that clearly presents perk value and next steps.

Challenges I ran into

One challenge was balancing speed with accuracy. Startup perk eligibility rules can be nuanced, but asking too many questions hurts usability. This was solved by designing a conversational flow where the agent asks only what is necessary and keeps internal context.

Another challenge was ensuring the results felt trustworthy and clear. This required careful structuring of perk data so founders can quickly understand why a perk applies to them.

Accomplishments that I am proud of

• Built a fully deployed and launch-ready AI application
  
• Designed an agentic flow that feels simple but delivers real value
  
• Successfully integrated Raindrop Smart Components where they make sense
  
• Delivered clear and actionable output instead of generic AI responses
  

What I learned

I learned that strong agentic experiences are defined by clarity, not complexity. Asking the right questions, maintaining context, and delivering concise outcomes creates more impact than adding unnecessary features.

Using Raindrop allowed me to focus on product behaviour instead of infrastructure, enabling faster iteration and cleaner execution.

What's next for Perkora

Next steps include expanding the perk database, improving savings estimation accuracy, and adding deeper personalisation based on company growth and spending patterns. The long-term goal is to evolve Perkora into a broader startup operations agent that helps small teams make smarter financial and tooling decisions.

Built With

• git
• nextjs
• raindrop
• smartinference
• smartmemory
• smartsql
• typescript