Latest edition of our newsletter, Notes from the forefront of digital engineering, guest illustrated by DALL·E 3.
Hello there,
It’s a while since we’ve updated, as we’ve been busy on a number of stealth projects. Meanwhile, we noticed that lots of companies out there using AWS (and probably other cloud service providers too) could be making significant savings. And, of course, we’ve been experimenting with generative AI. This edition of our newsletter is guest illustrated by Open AI’s DALL·E 3.
Read on for details, and enjoy!
Five steps to reduce your monthly AWS expenses 💰
We do a lot of work with AWS. When we’re reviewing a client’s configuration, we often spot opportunities for them to significantly reduce their monthly expenses. In our latest blog post, Loft’s lead engineer, Rimvydas Zilinskas, shares five essential steps you can take to ensure you are making your money work hard for you. If you haven’t implemented any of these yet, you could easily find yourself saving 30-50%.
Here’s the blog post: Our 5-step plan to reduce your monthly AWS spend by 30-50%
❤️ Complimentary AWS account review
If you are spending a minimum of £300 per month on AWS, we can offer a complimentary initial review of your AWS setup. We recently helped one client save £10,000+ per year with just a few of hours work. If you are just getting started with AWS or planning a new project, we can help you plan it in the most efficient way. Please reach out to me, [email protected] to get started.
Stealth mode 🤐
We’ve been working on a number of stealth projects since mid-last year.
One is a collaboration platform for a top 20 global pharma giant. Another is a data automation project to help one of the UK’s most exciting cycling brands grow their business faster.
Keep your eyes peeled as we update on work in due course.
Fun with DALL·E 3 🤖
We’ve been having some fun with AI of late, and in particular experimenting with and comparing different generative AI models to examine the results and opportunities.
All the images in this newsletter have been created with DALL·E 3, Open AI’s latest text-to-image system built on ChatGPT. It will be available via their API and in Labs later this year.
It took about half a day in total to create the images, mostly time spent refining and developing text prompts. I asked the model to create the images in the style of a tech comic. I’ll let you judge how well that’s been achieved – but here are a few initial observations:
Pros
- Based on generative artificial intelligence, ChatGPT.
- Quick and easy to create complex, professional looking images, tailored to specific needs.
- Responds to detailed prompts about what elements to include in the image, what style to use, lighting, colour, and production info such as dimensions or aspect ratio.
- Website states that you own the images created (subject to certain things).
- Use ChatGPT to brainstorm image ideas and then feed into DALL·E 3 to generate the image.
Cons
- Tendency to generate derivative-looking or same-y output in response to basic prompts; you need to keep iterating on ideas and provide rich prompts to get something more interesting.
- Tendency to put too many elements in the image, even when you instruct it to keep things simple – rather than home in on the one or two elements that would make it really powerful.
- Text on images tends to go wrong.
- A challenge is to get the right sort of look and feel, and consistently so. Need to explore how a brand might be able to use this to ensure all output aligns to brand guidelines.
Given the extraordinary cost savings and time savings of being able to create images like this in a few hours, it seems well worthwhile.
DALL·E 3 is paid for, but you can experiment with Microsoft Image Creator or via Microsoft Copilot, which are both free and powered by DALL·E 3, but the licence terms are unclear to me.
That’s all for now… onwards and upwards!
Nick and the team
