The Lossless Group

Laerdal Strategic Content

3 strategic insights from collections: essays

A New API Standard for chaining AI -- Model Context Protocol

Anthropic launched the Model Context Protocol on November 25, 2024 as an Open Specification, [8cac05] it was a game-changer for AI use and code generation. It allowed for the chaining of AI operations, which made it possible to create complex workflows that could be used to generate code, documents, and other content.
Related: Model Context Protocol, Anthropic, Chain of Draft

What is the Model Context Protocol (MCP)?

The Model Context Protocol (MCP) is an open standard developed to enable secure, two-way integration between AI-powered applications (such as large language models and agents) and external data sources, tools, and services. Its primary goal is to standardize how AI systems access, retrieve, and act on real-world data, replacing the fragmented, custom integrations that previously dominated the ecosystem. [8cac05] [b51563] [b479a8] [081ad2]
MCP draws inspiration from the Language Server Protocol (LSP), which unified how code editors interact with programming languages. Similarly, MCP provides a universal API for connecting AI models with external systems, transforming the integration challenge from an “M×N” problem (each app to each tool) to an “M+N” problem (apps and tools each implement MCP once). [b479a8] [026f85] [081ad2]

How MCP Works

  • Architecture: MCP uses a client-server model:
  • Host applications: LLM-powered tools (e.g., Claude Desktop, Copilot, AI IDEs).
  • MCP clients: Embedded in hosts, these handle communication with MCP servers.
  • MCP Servers: Expose data, tools, or functions to AI systems via a standardized interface.
  • Transport: Communication occurs over JSON-RPC 2.0, using either local (STDIO) or remote (HTTP + SSE) channels. [b51563] [026f85]
  • Core Features:
  • Resources: Read-only data sources (like files, databases, Knowledge Base).
  • Tools: Functions or APIs the AI can call to perform actions.
  • Prompts: Predefined templates to optimize how tools/resources are used. [b479a8] [026f85]
  • Security & Consent: MCP emphasizes explicit user consent, robust authorization, and clear UI for data access and tool usage. [026f85] [91430e]

Who Has Adopted MCP?

MCP has seen rapid adoption across major technology companies, AI platforms, and open-source communities. Below is a summary of notable adopters and their use cases:
Company/ServiceAdoption DetailsSource/Announcement
AnthropicCreator of MCP; integrated into Claude Desktop and open-sourced core SDKs and servers [8cac05] [b9fb3b]
Block (Square)Early adopter, using MCP to build agentic systems for business automation [8cac05]
ApolloIntegrated MCP for enhanced AI-driven workflows [8cac05]
MicrosoftAdopted MCP in Copilot Studio and Azure AI Foundry Agent Service for seamless agent integration across Microsoft 365 [b778b7] [3222d7]
Amazon AWSIntegrated MCP into AWS Bedrock agents for enterprise-scale, context-aware AI [b778b7] [8d2e1f]
GitHubAdded MCP server support to GitHub AI assistants and VS Code extensions [b778b7] [cecf76]
DeepsetUses MCP to power context-aware RAG and AI pipelines [b778b7]
AtlassianAnnounced MCP support for connecting structured knowledge to AI tools [b055c1]
CloudflareIntegrated MCP for AI-driven security and automation workflows [b9fb3b]
Zed, Replit, Codeium, SourcegraphEnhanced AI coding assistants with MCP for deeper context and tool integration [8cac05] [ce07e9]
OpenAI, Google DeepMindAnnounced MCP support for their AI platforms [081ad2] [cecf76]
Windows 11Previewed MCP as a foundational layer for secure, interoperable agentic apps [91430e]

