Archive
44 posts on AI, private equity, and building.
mdEdit - Repo
mdEdit is a local-first markdown editor for macOS with a twist: the documents you have open are directly readable and editable by AI over a local MCP connection — no cloud in between, and every AI edit happens right on your screen. It pairs a clean, iA Writer-inspired writing surface with fully native macOS chrome, live preview, and stale-write protection so the AI never clobbers your typing. You can download the notarized app free, or grab the source on GitHub (MIT licensed) and build your own.
Half of Lower-Middle-Market PE Firms Have No Visible AI Strategy
No AI roles open. No AI or digital staff. No AI value creation. No recognition of AI at all. That this described half of the firms in our research was surprising.
Your Data Is Your AI Advantage, Keep it
Satya Nadella's latest essay is getting a lot of buzz. For the middle market, it oversells the problem to pitch the fix.
Weekend Listens: Three podcasts for your weekend PE listening pleasure
#Private Equity is Coming For Your 401(k) — Private Equity Funcast, Jul 1 Ted Seides and PitchBook's Nizar Tarhuni on the push to open retail 401(k)s to private assets. The upside, the real risks, and why "democratizing" private equity is harder than the headlines suggest. Blunt and data-backed. #What LPs Are Looking for in Defence-Related PE Strategies — Private Equity Spotlight (PEI), Jul 2 PEI's Adam Leigh on the money rushing into defence and security. How LPs are underwriting the theme, where the real opportunity sits, and where it is just a hot label. How Opendoor, Datadog, and PwC Are Using AI in Finance — Run the Numbers, Jul 2 #Finance leaders on what AI actually does in the numbers function today. No hype. Practical: what to build, what to buy, and why waiting keeps getting more expensive.
AI Impact from the Ops seat
The AI headlines are about models, operators talk impact. The impact of AI extends beyond just numbers, it raises a fundamental question, how many people does it take to run a company.
Every CIM Says AI Now
On a recent Private Equity Funcast, ParkerGale partners Devin Mathews and Jim Milbery spent an episode on a problem every middle-market buyer now shares. Every CIM that crosses the desk claims AI.
$33 in incremental AI Spend
That's what it costs to run a full AI B2B marketing engine for a client right now. Content for LinkedIn. Email. Blog. SEO. Targeting and Pipeline prep for the sales team.
AI is finally giving us the intranet we were promised
In November 1998, two equity analysts at Merrill Lynch described the future of work in a single sentence. Christopher Shilakes and Julie Tylman were not futurists or vendors with something to sell.
AI Is Rewriting the PE Playbook
For years, when private equity talked about technology, it talked about cost. Squeeze the back office. Consolidate the stack. Expand the margin. Listen to how PE leaders talk about AI now and you hear something different. Something more existential.
On-Demand Apps from LI
Traditional software is built for thousands of people. That means it fits no one perfectly. AI enables software with an audience as small as one.
Rise of the AI Operating Partner
Here is a recent set of Posts on my LinkedIn.
Technical Journey
Over my career I've built a strong foundation in computer science, systems design, and the architecture of complex business platforms.
The Lot Line
Try out thelotline.com. We are looking at how to build a site to make real estate browsing a leaned back on the couch fun experience with a simple game mechanic.
What is the ICP tool in MeetBri?
This is the story of how a test became a product. While I was building the original AI interviewer, I needed some long form interviews to use as models, they had to be real, generally business oriented and available at volume.
How did MeetBri evolve?
The concept for the current iteration of MeetBri came about when I got asked by Anthropic to participate in a research survey. It was an in-depth survey that was administered by Claude. Being curious, I thought, "Let me build that so I can see how they did it."
What Is a Fractional Chief AI Officer?
Let's start with what a fractional Chief AI Officer is not. It's not a technical role. It's not a glorified IT consultant. And it's not an administrative layer you bolt on to check a box. Those are the easy misconceptions to clear up. The harder one is explaining what it actually is — because it lives in a space most organizations haven't had to think about before.
Meet the new MeetBri
Here is a posse from LinkedIn: https://www.linkedin.com/pulse/mood-changed-feb26-meetbri-gznje/
Anatomy of a Context Seed
A context seed file is written for AI, not humans. It provides everything an AI needs to understand a topic, person, product, or process — and know where to find more when needed. Whether the seed describes a person, a software product, a cooking technique, or a company, the structure remains the same: three sections, each with a distinct purpose. Read More
Why do some AI chats feel smarter than others
LinkedIn: https://www.linkedin.com/posts/scottrhall_why-do-some-ai-chats-feel-smarter-than-others-activity-7422672850714632192-rKki
Instant Context
Introducing InstantContext.ai a method for putting context directly in a link.
Ai B2B Purchasing Decision Data
Ai is the new top of funnel for B2B sales and here is the data.
Your Opening Message Determines Everything With AI
Article on LinkedIn link
Panels for Strategic Insight
Credible strategic support and blind-spot detection remain serious challenges for the current generation of AI systems. AI models, trained for helpfulness and likeability, often drift toward what can only be described as “performative agreeableness.” It’s why people push them to “get real with me”—they sense the flattening of truth beneath the friendliness.
