Pro-Tips to Master AI

How Nate B. Jones Advises Our Product Roadmap at 4 AM Every Morning (While We're Both Asleep)

I'm writing my first Storytell "Action" take have Nate B. Jones' latest AI thinking and apply it to Storytell's product roadmap This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit drodio.substack.com [https://drodio.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Writing In Your Voice, On Demand: Using Storytell's New "Persona Skill" Feature

I have a terrible habit of having my best thoughts in the worst places. Driving to pick up the kids. Mid-workout. Halfway through making coffee. The thought is vivid and clear, and then — almost immediately — it starts to fade at the edges. I’ve lost more genuinely interesting ideas to the friction of “I’ll write that up later” than I care to admit. So when I realized I could literally speak a blog post into existence — record a few minutes of rambling audio, hand it to Storytell, and use Storytell’s new Persona Skills feature [https://go.storytell.ai/learn-skills] to get back a fully structured Substack post that sounds exactly like me, I became a little obsessed. This post is the companion guide to the video above I made to walk you through this process. If you watched want a step-by-step written reference to follow along, you’re in the right place. Let’s go. Why This Works The core idea is simple: your thoughts are more fluid when you speak than when you type. So instead of staring at a blank page, you talk — to others in meetings, to yourself, to nobody in particular. Then you let Storytell do two things: * Learn your voice — your patterns, frameworks, tone, the way you build an argument * Turn your video, audio., transcripts or notes into written output — structured, in your style, ready to edit and publish The Skills feature is the mechanism that ties this together. Think of it as a pre-loaded, reusable set of instructions that knows exactly how you work, or speak, and can apply that understanding to your raw unstructured material. Pro-tip: Having Storytell write in your voice is just one small example of teaching Storytell a skill that it can then re-use at scale. You might also have Storytell learn how to speak like your CEO, your sales engineer, your data analyst. Skills aren’t limited to speaking “in the voice of,” either. You might give Storytell specialized knowledge about how your company likes to respond to customers and invoke that skill when doing churn analysis or customer success work. Phase 1: Build Your “Voice Foundation” This is the one-time setup. Do this right, and everything you have Storytell write in your voice becomes tuned to the way you write. Using the “Skill Creator” The Skill Creator is an easy way to have Storytell build a new skill, like “writing in my voice.” Invoke it by typing the @ symbol and then start typing “Skill Creator” — you’ll see it in a pop up menu. Then type a prompt like the one below to create a “Voice of” persona-based skill that will output a detailed brand voice and messaging guideline — things like your communication style, your values, how you frame problems, your recurring themes. Storytell will craft a “Voice of” Skill based on your instructions, which you can then save for future use. You’ll get back output that feels almost uncomfortably accurate. Mine had things like “vulnerable over polished” and “problem-focused over solution-obsessed.” That’s the goal — a document that captures you, not a generic professional persona. Save the Skill to Your Project and Put It To Work Once Storytell generates your “Voice of” persona skill, you can start to use it like I did in the video above to create this Substack post [https://drodio.substack.com/p/from-zoom-call-to-shipped-code-my]. Here’s a screenshot of the output from Storytell that formed the basis of that post: Storytell offers many pre-built skills, and you can add as many of your own as you’d like. But Wait, There’s More Storytell also just launched a Prompt Library [https://go.storytell.ai/learn-prompt-library] that lets you save and re-use your favorite prompts. If you switch to “advanced user mode [https://go.storytell.ai/help-user-mode],” you can even put field variables into your saved prompts. This lets you create a saved prompt that uses one of your skills, like this: The next time you want to run that prompt, you just select it from the Prompt Library and it’s ready to go: The Real Math Here Before I built this workflow, the barrier between “having a thought” and “publishing a post” was high enough that most thoughts never made it. The friction — the blank page, the organizing, the writing, the editing — added up to hours I often didn’t have. Now it looks like this: * 5 minutes of audio → captured thought * 1 click → structured draft in my voice * Edit to taste → polished post, ready to publish I’ve built the project once. The “Voice of” skill is ready. The labels are set. Every time I want to publish now, I start with a system that knows my voice. The whole system compounds. That’s the part I couldn’t have predicted when I started experimenting with this: how much easier the tenth post becomes compared to the first. The project just keeps getting richer, more context-aware, more me. I hope you can experience the same result! This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit drodio.substack.com [https://drodio.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

From Zoom call to shipped code: My workflow for PRDs

Most brilliant engineering conversations evaporate into Slack history. I spent 20 years watching velocity die in that gap. Here’s my exact workflow for turning a 15-minute call with an engineer into shipped code using Storytell + Cursor to bridge human conversation and AI execution. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit drodio.substack.com [https://drodio.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Turning a 12 minute conversation between CEO + Engineer into investor updates, technical and help documentation

