@GergelyOrosz: A good reminder why you can pi...
@GergelyOrosz
7 views
Mar 14, 2025
1
A good reminder why you can pick up GenAI - and you probably should. Real story:
Small company, 5 devs. Last time they hired was 12 years ago. AI comes out: company wants to add AI feature. But they don't have the expertise. So hire an AI agency.
Agency spend 3 months planning:
Small company, 5 devs. Last time they hired was 12 years ago. AI comes out: company wants to add AI feature. But they don't have the expertise. So hire an AI agency.
Agency spend 3 months planning:
2
After 3 months, the present a very complex architecture to build: several services multiple databases, SageMaker models etc, using a language a company is not using (Python - this is a Java shop)
It will take 6-9 months to build
Operational costs will be higher fort this one feature than all of the SaaS operational costs for the company!
It will take 6-9 months to build
Operational costs will be higher fort this one feature than all of the SaaS operational costs for the company!
3
Lead dev who is close to retiring (and has been at the company for 25 years) thinks "this cannot be right, surely."
So he says "screw it." Reads up on GenAI, builds a few prototypes and tells company to drop the agency: they will build it in ~3-4 months, much faster and cheaper.
So he says "screw it." Reads up on GenAI, builds a few prototypes and tells company to drop the agency: they will build it in ~3-4 months, much faster and cheaper.
4
Builds it in less than 2 months. Other devs get excited and all of them pick up how to work with LLMs.
They use Bedrock, embeddings, Postgres for embedding search and vector search.
Don't forget: these devs assumed "we need an expert, surely!"
But "experts" often don't know what they are doing. At least with GenAI where it's relatively new to everyone!
They use Bedrock, embeddings, Postgres for embedding search and vector search.
Don't forget: these devs assumed "we need an expert, surely!"
But "experts" often don't know what they are doing. At least with GenAI where it's relatively new to everyone!
5
And look, this is what you'd call a "more traditional" company. Profitable, small, little to no growth in HC. They use what I'd call "pragmatic" tech (Java on BE - it's a 25 year old service! - and React on FE) It works for them!
But picked up GenAI quick, and made far more pragmatic choices vs "experts"
But picked up GenAI quick, and made far more pragmatic choices vs "experts"
6
Source: I talked with this dev and will write up his and his company's story in a newsletter on an issue about how dev teams are doing "AI engineering" out there.
(Will also name them and the dev in the newsletter - it's a really nice story IMO and shows you can quickly learn new stuff!)
(Will also name them and the dev in the newsletter - it's a really nice story IMO and shows you can quickly learn new stuff!)
7
OK, dev who stepped up to build it and has been with the company for 25 years is @ryanecogswell (he agreed to be shared). Engineer #1 at the co.
I got it wrong: he is not close to retiring (sorry Ryan!): but several of his colleagues are over 65 and are.
This is his story
I got it wrong: he is not close to retiring (sorry Ryan!): but several of his colleagues are over 65 and are.
This is his story
8
THIS is what I see different about AI than any other technology wave the last ~20 years.
This time:
1. Devs at startups jumping on it (obv)
2. Devs at scaleups and Big Tech jumping on it
But also devs jumping on it at:
3. "Traditional" companies
4. Heavily regulated companies
With pretty much the same speed as startups are! Perhaps a little bit slower. Thanks to the practical utility it can bring, everywhere.
Compare this with eg the last major tech wave: crypto. It was mostly #1 and some #2. #3 and #4 almost entirely ignored it
This time:
1. Devs at startups jumping on it (obv)
2. Devs at scaleups and Big Tech jumping on it
But also devs jumping on it at:
3. "Traditional" companies
4. Heavily regulated companies
With pretty much the same speed as startups are! Perhaps a little bit slower. Thanks to the practical utility it can bring, everywhere.
Compare this with eg the last major tech wave: crypto. It was mostly #1 and some #2. #3 and #4 almost entirely ignored it
9
Other takeaway is how AWS seems to have played GenAI just right with Bedrock (even if they have no own high-quality LLMs) This team looked at options and settled on Bedrock. It's secure, they trust AWS, & doesn't train on your data. You can choose your model.
I hear so many companies who are a bit more conservative/worried about their data and:
1. Would want to host their own LLMs...
2. ... but it's a lot of work and is expensive
3. Hear about Bedrock. "Oh, it's what we need"
Home run by AWS
I hear so many companies who are a bit more conservative/worried about their data and:
1. Would want to host their own LLMs...
2. ... but it's a lot of work and is expensive
3. Hear about Bedrock. "Oh, it's what we need"
Home run by AWS