AI Assistants

Integrating ChatGPT into your business can have significant benefits - see our use-case examples below. Blueberry can harness Its capabilities to enhance customer engagement, boost satisfaction levels, and improve your overall bottom line.

Service Industry

• A website for the beauty industry, which was linked to the ChatGPT API to answer customer questions about the services offered on the portal - such as treatments available, how they are performed and available appointments. The AI chatbot was fine-tuned specifically on the industry's own data, enabling ChatGPT to deliver precise answers using a library of supplied reference text. The integration combined chat with data – so it could answer questions like, “who can blow dry my hair on Tuesday” by referencing SQL data on skills and holidays. Additionally, the ChatGPT language model was able to retain context though the conversation!

Transcribing Spoken Language

• ChatGPT was trained to transcribe spoken language into text and translate it into another language in real time during live meetings with international teams. This feature eliminated language barriers and ensured that everyone was on the same page, regardless of their native language.

Customer Queries

• ChatGPT was used to create a chatbot that could answer questions about a company's shipping and return process, and to provide support to customers who were having problems with their orders. The chatbot was available 24/7 and could handle a wide range of customer queries.

Customer Support

• A large company integrated ChatGPT into its customer support system to reduce costs and improve customer satisfaction. ChatGPT was configured to use the company's knowledge base as the main source material. Customers could then use a live chat interface to ask questions about the company's products and services. If ChatGPT was unable to answer the customer's question, it would escalate the issue to a human agent.

Product Catalogue

• ChatGPT was configured to use a client's product catalogue and other website content as the main source material. Visitors to the site could then use a built-in search box to ask questions about products, such as "What is the difference between these two different models?" or "What are the customer reviews for this product?" ChatGPT would then generate a personalised response that answered the user's question in a comprehensive and informative way.

Although the most obvious use case for ChatGPT integration is customer service and helpdesk applications, businesses across many industries have also started looking at integrating AI chatbots into their systems - and exploring different applications. As ChatGPT and its alternatives continue to develop, we can expect to see even more use cases for these tools.

There is no one-size-fits-all approach to ChatGPT integration, and the best way to use ChatGPT will vary depending on the specific business and its needs. Companies that have already integrated ChatGPT are seeing measurable results, despite there still being some concerns about its accuracy and ethics. However, these concerns should not prevent businesses from considering ChatGPT integration, as in most cases, the benefits will far outweigh the concerns.

There are five categories of use cases for ChatGPT:

Content Generation:

• ChatGPT can generate content for blogs, articles, ads, marketing materials, and product descriptions. It can also assist with research, generate topic ideas, and write in specific styles or tones.
• Automatic summarisation of long articles or documents helps users quickly understand the main points without reading the entire text.
• The bot can transcribe spoken words into text, enabling businesses to analyse customer conversations.
• Real-time translation of text is useful for messaging apps, social media platforms, and other communication channels.

Workflow Management:

• The bot serves as a virtual assistant for task management, including scheduling, reminders, and to-do lists.
• It helps users sort, prioritise, and respond to emails, increasing productivity and reducing email overload.
• ChatGPT assists with scheduling posts, responding to comments and messages, and providing content and engagement recommendations.
• Managing knowledge and information, such as FAQ pages or employee manuals, becomes easier with the bot, enabling employees and customers to find answers quickly.

Customer Experience & Interaction:

• The bot answers questions, troubleshoots technical issues, and provides information about products and services.
• ChatGPT personalises user experiences by offering recommendations based on preferences and behaviour.
• It assists customers in making purchase decisions through product comparison tools or chatbots that provide information and reviews.
• Analysing customer feedback through sentiment analysis tools or chatbots collecting feedback provides actionable insights.

Security and Compliance:

• ChatGPT detects fraud by analysing data, identifying patterns and anomalies indicating fraudulent activity.
• The bot monitors networks for suspicious activity, alerting security personnel to potential threats. It can also analyse network traffic, identifying unusual login attempts or data transfers.
• Monitoring compliance with industry regulations or internal policies helps businesses avoid legal or ethical issues.
• It helps identify and assess potential risks, analysing data from various sources to detect cyber threats.

Workflow Optimisation:

• ChatGPT analyses data using machine learning algorithms and natural language processing, providing insights for decision-making. For example, a virtual assistant powered by ChatGPT can analyse market trends and financial statements, offering investment recommendations.
• The bot analyses data from databases, spreadsheets, and social media platforms, providing insights on consumer behaviour, market trends, and relevant information. It can assist with data visualization, creating charts and graphs for better accessibility.
• Tracking inventory levels, monitoring order fulfilment, and managing the supply chain are tasks the bot can handle using natural language processing and machine learning algorithms.
• It assists with budgeting, bill payment, financial planning, and provides recommendations for investment strategies and opportunities.

Generally, there are two ways to integrate ChatGPT into your systems: using an API, or a custom implementation. How you integrate ChatGPT into your app depends on your specific needs and requirements. If you need a quick and easy way to get started, using an API is a good option. However, if you need the flexibility to customise the bot's responses and workflows, a custom implementation is the way to go.

