Getting Started
Why do we need AI for business?
Enterprise AI can greatly enhance your company’s productivity and reduce cost especially when you have vast amount of existing data. We are sure you have used OpenAI’s ChatGPT and experienced its jaw-dropping intelligence in providing you with enhanced information. Imagine having a ChatGPT for your own enterprise use. That is the benefit of enterprise AI – to help your organization save time and exponentially boost output.
We can take your enterprise data and enhance it with AI capabilities so that your team can save time from sipping through data but instead focus on strategic thinking and decision-making.
In what areas can we apply AI in our business?
If you have a vast amount of data and wish to make sense of it, AI is the right choice. For instance, if you have multiple data points of your manufacturing line, you can use AI to inform you on where you can optimize your manufacturing process.
If you have an internal knowledge base and wish to enhance it to provide more insightful information to your staff or customers, you may implement AI-powered semantic search or AI chatbots.
Do you practice ethical AI?
In our practice, we strive to develop AI systems that are accurate, impartial, transparent, respectful, and safe for anyone to use. So, yes, we do embrace ethical AI.
For example, we do not process sensitive data such as financial information (bank account details, health records, social security numbers) or personally identifiable information (passports, driver’s licenses).
Processing Data
What is data preprocessing and how do you do it?
Data pre-processing is a process of cleaning and refining your data to ensure that it is ready to be used to train an AI model. The cleaning process involves removing unwanted information or facts that could mislead the AI algorithm into thinking something is true. For example, you may have personally identifiable information (PII) that you do not want the AI to know or tell others.
You may also refine your data in this step by adding more information such as metadata to enhance its context.
Will we have a copy of the preprocessed data ?
Yes, absolutely. We will provide you a copy of the data we have cleaned and preprocessed so that you may use it for other purposes or store it in your archive.
What databases do you use?
The vector databases that we often use are Milvius, Pinecone, Chroma and more. The relational databases we often work on are MySQL, MariaDB, PostgreSQL, Microsoft SQL, Oracle and more.
What data formats do you support?
We prefer JSON, CSV and XML formats. But we can also support Excel, PDF, HTML, Word, PPT, PPTX and even plain text.
We could convert your less ideal data formats to something that is more easy for AI to work with.
Do you support PDF, Word or HTML?
Yes, we can support PDF, Word or HTML. We shall utilize some programming libraries to convert the text in the document to JSON or CSV.
Who owns the copyright to the data after preprocessing?
You do.
Where do you store our data?
We can store your data in our cloud infrastructure or yours. If you use our facility, your data shall be safely stored in our cloud hosting facilities powered by AWS or Google Cloud. In terms of geography, we can host your data in specific locations that you require.
Will OpenAI be able to take our data to train their AI?
No, your data will not be made public to anyone. Thus, AI companies like OpenAI will not have access to it.
AI Algorithm and Models
Do we need to build an AI algorithm?
No, you do not need to build your own AI algorithm or model when embracing enterprise AI because we can use existing AI models made available by the open-source community (BERT, GPT-2) or other premium models (GPT-4, Llama 3).
Nonetheless, you may wish to build your own AI models if there’s a need to do so.
What is AI hallucination and how to prevent it?
Hallucination in AI is a phenomenon where the AI system provides wrong or misleading information to the user. In other words, it lied. This happens when the AI model is fed with too much or too little information which it learned from. It also could hallucinate when trained on wrong or poor-quality data.
We prevent hallucination by implementing a method called RAG – Retrieval-Augmented Generation. This is a technique that allows AI models to access and include information from your databases, before generating a response, resulting in more accurate and relevant outputs tailored to a specific context. Basically, it enables the AI system to look up information before responding to the user.
We fully embrace RAG architecture in our AI development process.
Do we need to train an AI?
Not necessarily. Most of the time, you do not have to. Because training an AI model is an expensive and time-consuming process. We prefer that you use existing AI models, especially open-source ones, to reduce your AI capital expenditure.
Technical Support
Who will manage our AI system once it has been deployed?
We can continue to manage your system by providing technical support during your working hours. Or you may choose to manage it yourself.
Either way above works for us, regardless if we are using your cloud infrastructure or ours.
How can I add future knowledge to our AI system?
You may hire a data scientist or let us do the work for you. If you wish to do it yourself, we can help host and manage a data collection system for you so that your team can collect future data.
Do I need to hire AI engineers or data scientists to support our AI system?
It depends. We can assist you in supporting your system in the long run. Or you may choose to build your own AI support team. We do not foresee that you need to recruit AI engineers but having data scientists can be useful particularly when you want to add more data into the AI system.
What is the job of a data scientist?
Interestingly, the job of a data scientist isn’t like a lab scientist at all. Their role, essentially, is to collect, analyze, and interpret data to provide you with insights into your business.
This job usually requires some programming and statistical skills. Oftentimes, we envisage that their role in your company shall be to preprocess data. But with our AI system in place, their job may be much easier since our system is built with zero to minimal coding requirements.
Where are you based at?
Our team is based in the U.S., France and South Korea. But we cover the world over.
Can you work from our office or data center?
Yes, we can.