As CEO, I need to understand what my engineering team is building—but I also can’t spend hours diving into each technical improvement. When my team shipped powerful new functionality around multi-fleet agents and tools at Storytell, I needed a deep understanding fast. Here’s exactly how I used our own product to turn a quick phone call and some engineering tickets into a comprehensive understanding I could use for investor updates, technical capabilities documentation and user help docs. CSVs and transcripts don’t usually play nice together. But with Storytell.ai, you can combine unstructured data (like audio transcripts) with structured data (like Linear tickets) to create polished, production-ready documentation. This isn’t just about saving time, it’s about scaling knowledge across your organization and specifically across teams. In this case, I used: * A conversation transcript about new functionality we’d shipped (I used a 12-minute phone call with my engineer, Alex) * Structured data related to that topic (I exported recent engineering tickets from Linear) * Existing documentation we wanted to update (we have a tech stack page and help documentation) What I did, Step-by-Step I spent 12 minutes on the phone with Alex, having him walk me through how our new tools, agents, and skills work. Pro tip: Storytell has the ability to directly record audio right in the project [https://help.storytell.ai/articles/3215753-record-and-upload-audio-directly-to-storytell], which can be very useful to capture conversations like this. The result: A raw, conversational transcript that contains the knowledge I needed. I exported from Linear all the engineering tickets and issues that had activity in the last week. This gave me detailed technical context on what was actually being built and shipped. Now I had two distinct data sources: * Unstructured: The conversation transcript * Structured: CSV data from Linear I already had this tech stack page [https://tech.storytell.ai/] that needed updating. Here’s what I asked Storytell to do: * Consume the existing URL (the old tech stack page) * Using the resources above (the transcript + Linear tickets), write an executive summary about how all the technology works * Write up detailed documentation * Rewrite the entire page I wrote this prompt and Storytell got to work—reasoning, doing web searches, doing knowledge base searches on my internal data. Phenomenal. I also had this existing help documentation [https://help.storytell.ai/articles/4111122-multi-agent-architecture-and-specialized-agents] that needed refreshing. I asked Storytell to: * Consume the existing webpage (I copied it as markdown) * Run the prompt to update our help pages I used a diff checker to see the differences between the old version and the new version. The results were clear: There was a whole new section on the Storytell tools ecosystem that came directly from my 12-minute conversation with Alex. Storytell pulled insights from the conversation, cross-referenced with the Linear tickets, and integrated everything into cohesive documentation. I now had three updated assets: * Human-readable documentation with information about specialized agents, tools, skills, etc. * LLM-optimized version — this is what we send to large language models so they can learn how Storytell works (meant to be read by machines, not humans) * Help page updates — specific user-facing documentation about how to use these features I sent the help page updates to our customer success team to publish. This entire workflow—from phone call to published documentation—took less than an hour and was based on time I’d already spent on a call anyway. The magic is in combining different data types that traditionally don’t work together: * Conversational, unstructured knowledge (transcript) * Structured, technical detail (Linear CSV exports) * Existing documentation (web pages) That’s how you make use of time you’re already spending to scale knowledge across departments. Hope it’s useful! This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit drodio.substack.com [https://drodio.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

