I couldn't sleep, so I decided to try building a Pulumi C# application that uses an existing MCP Server. My forms will utilise the client to allow me to access my Azure subscriptions and resources - wow. Build a really cool tool quickly - Claude Sonnet 4 is once again significantly better than GPT-4.1 for programming with GitHub Copilot.
Thursday, 7 August 2025
Wednesday, 30 July 2025
AI for developers and Architects
The cost of prototypes is unbelievably low using AI.
Rapidly creating a prototype, especially with new or less well-known technology, is where I derive significant benefits from AI.
How to build application prototypes?
- Write /reverse prompt/Adjust instructions into md file
- Agentic AI (specialising in Doc Extraction) to extract and refine from md file
- Run using IDE-based copilot (VS Code with GitHub Copilot) (AmazonQ) (Cursor, Windsurf, Steamlit)
Thoughts: Developers are adjusting to using Ai to support software solutions. The developer role will continue the trend of making technical implementation more accessible, allowing knowledgeable IT engineers or domain experts to build faster and better than citizen/amateur developers. Ai assists in complex decisions!
What needs to improve?
- Knowledge is key. AI needs to have narrow expertise at the right time. i.e. only domain knowledge, not influenced by other data. Quality of input data used to train. Allows for dynamic reasoning.
- Session/long-term contact agreement/understanding to improve the understanding between your IDE and me. Remember how I prompt and provide feedback on how I digest information. Context between the human developer and Ai is Paramount.
- Control of IDE integration with coding copilots, clear return to the user developer to make better decisions. Context is Paramount.
- Governance & Data (Connectors, API's, code complex processes (MCP maybe), quality of data).
Retrieval Augmentation Generate (RAG)
Model Context Protocol (MCP)
MCP is a protocol (created by Anthropic) that allows an MCP client to connect to an MCP server, which in turn has specialist knowledge that the MCP server will expose. Authentication uses OAuth to secure access.
My Applications/Agents use the MCP to ask the MCP Server, 'What can you do?' so that they are aware of how to use the MCP Server.
The MCP Server, when built, informs the client of its capabilities and then performs actions such as updates using an API.
Summary: Use MCP to allow the client to talk to other resources/tools
Agents-to-agent (A2A)
A2A allows agents to work together. So two agents can leverage each other; the other agent solves the issue and returns the answer for the first agent to use. Whereas MCP allows any agent to speak to a source. Agents complete a task and give it back to the calling agent.
Summary: Use A2A to talk to specialised Agents, and the agent returns the calling agent's answers.
Wednesday, 2 July 2025
Artificial Intelligence as a mega trend
Overview of AI
The world has gone mad for AI. Constantly, I see overhype and poor messaging leading to a misunderstanding of potential.
AI is not new; the ability to use it for commercial gain is new at the scale we now have. AI excels at helping us identify patterns, gather information, and is primarily a decision support system (DSS).
AI is not intelligent, but rather good at making complex decisions, and it has biases that we teach it.
This means AI is useful for specialisation, not a generalisation of "smartness", now that ChatGPT et al. are wide-ranging, people are assuming it is a general-purpose tool. Actually, ChatGPT is a specialist in breaking down and grouping language based on a data source. For those in the technology industry, we pretty much know that ChatGPT is a good Google (search engine).
So, what is AI going to be successful at? Well, this is my prediction:
AI will have a massive impact on many industries:
1. Healthcare - guess what? More surgeons and people will be needed, not fewer. Here, I focus on Healthcare examples. People need to interact with others; avatars are a joke. I can talk to Alexa already. There is very little to nothing in this space except for snake oil salesmen. Please prove me wrong! More skilled people are needed.
2. Software Development/IT - This is a significant one. Programmers' roles will change significantly; those with a good understanding and knowledge will thrive, while those with superficial knowledge and a lack of ability to truly understand and work through challenges will likely disappear. Technologists will focus on challenging problems and introduce significant improvements to all business processes. The amount will continue to grow. There is not a lot of agentic, "smart AI" in the space, and we are 50 years away from this, imo.
