Well it started badly/well, instead of traveling to London for the event, I opted to watch remotely. It is a pity I would have enjoyed meeting people and talking face to face. Still AWS have done a great job of making the process easy, and the tracks are interesting, I’ll let you know what the follow up with suppliers, vendors etc is like, I’ve reached out to a few, as I’d like to check out the offerings/understand the material and learnings available.
I’m hoping you will be able to catch / watch the recording after the event. The link is here AWS Summit London 2025. / 30 April
The link will be emailed to you, the direct url at this moment is here https://summitlondon.awslivestream.com/
Below are the list of on Demand sessions and then the “live show starts at 10:00 GMT”. There are a lot of sessions, and all of the tracks are available for consumption.
There are Tracks dedicated to each “persona/user type” it is nicely laid out. (examples:- Business, Data, Developers, and something called Generative AI – what ever that is….) The technology which is going to change the world, and with the reducing in Compute requirements for new more efficient models, the sale of the Data Center needs/requirements are not going to be as “large” as initially expected, don’t get me wrong, if everyone on the planet needs every a small virtual compute increase, well, it’s that a lot of small numbers mounts up….
AWS are investing in Skills for all, skills for life.
The keynote was amazing. The speed/pace and use of Generative AI
David (Dave) Brown – AWS head of Infrastructure
Largest network provider, in the world – (claims)
Outpost enhancements are a big call out, Edge computing with the same controls and toolsets
Graviton – customer CPU
new cluster – Project for the above
S3 storage – The size/scale and tiers….
Meta data for tagging/flagging of data types/location/need
New table in S3 for Iceberg table
Trend analysis is now possible.
WS Sagemaker enhancements
New single unified studio
Partners and consumer stories :-
Michelle From Untold Studios
Amazon Bedrock:- Models at this moment – more coming.
Data within your model, customisation
RAG original made for unstructured data. diversifying Bedrock knowledge base
Guard rail to limit/block data leak.
Robin.AI – UK based startup. – highly regulated. Legal model injection, and model distillation for speed/enhancements.
AI Hallucinations. Automated reasoning, (for logic testing, crypt testing). – same models can drive down the uncertainty
Blocking possible hallucinations
Inference costs are still high, AWS is working on this, cost against performance. Cost optimisation work is ongoing. Bedrock has a number of guidance
Model distillation (reducing the model size/scope)
Intelligent prompt caching/routing
Utility – AI Agents (Agentic)
Retail, commercial and private banking
UK Finance
Fraud is apparently the biggest crime by volume/number in the UK
Looking at Amazon Q for later this year.
Look what Q can do….
Mainframe / code move
Cost saving of removing legacy code.
Migrate Vmware networking and firewalls to AWS controls
refactor cobol to Java
LSEG
Daniel Maguire LSEG London Stock Exchange Group
AWS Outpost will be created, consumed this year.
Mission critical business cases, governance
LSEG Product: Wellcheck
Data quality will be a focus for 2025