How To Create Pre-Filled 1-Click Prompts in Storytell

We just shipped our Prompt Launchpad [https://web.storytell.ai/blog/introducing-storytell-prompt-launchpad] with 1,659 curated, pre-filled prompts that users can easily submit with just one click. You can find them all right here [https://go.Storytell.ai/launchpad]. If you’ve visited Storytell.ai [https://storytell.ai/] recently, you’ve seen them on the homepage—featured prompts like “Churn Risk Analysis [https://storytell.ai/?use-case=Work%2FCustomer%2520Success%2FchurnRiskAnalysis&side=scope-picker]” or “Find Anyone’s Work Email Address [https://storytell.ai/?chat=Find+the+email+address+for+%23%7Btext%3APerson%27s+name%7D+.+Then+go+do+deep+research+on+them+using+all+available+public+data+sources+and+tell+me+what+their+current+top+priorities+are.]” where you can literally upload one file, hit go, and get high-quality output immediately. Here’s how you can also create these pre-filled 1-click prompts to share with others. Step 1: Enable Advanced Mode First, you need to have Advanced Mode [https://go.storytell.ai/help-user-mode] turned on. This isn’t enabled by default, but it’s available to all users: * Go to your user settings in Storytell * Switch on “Advanced Mode” Once you do this, you’ll see extra icons and options that let you create template variables, share draft prompts, and build the sophisticated prompts I’m about to show you. Step 2: Write Your 1-Click Prompt Using Template Variables To create a pre-filled prompt, you just use the # hashtag when writing a prompt, like this: You’ll be able to add one of four types of variables, and to choose whether they should be optional or required: Here’s what it looks like after you add the variable: Once you’ve put the fields in that you want your user to fill in, just choose Share this draft prompt from the share menu at the top right of the prompt box. That’s it! You can now paste this shareable URL anywhere (into an email, a Slack or Teams message, etc) or you can bookmark it for later use. Here’s the 1-click prompt I just created in the screenshots above for you to try: https://storytell.ai/?chat=Analyze+our+customer+accounts+using+%23%7Bfile%3ASales+Data%7D [https://storytell.ai/?chat=Analyze+our+customer+accounts+using+%23%7Bfile%3ASales+Data%7D] Going Way Deeper: Giving Storytell Specialized Knowledge In the video above, I do a deeper dive showing how I did way more than just create a 1-click prompt: I gave Storytell specialized knowledge to use when answering this “Help Me Resolve Conflict [https://storytell.ai/?use-case=Work%2FExecutive%2FChief%2520Executive%2520Officer%2FhelpMeResolveConflict&side=scope-picker]” prompt. If you want to learn how to do that, too, read on! Helping Humans Navigate Conflict (with one click) Most of our featured prompts are work-focused, but I wanted to tackle something different in the last spot I had available for the top six featured prompts on the Storytell.ai home page: Helping humans with conflict resolution. Why? Because conflict is fundamentally unstructured data—messy emotions, unclear needs, communication breakdowns. This is an area where Storytell excels. Using the process outlined below, I was able to build a prompt that uses the Clean Communication Framework developed by Erika Anderson combined with Stan Tatkin’s PACT framework without fine-tuning or training a custom LLM for this expertise. Step 1: Gathering Source Materials As I said above, everything from here on is optional. You don’t need to do any of this to make a pre-filled 1-click prompt. I first needed to assemble the raw “knowledge” Storytell would need to expertly answer the prompt query. This step is optional — out of the box Storytell is trained on the world’s knowledge, so if you’re not working with a bespoke framework like I was, you can skip this work. I created a project in Storytell [https://help.storytell.ai/articles/4759639-projects-overview] called “Clean Communication” and uploaded multiple assets: * The Clean Communication Framework PDF [https://go.Storytell.ai/clean-communication] (we literally use this at Storytell internally when we need to work through challenging conversations) * Active listening frameworks * Various other related materials Think of this step as loading the knowledge base that Storytell will synthesize its prompt expertise from. Step 2: Have Storytell Synthesize the Training Document Here’s where it gets powerful. Instead of manually compiling all the knowledge I’d need to have Storytell respond expertly, I just asked Storytell to do the synthesis work: My prompt to Storytell: Synthesize all the data in this project and create a comprehensive framework document for an LLM to consume about clean communication. Storytell went to work. You can actually watch it process in the video above —searching through the knowledge base, pulling relevant sections, organizing the material. What it created was a detailed training guide that included: * The different levels of communication (logic, emotions, vulnerability) * Tables of feelings associated with met and unmet needs * Tools like “observations vs. judgments” (e.g., “You’re always late to meetings” vs. “I’ve noticed you arrive 10 minutes after the hour for our last five meetings”) * Guidance on translating emotions into unmet needs This is the key insight: humans get stuck at level two (emotions). “I’m angry you didn’t show up to the meeting we scheduled.” But there’s no resolution at the level of emotions. The work happens at level three—vulnerability and unmet needs. “I felt angry and disappointed. I have an unmet need for consideration and partnership. I need to know that our time together matters to both of us.” The training document I had Storytell create teaches the LLM how to guide humans down to level three. Step 3: Store the Training Document in GitHub Gist Once Storytell generated the comprehensive training document (about 1,300 lines of markdown), I needed to make it accessible to the prompt. Here’s my process: * Download the document as markdown (the native language of LLMs) * Create a new GitHub Gist * Paste the markdown content into the gist * Grab the raw URL from the gist I went one step further: I created a subdomain at https://framework.cleancommunication.com [https://framework.cleancommunication.com] that redirects to the raw gist URL. This makes the prompt more readable and maintainable. Step 4: Build the Predefined Prompt Now comes the actual prompt construction — like above, but with a twist. In Advanced Mode, you can use template variables to make your prompts dynamic: * Type your prompt in the prompt bar * Use # to insert template variables (text input, file upload, dropdown, or checkbox) * Mark variables as required or optional * But this time, I included the instruction to consume my training document: Go learn about the Clean Communication Framework by consuming it at https://framework.cleancommunication.com Try Using the Prompt! You can try this Help Me Resolve Conflict prompt by going to the home page of Storytell.ai [https://Storytell.ai] — you’ll see it in the Prompt Launcher’s “Featured Prompts” section. You can also try it by clicking right here on this shareable URL [https://storytell.ai/?use-case=Work%2FExecutive%2FChief%2520Executive%2520Officer%2FhelpMeResolveConflict&side=scope-picker]. Why Do This? Traditional approaches would require fine-tuning a model, maintaining custom infrastructure, or using limited prompt space to create this level of specialized expertise for an anser. This approach is different: * The LLM expertise training document lives externally (GitHub gist) * Any foundational model can consume it on-demand * Updates to the framework just require updating the gist * The prompt stays clean and focused You can use this for any domain where you need an LLM to quickly ramp up on a specific skill set. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit drodio.substack.com [https://drodio.substack.com?utm_medium=podcast&utm_campaign=CTA_1]