A company’s operation creates a lot of unstructured data. Like refining crude oil Suppeco refines unstructured ‘crude’ data to extract an unlimited range of value by-products. SuppEQ analyses customer, supplier and stakeholder data (commentary, chat email, voice, documents, other BI sources) tracking sentiment, interpretation emotion, nuance, opinion behaviour, and performance as well as propensity to types of risk, and culture. Whilst retrospective analytics are good at identifying problems and predictive analytics are good at anticipating them additionally, our prospective approach means we can integrate multiple contextual variables of supplier and stakeholder data to identify likely outcomes.

 

We’re all pretty familiar now with using a range of Gen-AI models as personal assistants, at school or work, or just because! We’ve seen what they can do. Now imagine leveraging a more advanced AI engine that draws its intelligence from across your entire organisation, your people, and all of your suppliers – your entire supply chain ecosystem. We’ll take a closer look. But first, some key info, for context.

 

ChatGPT’s responses are generated based on several types of training data. Here are its key sources:

 

• Licensed Data: OpenAI licenses datasets that are used to train the model. These datasets come from various sources and cover a wide range of topics and languages.
• Publicly Available Data: This includes information from books, websites, and other texts that are publicly accessible on the internet. This broad base of information helps the model generate responses on a wide array of subjects.
• Human Trainers: OpenAI uses human trainers to provide demonstration data and to fine-tune the model’s responses.
• ChatGPT has limited internet browsing capabilities, allowing it to access real-time information or search for data online. It does not have access to personal data unless it has been shared with it during a conversation.

 

The training process involves using this large and diverse array of data to teach the model to understand and generate human-like text by predicting the next word in a sentence, which, over time, allows it to respond in a coherent and contextually appropriate manner. This is what’s meant by the terms Large Language Model and Generative AI.

 

Unique and Infrastructurally Pervasive

 

Modern Business relationships require structure and boundaries, less they become dysfunctional; just in fact like our personal relationships. Remember, that business relationships are lots of individual relationships, all joined up.

 

Until about 2022 (post-pandemic) Supplier Relationship Management (SRM) was often overlooked in favour of traditional performance management for achieving measurable results. The value of SRM was perceived as subjective and anecdotal, akin to nailing jelly to a wall. However, much has changed since then. In today’s volatile market, influenced by macroeconomic and geopolitical factors, gaining a competitive edge has become crucial. This edge relies on achieving higher levels of engagement, where aspirations like becoming a supplier’s “customer of choice” are frequently discussed but challenging to attain. For further insights on becoming Customer of Choice, read the BAE Systems Digital Intelligence Case Study.

 

As a relationship-first platform, Suppeco assembles relationship structures and those all-important boundaries to support functional and measurable engagement, for today’s operations, at scale. This is what we call the 4 pillars digital relationship layer. You can find more information here on the digital relationship layer.

 

SuppEQ is unique because it analyses entirely private and specific data to each customer and their supply chain, focusing on the vast amount of unstructured operational data created every day across the ecosystem, which is curated into the customer’s digital relationship layer on Suppeco.

 

So, what is Training Data?

 

Training data refers to the dataset used to teach a machine learning model how to make predictions or decisions. In SuppEQ, this process is optimised and fine-tuning using the unstructured data uploaded (manually, and systemically via API), into the customer’s Suppeco tenant. This data can include:

 

• Commentary: In-platform discussion between customers, suppliers, and all individuals.
• Emails: Connected correspondence between customers, suppliers, and internal teams.
• Chat 1: In-platform conversations between customers, suppliers, and internal teams.
• Chat 2: Connected chat between customers, suppliers, and internal teams (Teams, Slack).
• Files: Reports, contracts, proposals, and other text-based files (including .doc, .xls, .pdf).
• Business Intelligence: Reports, metrics, approach scores, cadence, relationship utilisation.
• Imagery: Photos, diagrams, and other visual data.
• Certificates: Compliance, standards, digital certification.
• Voice Data: Recorded meetings, calls, note takers, other audio files.

 

Benefits of Using Customer-Specific Data when Training:

 

• Personalisation: Because SuppEQ contextualises with customer-specific data it can provide entirely more tailored insights and recommendations.
• Relevance: SuppEQ understands the unique context and nuances of the customer’s supply chain and all its personal relationships.
• No Noise: Because SuppEQ is utilising customer specific data to support its training, by default it eliminates outside noise, “irrelevant chit chat”.
• No unwanted bias: SuppEQ utilises customer specific data, so by default rules out “unwanted” bias. Whereas bias observed in the data may be worthy of evaluation.
• Privacy: By confining private datasets within the customer’s secure environment, Suppeco ensures data confidentiality and compliance with privacy regulations.

 

Quick Examples of SuppEQ

 

Within a Suppeco tenant, a customer and its suppliers engage in various forms of communication – ranging from commentary, chat, and connected emails to voice recordings – all of course relating to contracted services. SuppEQ can leverage this diverse dataset to unlock actionable insights, identified for example within specific product lines or service areas; assessing stakeholder contributions, and analysing KPI metrics alongside other business intelligence data.

