AI becomes your personal mentor and builds custom dashboards

Most of us still type prompts into AI. Rob Williams, Chief AI Officer at Soft House Advisory, doesn’t. He speaks to every AI agent he uses, and it’s transforming his output.

In this episode of This New Way, Rob shares how going voice-first eliminates prompt engineering, why AI lowers the barrier to building custom tools, and the future of dynamic software that creates itself in real time.

→ Listen to Rob’s episode

→ Subscribe to our YouTube Channel

Get the step-by-step tutorials and AI tools mentioned in this episode. 👇

Tutorial 1: Build a custom app in minutes with Claude Code

Busy execs don’t have time for coding, but Rob shows how Claude Code acts like a full development team on your desktop.

1. Install Claude Code.
   Download the command-line tool from Anthropic’s site. It connects to Claude and runs locally on your machine.
   

2. Drop in your files.
   Place a spreadsheet or data file into a folder. This becomes the source for your custom app.
   

3. Talk to Claude.
   Rob uses voice, not typing, to describe the app he wants. Include context: what the app should display, who will use it, and any features you need.
   

4. Let the agent work.
   Claude Code spins up mini AIs, installs dependencies, and writes code autonomously. Rob watched it build an interactive dashboard in minutes.
   

5. Iterate by conversation.
   If something breaks or you want changes, just paste the error or speak your update. Claude adjusts instantly.

💡 Pro tip: Keep your prompts natural and conversational. Rob says voice removes the need for “perfect” prompting.

 Watch the demo here

Tutorial 2: Automate discovery to proposal in one day

Rob cut a five-day consulting workflow down to a single day using AI and purposeful context overload.

1. Record discovery calls intentionally.
   Narrate screens and actions out loud. Avoid vague references. Describe everything in detail for the AI.
   

2. Export transcripts.
   Use a high-quality transcription tool like Fellow. Rob feeds the raw transcripts into his AI workflow.
   

3. Load into a deep research prompt.
   His prompt includes structured markdown for clarity. The AI digests hours of calls into process maps, proposals, or reports.
   

4. Review and refine.
   Spend 30–60 minutes marking up the AI output, then feed corrections back in for a final pass.
   

5. Auto-generate deliverables.
   Rob’s workflow spins out a proposal and a PowerPoint deck with slide prompts and images in under an hour.

💡 Pro tip: Don’t just layer AI on old workflows. Redesign them around AI’s strengths: full context upfront and bulk processing.

See Rob’s workflow here

Tools mentioned in this episode:

• Claude Code (Anthropic) – AI-powered coding agent that runs locally
  

• Limitless AI Pin – Wearable AI that logs and analyzes conversations
  

• Fellow – The most secure AI meeting assistant and notetaker

Workflows like these show just how far AI can take us. Don’t miss out: subscribe to This New Way on YouTube and stay ahead.

👉 Subscribe to This New Way: YouTube Channel

 📩 Was this email forwarded to you? Subscribe here

Until next time,

Aydin Mirzaee
CEO at Fellow & Host of This New Way