Projects

Challenge: Intensely manual process to cross-check project documentation against requirements, tracking exceptions to ensure project success.

Solution: Evolving to AI Agents by starting with Q&A chatbots to enable source cross-checking of documents uploaded into SharePoint. Agents are dynamically selected based on the document type. Once analysed, the AI agent notifies the person who uploaded the document with actionable guidance.

Technologies: Microsoft SharePoint, Azure AI Foundry, Azure AI Search, Llamaindex, GPT4o, o3-mini, Copilot Studio, Azure CosmosDB

ROI:

• Increased market value of this firm to assist in upcoming sale
• Reduced cost of servicing projects drives improved margins
• Reduced risk of projects impacting reputation and outcomes for stakeholders

Challenge: Build an AI app which is customer-ready and marketable in a short period of time.

Solution: We built Jummbo, an app that enables salespeople to quickly identify new prospects they can target. The AI agent guides the user to find suitable business types and locations, searches the internet, and then researches each prospect. It returns a spreadsheet complete with customer names, social media links and tailored cold email and cold call scripts ready to action.

Technologies: GPT4o and GPT4o mini, MongoDB, NodeJS, React

Outcome: Jummbo.ai is a fast-growing SaaS app designed for sales and marketing teams to accelerate their sales pipeline development. www.jummbo.ai

Challenge: This company manages tens of thousands of deliveries each year. Although they have stringent safety standards, including manually inspecting each vehicle to ensure loads are securely fastened prior to departure, they wanted to leverage AI to minimise risk further.

Solution: We helped the customer understand how multimodal AI models can analyse images. Through a short proof of concept we showed them how to perform a red/amber/green assessment of photos collected at the warehouse prior to departure, identifying potential safety issues. We then helped their IT team in a "build with" capacity to apply these learnings to their technology stack.

Technologies: GPT4o on Azure dedicated compute (PTUs)

Outcome: The organisation has made significant advances across multiple domains in leveraging AI to reduce risk, improve safety and improve general operational performance.

Challenge: Call centre operators are seeking ways to leverage AI to improve all parts of their operations, including reducing average handling time (AHT) and better understanding the needs of their customers.

Solution: Solutions vary across customers. Typically organisations start with a Q&A knowledge chatbot internally, then extend it to web chat and voice channels to deflect calls and reduce AHT. Setting up a "data product" using speech transcriptions enables these organisations to better understand customer needs to inform better design of products, services and experiences.

Technologies: AWS Connect, Salesforce Service Cloud Voice, and multiple Azure services

Outcome: Improved customer satisfaction, reduction in cost to serve, improved marketing outcomes through a deeper understanding of customer needs.

Challenge: The customer runs a popular directory related to the primary and secondary agricultural industries in Australia. They wished to enrich the directory with more information about each member listing without expending significant manual effort.

Solution: We developed a crawler for the organisation that researches each member's website, capturing an overview of the company, key products they sell, information about their region and links to their social media sites.

Technologies: MySQL, OpenAI, NodeJS, Python, Langchain

Outcome: Improved member listings deliver a better consumer experience. Higher quality content on the website has also delivered an increase in web traffic from search, and longer average time on site from visitors.