Eric D. Boyd

Business Architecture is often misunderstood and incorrectly categorized as a mere component of Enterprise Architecture. This concise exploration aims to dispel the marketing hype and provide clarity on this crucial yet frequently misinterpreted field. Business Architecture - A Distinct Discipline from Enterprise Architecture Watch on Youtube: https://lnkd.in/ehViR8Zj Watch or listen on the ArchitecturesCOE website: https://lnkd.in/e5uNQz7d Listen on Apple Podcasts: https://lnkd.in/efJ_JGev Listen on Spotify: https://lnkd.in/eY9eYDuq Listen on Audible: https://lnkd.in/eqqwhmCm #RealTalkWithSamHolcman #TipsFromSam #EACOE #BACOE #EnterpriseArchitecture #BusinessArchitecture

29

This is a useful dashboard if you are considering what models to use for your agentic architecture implementations. Having the cost profile is handy as well. This is obviously no substitute for experimentation, but useful to have a *more* objective confirmation of what you are experiencing, or perhaps a useful guide on what to play with when getting started. Full link below.

35

2 Comments

A consistent set of components is required for Business Architecture success; there are four “keys:” • A thinking tool - a frame of reference - a framework • A methodology that is a "recipe" (Not just a general body of knowledge) • Human-consumable Architecture models • Human-consumable Relationship models These four elements are used to develop the true result of Business Architecture - a prioritized set of Capabilities that enable the Business strategy. The techniques outlined in BACOE Business Architecture Programs are universally effective techniques, regardless of the planning domain. Learn more about developing, delivering, and managing Business Architecture on our website: https://lnkd.in/gW63fFzd #BACOE #BusinessArchitecture #BusinessArchitect

22

New version of #OutboxKit out, focused on resiliency. Not much user facing going on (other than renaming some metrics), but the new version has some resiliency improvements to help when there are infrastructure issues (i.e. problems with the database or message broker). #DotNet #CSharp #OutboxPattern (link in the comments)

12

2 Comments

Consider the humble indicator. This column specifies whether an entity instance does or does not have a particular attribute. Which is bizarre if you adhere to third normal form. More importantly, the indicator sits one sematic level above the attribute it is referring to. Thus, the definition of the indicator is whether the attribute is found in the entity instance, but no definition of the attribute is (or need be) provided. This is a nice piece of ontological sleight of hand. Further, no value of the underlying attribute is possible in an indicator column. There are many lessons in this, but one is that data is not an exact representation of reality; rather it is an abstracted model of it that serves one or more purposes. #data #datagoverance #datamodel #database

12

6 Comments

This is neat, a tool to automate migration to central package management for .NET solutions: https://lnkd.in/gXiEUNVu Central package management (https://lnkd.in/gxPt3ds8) lets you centralize the package versions used in a solution with many projects, eliminating mismatched versions. For a large solution, manual migration is tedious but this .NET tool made it a snap.

128

7 Comments

A new curriculum on ChapterPal: Must-read papers on transformer architectures This curriculum curates the seminal papers on transformer architectures, designed to allow the reader to learn this critical scientific domain from scratch. Learners will gain a comprehensive understanding of core concepts, architectural innovations, and practical scaling techniques, essential for building advanced AI. A link to the curriculum: https://lnkd.in/ejVfNduM

37

2 Comments

APIs were never built for agents. MCP is. Remember when HTML standardized how browsers render content? Or when drivers abstracted hardware differences so operating systems didn’t need to care which printer or keyboard you plugged in? Those shifts didn’t just make life easier—they unlocked entire ecosystems. We’re at a similar turning point now. APIs are great—for humans. We built them so developers could manually wire up integrations. Every service—Gmail, Slack, Calendar—requires its own key, its own logic, its own custom fit. That’s fine when you’re coding one integration. But what about agents? Agents think differently. They don’t care how to fetch an email or find a file. They just want to do the job. But today, they can’t—because they’re stuck behind the same walls we are: auth flows, brittle APIs, service-by-service logic. Enter MCP. MCP is like HTML or drivers, but for agents. It provides a consistent interface to interact with everything—from cloud services to local files. You still authenticate once per service, but MCP handles the rest. So the agent doesn’t need to “know” how Gmail works. It just asks MCP for what it needs. Security: centralized, not simplified With MCP, each service—Gmail, Slack, local files—still requires its own login. OAuth flows, API tokens, and permission scopes remain in place. But instead of every agent handling credentials separately (and inconsistently), MCP centralizes access. This means: - You authenticate once per service, securely - Permissions are managed in one place - Access can be logged, monitored, and revoked easily For security teams, it’s a win: fewer custom scripts, more control, and clearer audit trails. For agents, it’s plug-and-play access—with guardrails. Why does this matter? Because agents won’t truly change how we work until they can operate across tools—autonomously, securely, and reliably. That’s the world MCP is making possible. A world where your agent checks your calendar, drafts your email, books your flight—and you didn’t have to alt-tab once. We’re not just automating tasks. We’re rethinking interaction itself. #AI #Agents #TechLeadership #MCP #APIs #FutureOfWork #Automation #EnterpriseAI The AI Executive WhitegloveAI

