Fast track your Generative AI with Daemon Ignite and AWS Bedrock

The recent growth of generative AI has been astounding. In just a few short years, the technology has gone from a curiosity to the cornerstone of our digital future. A third of companies are already using generative AI regularly, and this is expected to reach 80% by 2026. 

Daemon is proud to be an official AWS Advanced Partner, trusted by Amazon to build generative AI applications within AWS. We were thrilled to recently showcase Amazon Bedrock, Amazon’s exciting new generative AI service.

As part of this, we collaborated with the AWS Startup Loft to hold a generative AI deep dive and game day. The event was a blast, and we left feeling more enthused than ever about the possibilities of generative AI. 

Rapid prototyping in teams

We believe the best way to learn a new technology is to get stuck in and use it, which was reflected in the event's design. After a brief introduction to the key concepts, attendees were invited to participate in a morning of hands-on workshops. These focused on how Bedrock can be used to harness the power of large language models, covering everything from chatbots to code generation. 

“We rotated between several workshops led by experts from Daemon and solutions architects from Amazon. We then divided everyone into groups and challenged them to build a prototype application. We found that a number of these groups were able to do this quite quickly.”

Find out more about Amazon Bedrock

After this, the real fun began. In just two hours, attendees were challenged to build a functioning generative AI solution using everything they'd learned so far. Daemon Senior Consultant Motse Lehata was on hand to help, along with several other members of the Daemon team.

Connect with Motse on LinkedIn

Don’t get stuck in the build stage

Building a working prototype in just two hours might sound far-fetched, but this is the beauty of Bedrock. Tasks that would once have taken hours can now be completed in minutes, allowing companies to reach the prototype stage at a much faster rate. As Motse explains, this comes down to two main factors:

“If you’re a startup, you want to bring your offering to maturity without getting stuck in the building stage. Bedrock does a lot of this for you, so you can fast-forward to the point where you’re rapidly building prototypes. 

You no longer have to train your model because it gives you access to a variety of pre-trained models. You also don’t need to find your own dataset, so you immediately eliminate two of the most expensive and time-consuming parts of the process.”

As well as saving time on the building side, Bedrock goes out of its way to be as easy to use as possible. This further accelerates the journey to a finished prototype by eliminating the need for extensive training. Motse tells us more:

“It works very well with these two open source software packages- LangChain and Streamlit. LangChain is an open source module for interacting with large language models. It means you don’t need to learn a whole new library to interact with your Bedrock models. 

Streamlit allows you to rapidly build out a user interface. You don't have to do any training or learn a new interface to build a working demo in very short order.”

Secure up to $50,000 in funding from AWS

We’re big fans of Bedrock, and this is partly because it fits so well with our existing offerings. Daemon Ignite is our rapid prototyping service which aims to build a minimum viable product in just two to four weeks. This is a four part process which closely mirrors the structure of the Bedrock event:

1. A half-day workshop in which we discuss our client’s goals and define their criteria for success
2. A one to two week discovery phase during which we drill down into the client’s organisation and systems to design the most appropriate solution
3. A two to eight week building phase during which we create a proof of value along with the tools to measure its impact
4. A final stage in which we turn the proof of value into an integrated solution, adding maximum value to the client’s business 

By harnessing the functionality of Bedrock and capitalising on our close connections with AWS, we can now offer a version of this service specifically for generative AI. As Motse tells us, this is the perfect way for smaller businesses to jumpstart their AI capabilities:

“With Generative AI Ignite we work with startup clients to rapidly explore, validate and build out their use cases for generative AI. As well as building an app prototype, we also work closely with the client to help them create a proposal to unlock funding from AWS. Up to $50,000 is available for startups, and this is typically in the form of Amazon credits. Because we’re an AWS partner, this can be used to pay for our time and the computing resources needed to run the application.”

Read more: why rapid prototyping is the key to innovation

Something for everyone

This service may be focused on startups, but we have plenty to offer more established businesses. A great example of this is our recent work with KnowMe. We were already working with this client to create an educational video chatbot allowing users to converse with historical figures. 

We initially built this using Amazon’s Sagemaker platform but, when Bedrock came out, we realised we could use it to simplify the architecture. This is one of many examples of existing projects that we’ve been able to streamline by incorporating generative AI. 

Find out more about our work with KnowMe

Another exciting use case is intelligent document processing. Over the years, businesses can accumulate reams of unstructured data. Items such as pictures of customer invoices have historically been very difficult to process, even with the help of an AI model. As Motse tells us, though, this is no longer the case:

“To do intelligent document processing, you used to have to gather a lot of representative data of the output you wanted. If you wanted to train a model to process a document in a particular way, you’d have to gather a lot of examples of that. With Bedrock, you have a generally intelligent model that can be applied to several different use cases. You no longer need to do any training. All that’s needed is some fine tuning.” 

Safety first

Many businesses are understandably cautious about allowing an AI to process their unstructured data. Some of this information will be extremely sensitive and could be extremely damaging if it falls into the wrong hands. 

We’re also mindful of these concerns, so security is baked into everything we do involving generative AI. Once again, this is an area in which Bedrock has proved invaluable. Its security features are second to none:

• All data transmitted to and from Bedrock is encrypted, ensuring that it’s protected during the transfer. 
• The communication between your AWS applications and these models happens entirely within the AWS network.
• AWS employs multi-tenancy principles and logical isolation mechanisms using techniques like IAM (Identity and Access Management) and VPC (Virtual Private Cloud) to ensure that each customer’s data is logically isolated to prevent access from different users and applications. 
• Finally, there’s an explicit agreement not to use your data to retrain any of the models you use.

Don’t get left behind

Generative AI is too important to ignore. Luckily, it’s never been easier to jump on board. By combining our Ignite framework with the features of Bedrock, we can have you up and running in a matter of weeks.

Book a meeting with one of our experts to find out more.