Recurmind

Recurmind transforms e-commerce operations into autonomous agentic workflows. The core problem we solve is the massive inefficiency in abandoned-cart recovery and manual CRM actions — tasks that currently require human operators or rigid automations. Our target customers are small and mid-size online merchants who lose revenue daily because they cannot follow up with every customer in real time.

What the system does

Our MVP delivers a fully autonomous Cart Recovery Agent that:

Detects an abandoned cart

Parses customer context (items, price, intent)

Decides the correct CRM action

Generates a personalized recovery message

Sends it automatically through a simulated WhatsApp → n8n flow

Waits for customer confirmation

Completes the order using Crossmint’s agentic, tokenized payment flow

Stores each action in Qdrant as vector memory (traceability + learning)

The demo flow:
A customer abandons a cart → the agent reasons → generates the right message → receives consent → triggers tokenized card payment → confirms order.

This shows autonomous reasoning, action, decision-making, and multi-step completion without human intervention.

How the project meets the judging criteria

1. Execution & Functionality

Full autonomous pipeline built end-to-end

Real agentic reasoning (intent → plan → action)

True tokenized payment execution with Crossmint

Functional UI: chat + suggested actions + agent status + logs

2. AI & Agents Usage

LLM-driven reasoning and decision-making

Context extraction from cart scenarios

Agent state machine with autonomous action selection

Vector memory (Qdrant) for logging and retrieval

3. Scalability — One-Person Model

Recurmind is built to be maintained by a solo founder.
The agent handles:

messaging

decision-making

payment flow

confirmations

The architecture is modular: new agents (PromoAgent, SupportAgent…) can be added without UI redesign.

4. Problem Clarity & Market Impact

Cart abandonment costs merchants trillions globally.
Today’s tools are static automations.
We replace them with autonomous agents able to:

decide

interact

execute payments

close revenue

This directly increases merchant revenue with zero extra work.

5. Demo & Product Narrative

The narrative is simple and strong:
“Your online store runs itself.”
The user interface shows:

the agent thinking

the message it prepares

the action it takes

the final payment success

Clear, visual, and convincing.

6. Partner Tech Usage

We use:

Crossmint Agentic Payments

Tokenized credit card demo (agentic flow)

Basis Theory tokenization

Purchase intent + verification + payment intent

WhatsApp- messaging (n8n triggers)

This demonstrates real agentic commerce in action.
Technologies Actually Used

These are the real technologies used in your prototype, nothing more:

AI & Logic

OpenAI GPT — context extraction, reasoning, message generation

Simple agent state machine — intent → plan → action

Frontend & UX

Next.js + React — base app

Tailwind CSS + shadcn/ui — layout, components, sidebar

Framer Motion — lightweight animations

Custom chat interface — user-facing agent interactions

Messaging (Simulation)

n8n workflow simulation — sends/receives WhatsApp-style messages

UNIPILE API for whatsapp

Payments

Crossmint Agentic Flow — tokenized card → purchase intent → payment

Basis Theory — card tokenization

HTTPS via ngrok — required for secure card flow

Memory & Logging

Qdrant — vector storage for actions, messages, timestamps

Hosting / Dev

Local Next.js app

ngrok for secure tunneling

Failure Modes & Next Steps

Failure modes:

Wrong intent classification

Message mismatch

Payment verification errors

Missing context from merchants

Next steps:

Real WhatsApp Business integration

Shopify MCP / Checkout Kit integration

Multi-agent orchestration (promo, support, advisor)

Real merchant dashboard

https://github.com/0xplug75/recurmind-v.1

https://www.loom.com/share/c76a8512029643d0bcf5bb14c7dbb7f8