Major airline holding company.
Empowering a major global business with two GenAI solutions:
- Safe GPT: Enabling their people to utilise ChatGPT in a safe and secure environment.
- Semantic search: Using ChatGPT to perform semantic search on internal company documents to retrieve key information.
- Security
Sensitive company data is leveraged effectively for ChatGPT use whilst being kept entirely protected. - Simplicity
Access to thousands of internal documents – and all their data – at speed. - Scalability
The solution can be rolled out across multiple departments, at the touch of a button.
Project Summary.
About.
With an emphasis on sustainable value, the major global airline holding company is always looking for opportunities to scale and operate as efficiently as possible.
- Manages 7 operating companies
- Has more than 66,000 employees
- Operates across 91 countries
Challenge.
This wasn’t a problem-solving endeavour, so much as the harnessing of an opportunity for innovation that would bring about business-wide operational efficiencies, by streamlining processes and saving both time and manpower, using generative AI (GenAI).
In parallel, the client wanted to explore possible new business areas, by encouraging staff to think about what innovation might be possible, when using GenAI.
The company also wanted to explore potential use cases to take forward to its operating companies. The scope of this included LangChain (a framework for GenAI) and image generation, and, more specifically, the deployment of image generation on Azure and AWS.
To achieve all this, the client wanted an off-the-shelf solution, which needed to be heavily engineered and customised, in order to meet its necessarily stringent cybersecurity requirements.
Safe GPT.
It wanted to deploy a working instance of ChatGPT 3.5 and ChatGPT 4 in its Azure environment, as a POC. This was to prevent any leakage of internal data, which had the potential to damage the brand.
Semantic search.
It was difficult to extract information from 1000s of documents – the client wanted to be able to Q&A data in a conversational manner. They wanted to use ChatGPT to perform semantic search on internal company documents to retrieve key information.
Solution priorities.
Simple.
A straightforward, clean, intuitive user interface, to make everything as functional, transparent and operable as possible.
Scalable.
The ability to replicate, expand and customise the project at the touch of a button, repeatedly if required.
Secure.
The need to retrieve, leverage and deploy sensitive company data using AI, all while being hosted safely in-house, with watertight cybersecurity.
Approach: SafeGPT.
1. Exploration.
Workshopping to understand the client’s requirements and concerns, and exploring viable options.
2. Accelerator demo.
Demo of accelerator to show initial look and feel of SafeGPT
3. Hardened security.
In network cybersecurity terms, all the new functions were hardened to pass all penetration testing.
4. Deployment & customisation.
Deploying on the Azure environment and customising to suit their needs – including branding, authentication, and authorisation
5. DevOps pipelines.
Creating Infrastructure-as-Code (IaC) DevOps pipelines for repeatable deployments and customisations.
6. Launch & support.
Soft launch of SafeGPT to selected senior executives, plus ongoing support with new releases, updates and modernisation.
Approach: Semantic Search.
1. Exploration.
Workshopping to understand the client’s requirements and concerns, and exploring viable options.
2. Accelerator demo.
Demo of accelerator to show initial look and feel of the semantic search PoC
3. Deployment & customisation.
Deploying on the Azure environment and customising to suit their needs – including branding, file type, SharePoint connection, and security implementations
4. DevOps pipelines.
Creating Infrastructure-as-Code (IaC) DevOps pipelines for repeatable deployments and customisations.
5. Launch & support.
Launched to multiple departments to extract specific use cases, and teams across the company were provided with technical sessions and demos, to ensure they were able to optimise these new functionalities. And Nordcloud provides ongoing support with new releases, updates and modernisation.
Need more tech details?
Leave us your email and our specialists will be happy to share their expertise and more project details. Not in the mood to talk?
Simply email your questions!
Results.
The company now has the ability to take advantage of the latest AI capabilities in a safe, secure, compliant way – through both the creation of an internal OpenAI platform, with a bespoke search portal that can upload documents, and through a protected SafeGPT solution.
Rapid PoCs.
These were completed from initial conversation to completion within a strict four-week deadline.
Enhanced UX.
Deployment using OpenAI + Cognitive Search in Azure makes for a simple, clean, intuitive UI, and all chat sessions can be stored and recalled safely. It can ingest, embed, index, and search through 100s of documents.
Easily scalable.
These new-found capabilities were so well-received that Nordcloud was asked to redeploy the search portal to four other departments. Thanks to the IaC set-up, this could be deployed at the touch of a button.
Exploring new tech.
There’s now a dedicated team focused on testing new GenAI technology for this client as they continue to define use cases and develop PoCs to test value.
Robust security.
Because these new functions are all hardened to pass penetration testing, they are both secure and stable, which has already mitigated a tangible number of breaches and vulnerabilities.
Jumpstart Your Data Platform.
Not yet ready for a chat? Download our guide and see how a cloud-native data estate helps you avoid common data management pitfalls