Recent Company Blog Announcements

  • Anthropic: Introducing the Model Context Protocol — Official announcement, architecture, and open-source SDKs. [8cac05]
  • AWS: Unlocking the power of Model Context Protocol (MCP) on AWS — Detailed overview and enterprise use cases. [8d2e1f]
  • Microsoft: Announcing Model Context Protocol Support (preview) in Azure AI Foundry Agent Service — Preview announcement for Azure AI Foundry. [3222d7]
  • Atlassian: Introducing Atlassian's Remote Model Context Protocol — Announcement of MCP integration for Atlassian products. [b055c1]
  • Windows Blog: Securing the Model Context Protocol: Building a safer agentic future on Windows — Early preview of MCP in Windows 11. [91430e]
  • Cloudflare: Early Adopters of the Model Context Protocol (MCP) & Open-Source Implementations — Overview of open-source MCP projects and industry adoption. [b9fb3b]

Summary

Model Context Protocol is rapidly becoming the backbone for connecting AI models with real-world data and tools in a secure, standardized way. Its adoption by leading cloud, productivity, and development platforms signals a shift toward more interoperable, Agentic AI systems across the industry. [8cac05] [b778b7] [081ad2]

Footnotes

AI is first a Trojan Horse

Business leaders will be eager at first, then shocked to the core.

LakeFS
Pretty much every business that wants a competitive edge is already knee deep trying to figure out how to best use AI.
There is something seductive about imaginging and experiencing even basic LLM capability. It feels super human, at first. It conjures science fiction. We were promised jetpacks, dammit.
ℹ️
We were promised jetpacks, dammit.
Add the Fear of Missing Out, but in business boardroom bingo. To not adapt is to foresee an existential threat.
Both of these motives are true, truer than even the most motivated of us can fathom.
Yet, adopting AI is inviting in a Trojan Horse. Maybe that metaphore is a bit belltristic. Perhaps it's more like Meet the Faulkners.
Imagine being excited for your birthday. Your friends blindfold you and say they have a surprise for you. When you arrive, it is a family reunion followed by a high school reunion. Your new emotional reality is that all the awkward business you were avoiding, or swept under the rug, will now be in full display.
For the most part, by mid-2025, we observe the following:
  • Almost everyone is using LLM's to generate content so they can send more email faster.
  • Many are using it to make presentations with more content and better looking without as much pixel pushing.
  • Students are using it to write papers, even fill out their homework.
  • Designers are using it to generate many concepts.
  • Software engineers are using it to generate boilerplate code.
  • Data analysts are using it to sort through a bunch of data, filter out the irrelevant, and come back with the relevant data that needs focus.
What do all the popular use cases of AI have in common?
Well, they are individuals motivated to better leverage their time and brainpower by delegating the more mundane parts of their job to AI.
As it should be. The 'Economy of Action' is one of the more predictable behavioral patterns of, not just humans, but all lifeforms.

From Rags to Riches

To implement RAG effectively, businesses will need to think about content, files, and databases in a way they have not. Because you get out what you put in.
The title "AI is First a Trojan Horse" suggests an exploration of the potential risks or hidden challenges associated with integrating AI into business operations. The starred block highlights that effectively implementing Retrieval-Augmented Generation (RAG) requires businesses to rethink their approach to handling content, files, and databases.

Key Considerations for RAG Implementation:

  1. Content Quality:
    • Ensure high-quality input data is used as the output quality of AI systems directly correlates with the input.
    • Regularly update and curate datasets to maintain relevance and accuracy.
  2. Data Organization:
    • Reorganize existing content, files, and databases for optimal retrieval by AI models.
    • Implement structured metadata tagging to enhance searchability and accessibility.
  3. Integration Strategy:
    • Develop a clear strategy on how RAG will integrate with current systems and workflows.
    • Assess compatibility with existing infrastructure and plan necessary upgrades or modifications.
  4. Security Concerns:
    • Consider the security implications of exposing sensitive data to AI models, especially when integrating external content sources.
    • Implement robust access controls and encryption measures.
  5. Scalability:
    • Plan for scalability in terms of both data volume and processing power as RAG systems evolve.
    • Ensure infrastructure can handle increased loads without performance degradation.
  6. Ethical Considerations:
    • Address ethical concerns related to bias, privacy, and transparency in AI-generated content.
    • Implement guidelines to ensure responsible use of AI technologies.
  7. Continuous Monitoring and Evaluation:
    • Establish mechanisms for ongoing monitoring of RAG system outputs to catch errors or biases early.
    • Regularly evaluate the effectiveness of RAG implementations against business goals.
By addressing these considerations, businesses can leverage the potential benefits of RAG while mitigating risks associated with AI integration.
We might have to consider Vector Databases
⚠️Deprecation Warning

