Inspiration

As the Founder & CEO of the award-winning global platform Women in Blockchain Talks (WiBT), I’m constantly approached by people from every walk of life—teachers, creatives, policy-makers, and budding developers—each trying to make sense of ever-evolving Web3 jargon. The space moves at break-neck speed, yet that shouldn’t lock newcomers out. I wanted to remove the friction of endless Googling and Twitter-thread diving by creating De3PA – Your Web3 Personal Assistant, an agent that answers “What does this mean?” right when the question sparks, using language that respects users’ time, resources, and diverse learning styles.

What I Learned

• Simplicity Wins
  The one-prompt challenge forced clarity: if a feature couldn’t be expressed in a single instruction block, it was probably scope-creep.

• Skill-Up by Shipping
  Building inside Bolt.new leveled up my confidence. I went from “curious hacker” to “I can model and deploy an AI agent in an afternoon.”

• Community Energy
  Every test chat with De3PA felt like a rehearsal for empowering my WiBT network; that feedback loop made learning joyful.

How I Built It

1. Ideation in ChatGPT
   Sketched user journeys, refined the agent’s tone, and mapped quiz logic—offline first to save API credits.

2. Ported to Bolt.new
   Converted the draft into a single JSON prompt, wired in a tiny tts() function for ElevenLabs, and kept memory footprints slim.

3. Rapid Iteration

   Bolt’s live preview let me A/B test wording, tighten the 250-word cap, and ensure the output schema stayed judge-friendly.

Challenges & Wins

• Letting Go of Perfection
  My biggest hurdle was resisting the urge to over-engineer. Accepting “good enough for M

Built With

• bolt
• daos
• defi
• json
• nfts