3. Manufacturing - it won't make the impact that the media says it will. We are good at manufacturing. The sub-functions that will benefit include machine maintenance, sensor usage, and performance/behaviour will change. This will allow you to improve Machine Maintenance (MM) and scheduling. Think of railway lines; they need to be shut down, and it costs millions to trim hedges. Imagine now that you know the level crossing "lifty uppy-doowny" box/bar is showing signs of fatigue. Shift the fix left and save the unscheduled breakdown; the train line and knock-on effects shall result in massive improvement. We are already proficient in manufacturing and, to some extent, automation. If the AI is not significantly better, it is not worthwhile. More skilled people are needed.
 |
| Machine Maintenance in Manufacturing. AI is needed to mature MM. |
Techniques such as defect detection are already well-established using Visual AI at the micron level. Rubbish detection. Using AI will be beneficial - sure, it will become cheaper and easier to acquire this system capability for less, but AI is merely an enabler, and it has been available for well over a decade. More skilled people are needed.
4. Service Industry - Robots serving people, please, it's mad, except at MacyD's (McDonald's) and honestly, minimum wage workers are pretty efficient there, and it will be too sterile. Pushing out patties, well, if you need AI for this, you don't know what AI tries to do. AI & automation are already in the processing and packaging processes. The big stuff with AI will be in social media and advertising (and don't get me started there, automated advertising will absolutely fail. We need to invent a missile to destroy non-human posts). More people will be required in these new and changed services.
Analogy:
1. Old technology: Hand weaving material was a big profitable business in Britain; along came looms; these workers got upset and broke the looms and ended up in prison or broken; these were the Luddites (who refused to embrace technology). The Luddites ended up broke, and all this could have been avoided by embracing technology, as they were the most knowledgeable about materials and production. They are the natural experts.
2. Trend jumpers on: Too many companies wanted to build looms, and a handful of players did brilliantly and still exist today. Think Microsoft and AWS; they are transitioning from being programming technology companies to AI technology. They still solve the same problem of process improvement. The weavers who decided to go into building and repairing looms did exceptionally well, but ultimately ran out of demand, and their prices were driven down as there was an excess of supply. Still a good change. Many people also got hurt here. Be careful inventing new technology in processes; you get it right, you are a hero; get it wrong, go find a new job. Lots of sales silver bullets are being produced. There are tons of "AI experts", but mostly this is absolute rubbish. With rare exceptions, you are not an AI expert unless AI was in your job description more than 5 years ago. Beware the snake oil salesmen, nowadays they come in many forms, sizes and shapes :)
3. Embrace change: Normal, common-sense (smart) people realised they actually had 4 options:
- Learn how to use a loom. Use the technology available and use it to build garments faster;
- Build looms and support the loom business;
- Do nothing, continue to offer hand-weaving labour to the market. So take your pension and hope like hell you win the lottery (I'm still backing this option for myself); or
- Expert hand craftsmen or women :) Become the best hand weaver in the world, and people pay you for your expertise; these people's descendants/business still exist. But big surprise: it's hard, it takes a long time, it's unlikely to make you rich.,, So, sure, go do this if you are a genius in your field and love it, but don't die of surprise when you go broke or don't get the return you deserve for all that hard work.
Summary: Embrace technology and AI, it is only a decision support system. More skilled people are needed, as you have the background, being professional and embracing change means you are more in demand. Sitting on your backside waiting for the lottery means you are like 90% of people, and you'll get 2 jet skis and a new husband! yipee.
Healthcare
Good Use Case: Diagnostic medicine
Diagnostic medicine has become the centre of healthcare, and the ability to use AI, which is better at detecting abnormalities than the best radiologist using a single trained model, yields results in near real-time. This means that consultant radiologists and specialists can receive reports in seconds that are of unparalleled quality. GPs have the best guess within seconds, rather than well... we all know this.
AI also provides probability, so it's easy to prioritise any reporting that's life-threatening to a specialist, allowing them to focus on the most challenging work and receive the in-depth information provided by the AI.
This is possible because we are dealing with a relatively narrow field of data that we have taught AI to handle. Think of X-rays; the results are far superior to an expensive resource (a Radiologist) that takes at least 12 years to train. And more to get brilliant.
Should we stop training Radiologists and diagnosticians and allocate our resources to AI? Absolutely not!!
Radiologists should be using the AI reports, validating, using the info and extrapolating, when an issue is detected, this must be added back into the learning model resulting in improving the AI. AI should not act. It must only be used to support. Acting should be restricted to notifying relying parties such as GPs.
Good Use case: Online GP Appointments and triage
If you have an issue, you go onto an NHS app that will ask you for your symptoms and ask a few follow-up questions. It will only give you its best guess (this is already amazing imo.), this in turn, will triage your call into "go to your emergency department, they know you are coming", "let's book you an emergency appointment", or "this is my recommendation, and why". Dr Google actually becomes useful (weird medical insider joke). Honestly, we could do so much more, but care is given to the right people, "shift-left" (the sooner you catch it, the cheaper and better the solution, 100% applies to healthcare).