 

Responding to agent-defined requirements, SuppEQ applies advanced sentiment analysis and psycholinguistic evaluation lenses to email and other communications. This helps gauge key dynamics within customer-supplier interactions and relationships. Additionally, it is able to extract critical terms and clauses from contracts and related documents to monitor compliance and performance metrics effectively.

 

SuppEQ’s capabilities extend to analysing transcribed voice recordings, enabling it to detect nuanced negotiation cues. By doing so, it predicts potential risks and identifies emerging opportunities, empowering agents to strengthen supplier relationships and drive value.

 

SuppEQ Amplifies Effective Output with Live Streaming Data

 

Suppeco creates and curates live, dynamic, naturally aging data, adding a valuable nuance. For this reason, Suppeco splits data to support training, into Background and Foreground. The background essentially refers to the overarching digital relationship layer infrastructure itself, whereas foreground data for training refers to the dynamic data that naturally ages and always changes.

Understanding Suppeco’s Data for Training

 

1. Background Data for Training

 

Definition:

 

• The foundational data that forms the digital relationship layer infrastructure. This includes static information that doesn’t change frequently.

 

Components:

 

• Historical Data: Long-term contracts. Shared documents and artifacts, any artifacts uploaded into 4 Pillars.
• Structural Data: Organisational hierarchies, supplier profiles. Standard and custom 4 Pillars.
• Contextual Data: Stakeholder mapping, stakeholder narratives.
• Static Documents: Policy documents, standard compliance requirements, and baseline metrics.

 

Purpose:

 

• Provides a stable context for SuppEQ to understand the overarching framework within which relationships and transactions occur. Helps the system recognise long-term patterns and trends.

 

2. Foreground Data for Training

 

Definition:

 

• The live, dynamic data that is continuously updated and naturally ages over time. This data captures real-time interactions and current states of relationships.

 

Components:

• Real-Time Communications: Connected emails, chat (in-platform) connected chat, and ongoing correspondence.
• Live Documents: Updated contracts, current reports, and real-time performance reviews.
• Voice and Visual Data: Current meeting recordings, transcriptions, recent images, and up to date multimedia data.
• Operational Data: Live transaction records, real-time compliance checks, and active project management data.

 

Purpose:

 

• Keeps SuppEQ updated with the latest developments, ensuring that recommendations and insights are based on the most current information. It allows the system to adapt to changes and provide timely, relevant advice.

 

How Suppeco Utilises Background and Foreground Data for Training

 

Intellectual and Contextual Segmentation:

 

• Background Data: Forms the stable foundation upon which dynamic interactions are understood. This data is less frequently updated but crucial for maintaining a contextual understanding of the supply chain ecosystem. This speaks to the necessity for functional structure and boundaries required to support all our relationships.
• Background Based Raw Semantic Indexing: The 4 Pillars digital relationship layer categorises the data to support the first critical semantic indexing sortation – ensuring a higher quality of data input for SuppEQ.
• Foreground Data: Continuously fed into the system and read (Streamed) in real-time to update SuppEQ’s understanding of latest states and trends. This ensures SuppEQ’s insights are timely and relevant.

 

Processing and Integration:

 

• Background Data Processing: Focuses on creating pathways and access to existing, as well as new and custom relationship structures, for extraction of emerging baseline patterns, and performance metrics.
• Foreground Data Processing: Emphasises latest or real-time analysis, sentiment detection, psycholinguistic analysis and dynamic trend identification.

 

Model Training and Adaptation:

 

• Training with Background Data: Helps SuppEQ understand the static framework and infrastructure, the basis for functional boundaries, historical trends, and baseline metrics.
• Training with Foreground Data: Ensures SuppEQ remains responsive to current changes, the natural data aging process, and capturing the evolving nature of relationships and their operational dynamics.

 

Insights and Recommendations:

 

• Stable Insights from Background Data: Provides a reliable foundation for strategic decisions, based on long-term norms, established patterns, and functional structures.
• Dynamic Insights from Foreground Data: Offers immediate, actionable recommendations and prospective suggestions based on the latest information, supporting agile decision-making.

 

Example Scenario

 

Imagine a supplier relationship that has always operated within its functional structures and relationship boundaries (background data) but has also recently demonstrated signs of strain due to delayed deliveries and negative sentiment in recent communications (foreground data).

 

SuppEQ will:

 

• Background Analysis: Refer to historical functioning status data to understand long-term role stability, and segmentation / importance of the supplier.
• Foreground Analysis: Analyse recent comms: emails, stakeholder commentary and chat, performance data, and other available BI, to identify the current issues and their impact.
• Combined Insight: Provide a nuanced recommendation and action plan that considers both the historical status of the relationship and the immediate need for intervention to address the current issues.

 

Suppliers on the Suppeco platform typically operate under bilateral customer contracts. However, the services they provide often go beyond bilateral frameworks, requiring multi-party collaborative engagement. By separating data for training into background and foreground, SuppEQ retains a detailed understanding of each bilateral digital relationship layer (the static framework) while dynamically adapting to the evolving collaborative landscape of customer-supplier interactions. Suppeco uses retrospective analytics to identify past issues, predictive analytics to anticipate future problems, and prospective analytics to integrate various contextual variables from supplier and stakeholder data. This comprehensive approach allows Suppeco to identify likely outcomes and offer actionable recommendations.