The Rise of Micro SaaS: Build Your Own AI Tools with No-Code!

Full video transcript

[Music] How’s it going everyone, Darby Rollins here, founder CEO of Gen AI University, joining today with Mike Gioia from Pickaxe. Mike will be joining us here momentarily inside this call — going to be doing some overviews and demonstrations of the tool. What you can expect on today’s call I’ll cover here in just a minute, but I’m going to bring Mike up to the stage and then we will get ready to rock and roll. Mike, great to see you, how are you doing today? Hey, good to see you, good to see you, hope you are doing well. Good — well, I like the hat, it’s a good look. It’s finally a little cooler temperature here in Austin, Texas, and so I’m taking advantage of my fall

beanie season, as short as it might last this year. Yeah, we get a very brief beanie season here in LA. All right guys, good to see everybody here joining us live. For those of you who don’t know, Mike is the co-founder of a tool called Pickaxe, which we’ve used ourselves and a number of others inside of our community have been users of over the past number of months, and wanted to bring Mike on today as a followup to our sessions through the Scale with AI Summit. Mike put together a few demonstrations on how to leverage Pickaxe as a tool for empowering knowledge entrepreneurs and turning your ideas into impact and monetizing those ideas, and then we had a special knowledge

entrepreneur panel with Travis and a number of other speakers on the topic of knowledge entrepreneurship, and so I asked Mike after the summit if he’d be open to coming back and sharing a little bit about the tool, having some open floor questions and answers with those inside of our community, because I know I see a few members on right now that actively use Project Pickaxe. If you go over to projectpickaxe.com and use the code from today’s call, you get a discount for your first three months when you do subscribe for Pickaxe. So we’re going to be diving into a hands-on workshop throughout the course of today’s call, showing you how to build these tools. And again,

if it’s not working for you after that time, no problem, but I think you’ll see once you get in and get to building some of these tools how powerful these can be for just getting those ideas out and turning them into ways you can make an impact. Mike, you’re obviously the expert here on the tool. I would love to hear where you’re at coming off the summit and if you got any good feedback from those knowledge entrepreneur panels, but if you have any ideas for maybe some examples of tools we can create on today’s call. Some of the preface coming in was that we’ll show people a little bit over the shoulder of Pickaxe and

we’d love to hear just at a high level what Pickaxe is from your point of view, who it’s for, and then we’ll field some questions as well throughout the call. Totally — hello everyone, good afternoon, or whatever the time is for you. We had a lot of fun at the knowledge entrepreneur panel a couple weeks ago; we had Travis, Rosin, and Tyrese also there talking, and it wasn’t all about Pickaxe, but you probably saw Pickaxe there and maybe even checked it out. I’ll give a quick explanation of what it is for people that might not know. Pickaxe is a no-code platform and it lets you do

everything you might need to do to build an AI app and launch it like a GPT-type app. So you can create AI tools in the form of chatbots or smart forms, you can train them with documents, you can connect them to other pieces of software, you can teach them to talk like you, then you can deploy and launch them, put them into your website as an embed, spin them up as a standalone website or web app, and then you can start to run it as a business or a lead-gen tool. So you can have people register, limit the amount of times they can use it, have people pay for more usage — that’s the full scope

of the product, and you bounce between those things: you create, you launch, you start selling it or using it as lead gen, you see what people are saying, you go back, you tweak it a little bit, and vice versa. So it’s a really fun loop to build and launch AI tools in. It’s particularly good for people that are what we would call knowledge entrepreneurs, which basically means someone who is turning their knowledge or expertise into products or services that they sell to people. Coaches are knowledge entrepreneurs, consultants, course creators, even people at normal companies — anyone with knowledge or

expertise that they’re trying to sell people through a video call, a course, blog posts, hands-on consulting. AI is a really cool way to scale that, and we see a lot of successful people in Pickaxe do that sort of thing — they take their knowledge and turn it into some sort of chatbot, probably the simplest example, and people pay to use it or use it and get interested and then contact them for the rest of their services. It’s not in any way a full replacement for a knowledge entrepreneur, it’s really just a way to offer a fraction of their services in a super scalable way, because if you put a chatbot on your website

tons of people can use it, as opposed to giving people one-on-one feedback or consulting where the constraint is obviously your time. On that note, Mike, just to put a little spotlight on the specific apps you’re able to scale — we talked during the knowledge entrepreneur panel and in the past about the idea of micro-SaaS applications solving super specific pain points, whether it’s a free mechanism for nurturing your audience or helping them solve a problem, and then they get to know you more and potentially join your other programs. Can you maybe talk through some of the strategies you’re seeing people use to monetize Pickaxe — how they’re using

this micro-SaaS approach to create something that’s monetizable? Totally, yeah, that’s a really good thing to talk about. Micro-SaaS — if Pickaxe is a SaaS product that lets you do no-code AI tool creation, that’s quite broad. Micro-SaaS is super specific. One I made that went viral was called Shakespeare Translator — very simple, you put in Elizabethan English, usually written by William Shakespeare (sometimes by other people), hit submit, and it translates it into modern plain-spoken English. That use case is so specific that, believe it or not, there are hundreds of thousands of people searching for that on Google. It’s not the hardest thing to