What is a Manifest and well-known
A manifest is the configuration file that tells clients like Claude what an MCP server can do. It declares the tools, resources, and prompts the server offers, plus some basic metadata like its name, version, and description. In short, it’s how an MCP server introduces itself and explains how to talk to it. Until now, you wouldn’t have seen one because clients had to actually connect to a server before discovering its capabilities. That’s changing with the November 2025 spec update, which introduces .well-known URL discovery. This is a big deal. It means MCP servers will be able to publish their manifest in a predictable public location—like how websites use sitemap.xml—so tools, registries, and even search engines for MCP can index what’s out there without needing a live connection. That shift makes the manifest more than just internal config. It becomes the public face of an MCP server—the thing that lets the ecosystem browse, catalog, and connect everything together. If you’ve never heard of manifests before, that’s normal. They’ve been working quietly in the background. But the new .well-known requirement is about to make them front and center in how AI systems discover and connect across the MCP network.
How We Test
LLMs don’t just need to answer questions—they need to choose the right tools when connected to an MCP server. Our testing system is designed to measure exactly that, giving us a clear picture of how well models navigate real-world tool use.
Case Study: How Manifest Alignment Boosted Tool Use Performance from 70% to 83% for Cooking MCP
Large language models don’t just rely on their training—they also depend on how tools are described and exposed through manifests. Even small wording choices in a manifest can make the difference between a model using the right tool confidently or skipping it altogether.
Deterministic vs. Interpretive MCP Servers: Speaking the Same Language
MCP isn’t just about exposing tools. It’s about creating an efficient and effective understanding of a server in the eye of the ClientAI. This communication is guided by the manifest and, by extension, the initialization and tool calls. For simplicity, we’ll use “manifest” to represent that entire package. In effect, the manifest becomes the dictionary of the language between server and AI.
Case Study: How a Single Manifest Change Boosted LLM Accuracy from 33% to 83% in a Key Use Category
## The Challenge We were testing a simple MCP todo server with three tools: createTodo, listTodos, and a completion tool. The original completion tool was named completeTodo with the bare description: “Mark a todo item as completed.”
Potential and Pitfalls: Security Lessons from Real-World MCP Tools
The Model Context Protocol (MCP) ecosystem has exploded. In just months, developers have published hundreds of servers and thousands of tools. It’s a remarkable show of creativity: everything from lightweight utilities that return the time of day to heavyweight servers that can administer databases, manage cloud infrastructure, or even control a computer.
MCP Tool Distribution
We analyzed 1,945 tools across 238 servers in the MCP Registry to understand how tools are distributed.
MCP Tool Research
This week, we’re launching a new phase of research at Lyr3: digging into the tools that power the MCP ecosystem. So far: - We’ve collected data on > 420 MCP servers. - Across ~300 of them, we’ve identified 1,400+ distinct tools, tool descriptions and inputs.
Link: You are Doing MCP Wrong
Good read of what MCP is and is not from Jim Clark at Docker. link.
MCP Registry
Lyr3.com is happy to present its MCP Registry as a beta.
NEWS: MCP Continues Reach Growth - Sept 9
MCP continues to expand its reach across enterprise and developer ecosystems, with new use cases, platform adoption, and benchmarks revealing both promise and challenges.
Do We Need Embeddable UIs in MCP?
Chat boxes have become the face of AI. We type, the AI replies, and when tools are connected through MCP (Model Context Protocol), the outputs are poured back into text. Clean, minimal plumbing. But MCP-Universe, a benchmarking project, shows just how brittle that model is. Only about one-third of complex interactions succeed when the AI has to juggle multi-step orchestration entirely through chat. That raises a question: are we overloading the AI, and forcing the chat box to do too much?
MCPs: The Choose-Your-Own-Adventure Games of Enterprise AI
When people hear “Model Context Protocol (MCP),” it sounds abstract and technical—like plumbing for AI systems. But if you strip it down, MCPs are really just goal-oriented choose-your-own-adventure games written for AI.
MCP Universe and AI limitations
On LinkedIn I wrote a short piece on how the limitations of AI impact the design and architecture of AIUX, the process of designing interfaces for AI. We will need to pay close attention to how this develops.
Making Excel AI Friendly
MDN (Markdown Notation) project defining a simple, AI-readable spec for representing tabular and contextual data for AI systems
Strategic decision feedback
An application that uses AI to provide critical feedback in go-to-market decisions. Outcome: Startup concept, effective technology but could not resolve on GTM. Retired.
Don't Build AI Tools. Build With AI.
The real opportunity in the AI era isn't building AI tools, but creating AI-native companies that fundamentally rethink how value gets created and delivered.
AI in Regulated Contexts
Exploring how compliance, audit, and defensibility change when AI becomes part of the workflow. Outcome: Developed AI tool to write compliance training scripts and documents. Client Project
The Future of Proofreading: AI Consensus Systems
How combining multiple AI proofreaders with consensus-based decision making can achieve better results than either humans or AI alone. Client Project
The Three Layers of AI Data
Understanding the three critical data layers that form the backbone of reliable AI decision-making: Training Data, Proprietary Data, and External Data.
Streamlining External Data with Preprocessed Repositories
How preprocessed data repositories can streamline external data integration for corporate AI systems, reducing complexity and accelerating time-to-value.