The ```yaml toolingGallery syntax is deprecated. Please use :::tooling-gallery directive syntax instead.

How GitHub Changed Everything

https://youtu.be/cdcjw5etCnw?si=MQWGQE_MLIbDaUun
GitHub launched in 2009, and almost immediately became the primary code version control tool of new technology startups. From there, it's growth was meteoric and exponential. [4] [5]
ℹ️
"GitHub is like the air we breathe. It’s such a natural part of the way we work that sometimes we don’t even notice it. We cannot imagine living without GitHub.” Ryuzo Yamamoto // Software Engineer, Souzoh [3]
Slowly but surely, GitHub has become and will continue to get better at being the "everything" tool for collaborative software development. Smaller development teams already find no reason to adopt Workflow Management tools like Linear. GitHub sports "just-good-enough" Task Management, written about on their Docs here. They use Markdown inspired writing functionality, which extends basic text in many versatile ways. ^[1]
You can find little libraries that can speed you up considerably, such as Swapy Major collaborations on web security are now performed over GitHub, including Portable Open SSH GitHub has also become a public repository for computational work in science, with similar functionality as peer review.
The social network and community features of GitHub enable new Decision Heuristics in evaluating potential software solutions in the adoption process. [2]
GitHub has become the mainstay in the development and adoption of Open Source software. Every year, they publish the premier industry report, the Octoverse, which uses data on the GitHub platform.
Source: https://seekingalpha.com/article/4515335-gitlab-too-much-competition-little-differentiation-elevated-valuation-sell-for-better-entry-point
Taking a step back, the Enlightenment and Scientific Revolution developed a complex, emergent system of codifying and sharing knowledge: the endgame of academic work became the publication of an article to a peer reviewed journal. For those now doing their work by using programming languages to do computational work, they actually just use GitHub (which probably doesn't affect hiring and tenure at academic institutions yet, but...). Take for example this code base for analyzing DNA -- Dart-Eval: A Comprehensive DNA Language Model Evaluation Benchmark on Regulatory DNA. GitHub is now serving the same function as staying up-to-speed with evolving knowledge as going to a University library and pouring over recent journal publications. Therefore, watching trending code bases on GitHub now has dramatic effects on knowledge dissemination.
AI Explains GitHub
📝AI Explains
GitHub has fundamentally transformed the software development landscape by accelerating development cycles, fostering a culture of open-source collaboration, and democratizing access to tools and knowledge. Its impact is evident in both quantitative and qualitative metrics, as well as in how it has shaped the ethos of software engineering globally.

Key Ways GitHub Has Changed the World

1. Streamlining Software Development

GitHub provides a centralized platform for version control, collaboration, and project management, making software development faster and more efficient.
  • Version Control with Git Integration: By integrating Git, a distributed version control system, GitHub allows developers to track changes, roll back to previous versions, and collaborate seamlessly.
  • Continuous Integration and Deployment (CI/CD): GitHub Actions (introduced in 2019) has enabled automation of workflows, reducing manual overhead and speeding up the development cycle.
  • Pull Requests and Code Reviews: GitHub popularized features like pull requests, which streamline the process of proposing, discussing, and integrating code changes, ensuring higher code quality.

Traction Data:

  • As of 2023, 100 million developers use GitHub, up from 28 million in 2018, reflecting its widespread adoption.
  • Over 372 million repositories are hosted on GitHub, a testament to its role in global software development.