Note that Microsoft has recently announced Power Virtual Agents, which lets customers create AI-powered chatbots for a range of uses - from providing simple answers to common questions to resolving issues requiring complex conversations. However, MS Power Virtual Assistant is probably more suited for organisations using many Microsoft tools, as it provides integrations to other Microsoft services. If you’re not one of these, it’s worth considering:

• There is a higher chance of lock-in – you may not be able to take the chatbot elsewhere should you stop using Microsoft or want to switch to another vendor as LLMs evolve and new players enter the market.
• There is a monthly fee.

In contrast, a custom ChatGPT integration means the code would be reusable for future projects, with no monthly fee (excluding the cost of using the ChatGPT API).

An API is a pre-built solution that you can integrate into your project without having to develop anything from scratch and is basically a set of instructions on how your app should communicate with ChatGPT.

When you use an API to integrate ChatGPT, you're essentially creating an interface within your app so users can directly ask the bot questions. This is a quick and easy way to get started with ChatGPT, but it doesn't give you the flexibility to customise the bot's responses or workflows.

A custom implementation of ChatGPT gives you the highest level of flexibility. This means creating your own chatbot and then implementing API integration. The process typically involves:

• Defining the chatbot's functionality
• Designing the conversation flow
• Creating the front-end interface
• Building the back-end logic to interface with the ChatGPT API

A custom implementation of ChatGPT can take anywhere from a few weeks to several months to complete, depending on the scope of the project. There is also the option of ongoing maintenance and updates as the chatbot is used and new use cases emerge.
 

Custom ChatGPT integration is a good fit for businesses that require a high level of customisation and control over their chatbot. For example, businesses in regulated industries like healthcare may need to adhere to strict compliance guidelines that require a custom implementation to ensure compliance. Additionally, businesses with unique user journeys and use cases may benefit from a custom implementation to create a chatbot that aligns with their brand and user experience.

Whilst ChatGPT offers many benefits, it's important to be aware of its limitations and potential flaws. As a result, ChatGPT still requires extensive fact-checking. For example:

• Inaccuracy: ChatGPT can sometimes generate inaccurate or misleading information. This is because it is trained on a massive dataset of text and code, which includes both accurate and inaccurate information.
• Bias: ChatGPT can also be biased, reflecting the biases that are present in the dataset it is trained on. This can be a problem if ChatGPT is used to generate text that is intended to be factual or unbiased.
• Lack of understanding: ChatGPT can sometimes misunderstand the context of a conversation, which can lead to it generating irrelevant or nonsensical responses.
• Computational requirements: ChatGPT is a large language model, which means that it requires a significant amount of computational resources to run. This can be a problem for businesses that do not have the necessary resources to run ChatGPT.

Although ChatGPT is a dominant player in the AI space, there are a number of open-source alternatives that offer similar functionality and performance – and which Blueberry can use instead. These alternatives are typically less expensive and easier to deploy than ChatGPT, and they can be customised to meet the specific needs of a business. Some of the most popular open-source alternatives to ChatGPT include:

LLAMA 

• LLaMA (Large Language Model Meta AI) is a large language model (LLM) released by Meta AI in February 2023. It’s trained on a massive dataset of text and code. And offers similar performance to ChatGPT. LLAMA has been shown to outperform ChatGPT on a number of benchmarks, including the GLUE benchmark and the SQuAD benchmark. LLAMA is available free under an open-source license, which means that it can be used by anyone.

ColossalChat 

• ColossalChat is an open-source project that is based on the LLAMA model. It offers a number of features that are not available in ChatGPT, such as the ability to generate text in different styles and tones. ColossalChat is available free under an open-source license.

Alpaca-LoRA 

• Alpaca-LoRA is a highly efficient tool for fine-tuning language models such as LLAMA. It uses the LoRA (Low-rank adaptation) technique, which offers a number of benefits over other fine-tuning methods, including: 
  - Greater speed and less memory consumption, making it compatible with consumer hardware.
  - Smaller output size (megabytes instead of gigabytes).
  - The ability to combine multiple fine-tuned models during runtime.
  Alpaca-LoRA is available free under an open-source license.

Vicuna 

• Vicuna is an open source chatbot that is powered by the RWKV language model. It offers similar performance to ChatGPT, and it is available for free under an open-source license. Vicuna is known for its excellent customisability and adaptability. It can be fine-tuned to perform a variety of tasks, and it can be integrated with a variety of third-party services.

OpenChatKit 

• OpenChatKit is a complete toolkit for chatbot application development. It offers a number of features that are not available in ChatGPT, such as the ability to train custom language models and the ability to integrate with a variety of third-party services. OpenChatKit is available free under an open-source license.

Blueberry specialises in building custom software solutions for businesses of all sizes, and we have recent experience of working on projects requiring integration of generative AI assistants. If you have a requirement for a custom chatbot for your business, please contact us. We can work with ChatGPT as well as its open-source equivalents, depending on your needs and tailored to your specific requirements.