spin up, and it’s certainly not something that has a huge moat — you wouldn’t want to make a whole business around it — but it’s something a lot of people are looking for all the time, and when they want it, they just want it. You can do that in ChatGPT, but you’d have to think of it and prompt it. By pulling that use case out of a large language model, labeling it well on its own web page in an SEO-friendly way, and doing minimal marketing, that thing has gotten over two million uses. I haven’t tried to monetize it, but many people try to monetize those

things — for us it’s just an interesting lead generation thing. In terms of your own services, you might think of something pretty simple that covers maybe 80% of the first couple questions clients ask. If you can pull that out as a micro-SaaS product and optimize it around SEO, it’s really helpful. Maybe you help people prepare their wills or estate documents, and the number one question they always ask is about how inheritance is taxed in their state, or something like that. You could

quickly make a little calculator that does pretty easy math and turn it into a micro-SaaS product, and you’re guaranteed to catch a lot of people Googling that stuff — and of course there are much more complicated parts of creating estates and wills that they’ll then want to consult you for. It’s a great way to take a very simple part of your service and pull it out as a micro-SaaS, and it’s probably the easiest way to get started with something like Pickaxe. When we talk about where competitive advantage lies here, you mention that some of these tools don’t have a huge moat around them,

so they might not be that difficult for someone to replicate or monetize, but you can also very quickly spin up some of these tools just to service your own clients — you’re talking less about monetizing on the front end and more about how it leads people back to consulting or coaching. So I think in perfect timing for this question — one is how credits work inside Pickaxe, and two, I’d love to see if we can jump in and make a Studio, if you’re in a spot to do that, Mike, showing how you’d

go in and set up a basic Studio, maybe a basic form or calculator we could use as an example, and give people a mechanism for going from start to launch in a very MVP fashion — just to walk people through how to get the ball rolling inside Pickaxe. Totally, let’s hop in. But to answer your first question, how do credits work — I’m pulling up Pickaxe. You’re able to plug in APIs to some degree, and there are a few different ways they might work, and I know some people create Studios just from context, and that’s totally correct. So if you sign

up for an account, we give you credits, which are made up — an abstraction — but basically it’s free LLM or model usage. Every time an AI generates a response, that’s one credit, whether you say “hi” or ask for a 5,000-word essay. But eventually, and actually immediately, you can put in your own API key from OpenAI or Anthropic, and then you’re just paying for usage directly from the model provider, as opposed to the credit system where we’re the middleman reselling you credits, which is a pretty common pricing format among

these AI platforms. Generally I’d say it’s always better to put in your own API key, because as someone who thinks about these pricing models, the AI platforms don’t know exactly how much usage you’re going to use, so we always price credits at the higher end of the usage spectrum, whereas with your own API key you pay exactly at cost. That way you can control the cost you’re getting charged, and you’re only charged when you’re actually using the tool for that credit specifically — separate from that, there are also a few different account tiers inside Pickaxe that get you access to more features. One other

common question we get a lot is people trying to figure out how to price their AI products for their customers. Some people worry about the cost, but generally these AI models — especially newer, cheaper ones like GPT-4o mini — are incredibly cheap, we’re talking hundredths of a cent per thousand tokens, and the trend is that they keep getting cheaper. So it’s generally not something to worry too much about. Here we are at the homepage of Pickaxe, and as mentioned, it’s a great way to scale yourself — this little thing here is actually a Pickaxe we made and embedded, and you can talk to it and it will ask you questions and

help you make a tool — it kind of interviews you. But we’ll start from scratch today. It would be helpful if there’s a use case someone here wants to explore, and we can make it together — otherwise I can make one up. Any ideas for a use case you’d like to see? One participant, Tyres, mentioned working on comparing a

resume to a job description — taking a CV from LinkedIn, comparing it to a job description, scoring it on a scale of one to 100, and then creating a custom resume based on what’s already there (pulling relevant keywords and phrases from the job description rather than rewriting from scratch), and also generating a custom cover letter based on the missing 25%, drawing out transferable skills to put into the

cover letter. Gioia suggested doing a simplified version: a tool that looks at a cover letter and resume together and explains what the resume is missing, giving actionable feedback on how to adjust it. To build the tool, you click Create, and there are two types you can make: a chatbot (a back-and-forth conversation) or a form, which is a bit different but in some

ways a much better experience. Pickaxe has a cover-letter-writer template, but the demo builds one from scratch. On the left side you write your prompt and configure settings and add documents; on the right side is a live preview of the form. Starting from the world’s simplest prompt — “write a cover letter for [blank] job” — whatever you type into the job-title input appears

in the preview, though the results are very generic since almost no work has been done yet. This demonstrates how the form works: build on one side, test on the other. You could also restrict a tool tightly (for example, a cover-letter writer only for software engineers and designers), but for this demo they build a more complete version: get a job description, get a resume, and write a cover letter based on both — “write a cover letter for a job with the following details… the person applying has the following resume…” You can adjust how many tokens or words people can paste into an input, and for