2. Creating an Open-Source Collaboration Ethos

GitHub has catalyzed the growth of the open-source movement by providing a platform where developers can share, contribute to, and reuse code.
  • Accessibility and Community: GitHub has made it easy for anyone to contribute to open-source projects, breaking down barriers to entry.
  • Popular Open-Source Projects: Projects like Linux, React, TensorFlow, and Kubernetes thrive on GitHub, powered by millions of contributors worldwide.
  • Knowledge Sharing: GitHub’s README files, wikis, and issue trackers have made it a hub for documenting and sharing software knowledge.

Traction Data:

  • GitHub's "State of the Octoverse 2022" report highlighted that 94 million developers contributed to open-source projects that year.
  • In 2022, 3.5 billion contributions (commits, pull requests, etc.) were made, showing the platform’s scale as a collaboration tool.

3. Accelerating Innovation

GitHub has reduced the time and effort required to bring new technologies to market by enabling developers to build on existing work rather than starting from scratch.
  • Code Reusability: Developers can fork repositories, clone projects, and use libraries or frameworks hosted on GitHub, significantly reducing development time.
  • Rapid Prototyping: GitHub makes it easy to experiment and iterate on ideas, promoting agile methodologies.
  • Open-Source Foundations for Startups: Many startups and tech giants, including Netflix, Airbnb, and Uber, have leveraged open-source technologies hosted on GitHub to build their core products.

Traction Data:

  • GitHub reports that 90% of companies use open-source software in some capacity, much of which is hosted on GitHub.
  • Over 20,000 open-source projects are actively maintained by Fortune 500 companies on GitHub.

4. Democratizing Software Development

GitHub has lowered the barrier to entry for learning to code and participating in software development by providing free tools, educational resources, and a vibrant community.
  • Free Access to Tools: GitHub offers free repositories, enabling anyone to start coding without financial constraints.
  • GitHub Education: Programs like GitHub Student Developer Pack provide free access to premium tools and resources for students.
  • Global Participation: Developers from around the world, including those in traditionally underrepresented regions, can now contribute to global projects.

Traction Data:

  • GitHub is used in over 200 countries, with significant growth in regions like Africa and Asia.
  • 40% of GitHub’s users are now outside the US, reflecting its global reach.

5. Fostering Corporate and Community Collaboration

GitHub bridges the gap between corporate development and the open-source community, creating a hybrid model of innovation.
  • Corporate Open Source: Companies like Microsoft, Google, and Facebook actively contribute to open-source projects on GitHub, benefiting from community contributions while advancing their own goals.
  • Cross-Disciplinary Collaboration: GitHub’s platform is now used beyond software, including in fields like data science, education, and even creative industries.

Traction Data:

  • Microsoft’s acquisition of GitHub in 2018 for $7.5 billion underscored its strategic importance.
  • Corporate repositories on GitHub have grown rapidly, with over 10,000 organizations using GitHub Enterprise.

Examples of GitHub’s Impact

  1. React (Meta): React, hosted on GitHub, has over 210,000 stars and is used by millions of developers to build web applications.
  2. Linux Kernel: The Linux kernel repository on GitHub has attracted contributions from thousands of developers, making it one of the most significant open-source projects in history.
  3. COVID-19 Research: During the pandemic, GitHub became a hub for sharing data, models, and software to combat COVID-19, accelerating global research efforts.

Conclusion

GitHub has redefined software development by fostering a collaborative, open, and innovative ecosystem. With its vast user base, growing repository count, and integration into every facet of the tech industry, GitHub has become the backbone of modern software engineering. Its role in accelerating development, enabling open-source collaboration, and democratizing access ensures its continued influence on the global tech landscape.

Footnotes

[2] 2024, Nov 04.

How to Pick the Best AI Open-source Projects for Production Use
Yifan Beyond the Hype, YouTube.

[4] Timeline of GitHub Wikipedia. Accessed 2025, Feb 12.

[5] GitHub Tutorial. PSL Models. Accessed 2025, Feb 16.