29

4 Comments

Today's learning insight: The Application Services Framework (ASF) offers a simple and powerful way to create REST provider services for #PeopleSoft! When developing ASF endpoints, I like to focus on REST URLs, particularly in terms of module and resource. For instance, if I were to create a set of services for managing User Profiles, I would assign the Application Service URL ID as "security," with "users" serving as the root resource. My URL pattern would then be .../security/users/RCHANNING, etc. With minimal metadata generation and a small amount of PeopleCode, I can establish a functioning REST endpoint, complete with an Open API specification for integration with other solutions. To learn more, check out our Integration Tools Update course, which is available live virtually or on-demand at https://lnkd.in/dQGGueVs.

26

Ten minutes to prototype. But technical debt accumulates faster than ever before. Yet I believe GitHub Copilot's 55% productivity gains fundamentally outweigh debugging overhead when examining comprehensive business impact data. I'm witnessing an unprecedented paradigm shift in software development that extends far beyond simple coding acceleration, fundamentally transforming how organizations conceptualize, validate, and iterate on complex products. Reddit's CPO confirms engineering teams conceptualize ideas one day and deploy functional prototypes the next morning. Traditional boundaries between discovery and testing have dissolved as AI enables "vibe coding" - exploratory development without extensive upfront architecture. Small companies with minimal engineering resources leverage these tools as competitive advantages, moving with unprecedented organizational agility while established enterprises struggle with legacy architectural constraints. However, Google's DORA report reveals 25% AI usage increase leads to 7.2% decrease in delivery stability across enterprise organizations. I counter that forward-thinking organizations treat AI prototypes as strategic starting points while maintaining rigorous architectural oversight and governance frameworks. Smart enterprises adapt without abandoning engineering best practices. Transformation accelerates.

16

1 Comment

⚙️ “Clicked a button — and an entire workflow just kicked off. No plugin. No backend queue. Just JavaScript + Flow.” Power Automate + Model-Driven App + JS = Instant Power. Want to trigger automation right from a form ribbon, command bar, or onChange event? You don’t need code-heavy plugins anymore. With just a few lines of JavaScript, you can fire a cloud flow securely. 💡 Why This Matters: 🔁 No need to create custom actions or background workflows 📲 Instant business logic tied to front-end events 🛡️ Secure and audit-compliant with proper authentication 💻 Seamless user experience — no page reloads, no waiting 🚀 Here’s How to Do It (High Level): Create a cloud flow with a HTTP Request trigger Make the flow accept dynamic parameters (like recordId, userId) In your model-driven app, use Web Resource JS: XMLHttpRequest or fetch() to post to Flow URL Secure it using Azure AD App Registration (avoid public endpoints) 🧠 Pro Tips from the Field: ✔️ Use Xrm.Page.data.entity.getId() to get record ID dynamically ✔️ Always include error handling for failed flows ✔️ Store flow URLs in environment variables, not hardcoded ✔️ Add telemetry via Application Insights for debugging #PowerPlatform #ModelDrivenApps #PowerAutomate #JavaScript #CustomRibbon #LowCodeEngineering #MicrosoftDynamics365 #SolutionArchitect #ShrikantSharma #IntegrationBestPractices Amit Dalakoti Saurabh Khosla Ananya Singh Bipin Bhandari Naman J. Rishabh khokher

13

This AIO usability study is absolutely amazing work from Kevin Indig https://lnkd.in/eQcd-eFc You should read it, but one of the most compelling validations is that younger folks prefer AIOs. Older folks prefer classic organic.

107

14 Comments