Preventive medicine and nudging technology will have profound improvements in people's lives and lifestyles. Hooking into ambulance services and driverless automated vehicles,.. people do the hard stuff and make the decisions. AI does the piece efficiently and quickly that we as humans aren't good at. Hopefully, you are understanding the difference between narrow and wide industries.
Bad Examples: Robot Surgery or treatment rooms
Robots replace people in operating theatres. It is insane!! A surgeon could use AI to obtain better diagnostic data more quickly; they could even utilise technology like AI-assisted operations, then send messages if it detects that the actions are not optimal or there is a clear risk that a priority has changed. It is brilliant for decision support. It's not a good idea to try to automate complex decision-making.
My AI Posts
AI Model Comparison (this post)
This post does not look at Strategy and Purpose
All AI solutions give themselves an exponentially better chance of success, regardless of industry, if they have a strategy, purpose, and FAIR data (Findable, Accessible, Interchangeable/exchangeable, and Reusable).
Saturday, 28 June 2025
UK Railway Industry for Dummies focusing on Rail Infrastructure
Rail assets are organised into large hierarchical asset classes that are interdependent to make up a rail system. These rail assets are organised using detailed, lower-level assets built from taxonomies and ontologies tailored to each jurisdiction within the rail industry. Railway interaction and operation of assets must conform to various stringent rail regulations. Safety has a massive focus.
Taxonomy organises data hierarchically, while ontology
models both hierarchies and complex relationships between entities and their
properties. In the rail industry, ontologies are crucial for successfully
modelling assets.
The picture shows examples of significant assets (high-level)
 |
| Main Railway Infrastructure Assets high-level overview. |
 |
| An AI-generated image to explain commonly used railway terms. |
The railways consist of "rolling stock, rail infrastructure, and environment"; these components have multiple relationships with one another.
1. Rolling stock is the trains.
2. Rail Infrastructure relates to:
2.1. Electrification/power/energy, generally used for power supply for signalling, train power, and telecoms.
2.2. Telecommunication, track-to-control, and train-to-control are used to communicate, including sensors and IoT devices.
2.3. Signalling relates to ensuring train safety so the train knows there is a train ahead of it, and issues when to slow down.
2.4. Track Engineering, also known as Rail Engineering and The Permanent Way, involves the rails, connectivity, support, extensive physics and geometry, steel rail installation and joining, ballast (the ground on which the track is laid), drainage, substructure, and sleepers. It gets detailed with rail joins (Fishplated) and even the welding process used. Fastening types, baseplates, sleepers, off-track maintenance such as hedge trimming (you won't believe the rules unless you work in the rail industry) ...
3. The environment refers to the existing conditions before the railway, including the topography and type of terrain, bridges, and rivers.
The interdependencies with the rail industry are perfect for numerous AI scenarios. With all AI, you need high-quality data, and it must be secured appropriately. Bring the information from the various business functions, allowing for automation, ML, AI and better decision-making.
Each country or jurisdiction has different rules for trains, and operators must comply with Health, Safety, and Environment (HSE) regulations. There are industry rules adapted to each jurisdiction and standards that vary between regions. For example, most jurisdictions have a gauge width requirement; in the UK, the standard gauge is 4 feet 8 1/2 inches (1435mm). There are exceptions, such as heritage railways in the UK. There are manufacturing standards for everything. EN13674 is the British Rail specification for the actual pure steel used to manufacture the track to be installed.
ISO 55000/1/2 addresses Physical Asset Management. Building Information Modelling (BIM) enhances the design and construction process, and both apply to Rail Infrastructure. There is generally a disconnect between Asset Management and BIM, and International Foundation Modelling (IFC) tries to help build a standardised set of assets for the railway business; we are on v4.3.
References used:
Permanent Way Institution (2023) Understanding Track Engineering. London: The PWI. Available at: https://www.thepwi.org/product/understanding-track-engineering/ (Accessed: 4 July 2025)
Camarazo, D., Roxin, A. and Lalou, M. (2024) Railway systems’ ontologies: A literature review and an alignment proposal. 26th International Conference on Information Integration and Web Intelligence (iiWAS2024), Bratislava, Slovakia, December 2024. Available at: https://ube.hal.science/hal-04797679/document (Accessed: 4 July 2025).