navigation there are Prompt, Configure, Learn, and Act tabs. Prompt is where you talk to the builder in natural language, similar to talking to ChatGPT or Claude — Gioia writes it in a structured, programmer-style format, but any style works, and generally the more consistent effort you put in, the more consistent the results. Configure changes settings, like how many words/tokens a user can paste into an input, or how many words the

output can contain — described as fairly advanced and not usually necessary to touch. The Learn tab matters more for some use cases: you can upload files, add web pages (which get scraped), and even add YouTube videos, which Pickaxe visits, scrapes, and uses the transcript as training material. You can upload up to 50 documents on some tiers and 100 on others. This is where you put knowledge the AI should have — it gets

turned into referenceable chunks so the tool can look at the relevant parts of the knowledge base to answer a question. Gioia gives the example of Pickaxe’s own support chatbot, trained on blog posts and sample Q&A about how Pickaxe works. Whenever it answers a support question it looks at that uploaded information. Importantly, the knowledge base is added on top of the underlying model’s existing training (a chatbot can already tell you about the Roman Empire without citations), but when you add documents

it can point back to the specific chunk it used — a citation feature Gioia calls important for trust and observability, since knowing when an AI answer is and isn’t grounded is one of the biggest roadblocks to adoption. Finally there’s the Act tab, added in the last couple of months, which lets you connect actions — essentially APIs — so the tool can do things a plain language model can’t on its own: search Google, generate images, generate PDFs, send emails, or hit a Zapier or Make.com webhook to trigger automation chains. You create a

labeled “button” in natural language that the AI learns to press and explain when to press it. In the demo, an action called “send summary email to owner” is set up so that whenever the tool generates a cover letter, it emails the tool owner the applicant’s name and the job applied for — described entirely in the prompt in natural language (“when you generate a cover letter, email the tool owner the person’s name and the job they are applying for”). Questions from participants covered styling and branding (addressed later, since the Builder handles functionality and a separate step handles design/branding after publishing) and whether the knowledge base

uses retrieval-augmented generation (RAG) — yes, described as fairly standard and state-of-the-art across AI platforms. RAG means the tool does not read an entire uploaded knowledge base on every question; instead it retrieves the most relevant chunks (you can adjust how many) using vector similarity, similar to a very complex indexing system, and only reads those chunks before generating a response. On formatting recommendations for uploaded documents: plain text, markdown, and HTML all work reasonably well; messy PDFs with complicated tables (for example, financial reports) tend to perform poorly because it’s unclear where information lives once text is stripped out — Gioia compares the AI’s ability to parse a document to a college intern’s. CSVs and spreadsheets are handled well, with each row broken into its own chunk. Pickaxe also has a “Knowledge Explorer” feature that lets you see exactly which chunks get retrieved and their relevance score for a given query, demonstrated live using an Airbnb-listings knowledge base — search terms like a host’s name or a neighborhood produce different, scored chunks of matching listings, and you can edit or add context sentences to chunks to influence how they’re retrieved. Returning to the cover-letter tool build, additional user inputs (resume text, etc.) were added, and a participant asked whether there’s a cap on the number of input variables — there isn’t a known limit, though Gioia doesn’t recommend adding too many. Design options let you use AI-generated images or upload your own branding, change the chatbot’s icon, and adjust various visual settings before publishing. After publishing, a tool can be deployed in two ways: embedded into an existing website (as an iframe, the simplest method, or as a script embed for more advanced customization of size, colors, buttons, and fonts — similar to how a YouTube video is embedded), or added to a Pickaxe Studio, which spins up a full web app with more of Pickaxe’s own technology. In the Studio, you can register users, verify emails, and — most importantly — limit and monetize usage: create different tiers (for example, guests get five free uses, users who share an email get ten, and a paid tier at $10/month gets unlimited use), set up subscription or one-time billing, or make a Studio invite-only for people already paying for another part of your service. A Studio can host multiple tools together, similar to how ChatGPT surfaces different tools, and you can customize branding, logos, and the “out of uses” popup messaging shown to users who hit their limit. The Monitor tab lets an owner see usage and conversations, sort by tool or user, manage and upgrade individual users (including manually granting higher tiers), export or bulk-add users, and — a newly released feature at the time of the call — set up “user memory,” where a Studio can be instructed to listen for and remember specific details a user mentions (such as their job or professional goal) across conversations, building a lightweight profile per user that tools can reference to personalize responses. Toward the end of the call, participants raised support-style questions: one attendee on a legacy AppSumo tier reported being unable to add an OpenAI API key (no dropdown appearing), which Gioia offered to help troubleshoot directly; another reported a recurring bug where a form kept repeating the same question back to a user regardless of prompt changes, which Gioia attributed to possibly needing prompt simplification or a look at memory settings, and offered to review the tool directly. The session closed with Gioia noting Pickaxe was continuing to expand its user-memory and observability features, Darby thanking Mike for the workshop and reiterating the discount code for the first three months of a Pickaxe subscription, and Gioia’s parting encouragement for aspiring knowledge entrepreneurs to “scale yourself” using these tools.