Explore

AI Roadmap

DiMuto AI

DiMuto AI

Product Quality AI

Description

Current: To detect visual defects (e.g. blemishes, cuts) in fruits.

End goal:

Buyer’s biggest concern is: are you sending me the right things I have requested for (e.g. size, grade).

DIMUTO PILLAR

Trade Management System

READINGS

Learnings from Clarifruit (taken from their website / blog)

Technology

AI tech runs on mobile

IDEA

Trade Health AI

Description

End goal: To give an assessment on the probability of default for DiMuto financier to make a better judgement on whether DiMuto should finance the trade

IP: changing the game by enabling trade health to be detected + sustainability score for green financing

DiMuto term for green financing for value proposition

DIMUTO PILLAR

Financial Services

Sustainability Report AI

Description

End goal: To generate credible and data-proven sustainability reports for customers to use for regulatory purposes, fundraising, brand awareness.

Phase 1: Sustainability reports in line with sustainability reporting guidelines (Q4 2023)

DIMUTO PILLAR

Trade Management System

Marketplace Chat AI

Description

End goal: To chat with customer about their buying and selling needs (replacing humans who do it manually). Make the marketplace alive!

AI to understand what the customer wants and gives suggestions e.g. instead of filling up the quotation form, customer will talk to the chatbot.

Buyer can ask "I want to set up a store in san paulo, what should i sell"; "i want to do something promotional. what do you recommend"; "i need asparagus between this time and within this price"

DIMUTO PILLAR

Marketplace

READINGS

Technical approach

A. Train LLM on own data - GPT index of your documents, Open AI key

B. Build QA chatbot with LLM - approximate nearest neighbour (ANN) models to find similar items + instruction-tuned LLM

IDEA

Product Quality AI Calculator

Description

End goal: To be able to predict the product quality of the products upon reaching the buyer. This is good for trades from new origins/ to new destinations.

Use cases:

DIMUTO PILLAR

TMS & Marketplace

IDEA

Q&A Siri AI

Description

“What is the best selling fruit?”

Content Suggestions AI

Description

Suggest better product photos

Suggest better writing in marketplace listings

Generate high quality product image listings on marketplace

ChatGPT on the platform

Description

Operations e.g. voice recognition “can you start the pack plan session now”

Reading charts e.g. “can you share with me the chart for the banana trades”

Report generation e.g. dispute report

READINGS

AFL — reading data/ charts; Q&A

Dispute AI

Description

Generate report formally and objectively, explaining the information presented and not a “data dump”

Quality Report

Use generative ai to give potential reasons for the deterioration based on different data points. e.g. "a berries trade is fragile and requires great care in temperature during shipping. Based on the iot device, the temperature during shipping increased above x degrees C, which can reduce the shelf life of the fruits."

DIMUTO PILLAR

Trade Management System

IDEA

AI methodology

1. Data Collection and Integration:

Integrate Data Points: Ensure all data points, including images, timestamps, IoT device readings, shipping information, and goods received information, are centrally stored and accessible.

Feature Engineering: Extract meaningful features from this data. For instance, if you're focusing on temperature, derive metrics like 'maximum temperature', 'duration above a certain temperature', etc.

⁠

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.