Claude As An AI Agent
While ChatGPT tends to get the majority of the AI news coverage, I use Anthropic’s Claude almost exclusively. This is mostly because Anthropic seems to have more thoughtful policies around privacy and safeguarding our collective future. I recently stumbled across something that has completely elevated how I work with Claude – the ability to edit system preferences that act like a system prompt for every conversation. I read something Tiago Forte wrote about this and looked into it.
You can think of it like programming the LLM with as much relevant information and context as you want. Much in the same way you might in a single chat with an LLM, where you tell it about yourself or your business or what kind of project you are working on - but with this system preferences method, Claude already has that context for every conversation. My experience has been that this dramatically improves the quality of the output while reducing the amount of interactions (or tokens) needed. The screenshot below is what you’ll want to look for under the Settings page.
I would hazard a guess that a lot of construction businesses might be missing out on this kind of value. The software landscape is shifting rapidly and there are lots of tools at our disposal that are fundamentally different than the software we are used to. While tools like BuilderTrend or Quickbooks are helpful, they are not the only type of tech we could be using. Picture this: your entire team running on a Claude Enterprise account, not just sharing projects but operating from the same foundational knowledge base. Fill your system prompt with protocols for writing contracts or drafting change order documents. This literally means whole text examples of what you want these documents to look like that are pasted into your system prompt. Then you can add in keywords so when a user types “drafting a contract” it will ask them relevant context questions and then draft a document in the exact format and style needed. Follow that with your team’s unique voice in editing, re-writing, and crafting this document to its authentic finish. While many will use these tools to generate sloppy text content with less effort, I believe we can program them to elevate our teams, allowing us to spend more time bringing humanity and trust to our work.
Testing Conversational Lead Qualification
Turning Chat Into Intelligence
I've been experimenting with something that I think could be really valuable in sales and marketing for contractors. I took existing client website chat assistants and started editing them to focus specifically on lead qualification rather than just answering questions.
Here's the approach: fewer knowledge sources, more structured conversation flows. When users chat with these systems, they're essentially filling out a conversational contact form. All of this data pushes to Google Sheets or your CRM, and then AI automatically structures and scores each potential lead.

For small high-performance contractors where every lead matters, this creates a systematic approach to understanding client readiness and project fit before you invest time in meetings or site visits.
Claude’s New Artifacts Builder
Experimenting with Team Dashboards
This month, Claude launched their new Artifacts builder, and it's opened up some interesting project opportunities. The Artifacts feature allows users to create interactive web applications, dashboards, and tools directly within conversations. I have been experimenting with building out team dashboards that can present all of an organization’s AI agents in one place. There are a lot of businesses right now that are barely using AI. Perhaps they have a few people internally that use it to proofread or do research. But I think the real leverage opportunity is when a company can commit to adopting it across their team in a staged and strategic way.
Many of these tools or assistants are quite simple to engage with, and I think dashboards for tools can help integrate these into a company with as little friction as possible. While I'm still exploring the possibilities, the rapid prototyping capability feels promising for creating team-specific tools.
Heading to ChippCon
AI in Construction

I'll be speaking at ChippCon in Fargo, ND on July 21-22 about AI applications in construction. I am excited to connect with the community of people using Chipp to create value in their organizations – from veterans to farmers to lawyers to contractors to teachers to coaches.
What I love about this community is the focus on practical implementation and the direct impact these tools can have. Real people solving real problems with the help of AI - it’s a beautiful thing, and one that has greatly expanded my own horizon for what work looks and feels like.
If you're attending, I'd love to connect and hear how you're approaching AI in your own work.
Drop me a line at [email protected]
If you’re interested in building your own AI tools and supporting this newsletter along the way, you can check out this link: https://chipp.ai/?via=murray
Chipp.ai is an amazing team with excellent support - focused on making these tools accessible to all.
Murray