Network Rail (2021) Asset Management: Weather Resilience and Climate Change Adaptation Plan. London: Network Rail. Available at: https://www.networkrail.co.uk/wp-content/uploads/2021/11/Asset-Management-WRCCA-Plan.pdf (Accessed: 4 July 2025).
Thursday, 26 June 2025
openBIM for AEC understanding
Within the AEC industry, standards are necessary to ensure that all project stakeholders are speaking the same language, thereby improving collaboration. We can also process data to automate various processes if the data is standardised.
BIM (Building Information Modelling) is used to improve collaboration on infrastructure projects. BIM is essentially ISO 19650, and it has various levels.
Building Models contain 3D information that shows how assets fit together. Each of these assets may contain properties that can be used to look for clash detections. Think of a CAD diagram, it lays out the plans for a building so all parties can see the proposed plan. As CAD technology advances, you can add more information about the project. For example, as an electrician, I only want to see the layers that affect my work. CAD can be further extrapolated to show products and material information.
closedBIM: These were the original big BIM systems, including AutoCAD, Revit, and Bentley ProjectWise. These tools feature visual editors and viewers, allowing them to securely store the files needed for a project and ensure that the appropriate people have access. These all have their own proprietary standards.
openBIM: Read other parties' data, improves collaboration and consensus. Easier to switch tools to reduce costs or get better features. Consists of:
- IFC (common language)
- bSDD (industry common language)
- IDS (Requirement specification)
- BCF (check)
- openCDE (sharing with APIS)
Industry Foundation Classes (IFC) serve as the basis for standardising how information is handled. Has standards for location, such as geographic information. Materials, Geometry, and Spatial Structures are covered by IFC classes. In each industry, these base IFCs are added to. The BuildingSmart bSDD is an extension of IFC for specialised industries and sectors, published to provide more specific, agreed-upon standards.
Project Requirements: These can vary, but having an agreed-upon format, such as an Information Delivery Specification (IDS), is helpful. Although it is not necessary or widely used, it ensures that precise details are provided. Therefore, collaboration allows all parties to clearly understand what is needed using IDS.
IDS uses bSDD, which is based on IFC, so that the requirement specifications are precisely laid out.
openCDE defines technical interfaces, .....
Thursday, 5 June 2025
AI Vendor Management - Formiti
AI is going crazy, and you can build your own but generally you need to look at a supplier, so it's worth understand management of Vendors, you as the controller using their service are at risk of them not making their AI operations transparent. It's a big business risk to my clients.
GDPR is closely linked to AI, and if you use a service/vendor, the reputation and fine risk may fall on you as the provider. Need visibility into each vendor, how they are using AI, in turn they are using vendors so it's a nice complex dependency problem. You need to be aware of what you are relying on.
Ensure contracts with vendors consider AI, how the process your data and how their sub process vendors do the same.
Track website customer behaviour, we use a vendor to clean up the data. In turn, I have no idea that they are using AI outside of the UK or EU. Follow the dependency chains as all this needs to be transparent to the end customer if needed.
Monday, 2 June 2025
Copilot Studio 2025 Notes
Copilot Studio is fantastic, the licensing is complex, and the AI integration is excellent. Architects really need to understand Licensing and billing, or AI will get out of control. The Purview and governance look very good. Copilot Studio Cost Estimator (preview June 2025)
MS Build 2025:
MCP Server in Preview - possible to collect data from other AI services or write back.
Connector Kit - So, you can add custom connectors from Power Platform Connectors, including Copilot Studio - great stuff.
Agent Flow - Added functionality to Power Automate flows (Copilot Studio aware), deployed via solutions.
Note: The M365 Agent Toolkit appears to be an interesting tool that allows agents to perform tasks using Office add-ins with VS Code.
Licensing
You need to be aware:
- M365 agents - require all end users to have M365 Copilot licences, retailing at $20/user. Alternatively, users can consume the agents using a PAYG model per message (it racks up quickly). I can add these to MS Teams, and it appears that people with licences can ask the M365 agent, while others can view the results (some more testing and understanding are needed here by me).
- Copilot Studio - Requires a Copilot Studio AI Studio/maker licence at $30/retail. Users don't need a licence to use it, but you pay per message, and this can rack up quickly, so watch your usage. Buying bulk message credits can help reduce costs.
- Each prompt generates multiple messages, which are all billable (complex to calculate)
- (If you use Copilot Studio and it calls Azure AI Foundry, it also bills Tokens (also complex to estimate)
- Copilot Studio utilises the AI Foundry connector through its Premium connector.
Subscribe to:
Posts (Atom)