Just Products #7: Using AI driven App Creation tools and Building in the HR Space
We explore different tools between red and blue ocean markets and end with a note on some examples of radical New Year resolutions.
Just Product brings you by-weekly Product Ideas, Stories, and Thoughts about Product Development.
Headlines this week
Using Bolt.new and lovable.web
Prompting when building
Some simple app ideas I tried to build
YouTube Channel Directory Builder
Movie Review Generator
White Paper Builder
Summary
Building in The HR space: Part 1
Thoughts on Red vs. Blue Ocean Idea Generation
Subscriber teaser answer
Using Bolt.new and lovable.web
Bolt, new, and lovable.dev are app creation tools for building simple web-based apps with back-end and front-end functionality. You can create your logins to sign in, but both platforms prefer GitHub SSO, which automatically signals, “You need some basic coding understanding.” This made me think, “What a win for GitHub! " I made a new profile because I can’t remember if I used my old one at any given time. It also made me think: Who’s going to rival Github?
Of the two, I used Lovable.dev because Bolt.new didn’t handle natural language commands. After three chats, I ran out of free interactivity, which you can’t manage except for a few errors. You can do much of the same using the Antropics Claude model, but the whole use case for building and deploying web apps seems like a UX thing you can’t have in the same interface as a general-purpose large language model.
My app ideas included data from existing platforms, so the building suggested I get and maintain the APIs myself. Here, you phased a level of technicality I hope someone obscures;
You must create AWS or Google Cloud accounts to request APIs for services built on top of respective cloud service providers. These account creation processes are not tailored for someone non-technical, even if lovable.dev and bolt.new are increasingly tailored towards such individuals.
Depending on the service, it may take up to a week for the service provider to accept your API request, and even then, they may deny it. This simple notion prevented me from testing some ideas further.
I imagine some marketplace along this app builder software where they make deals with different service providers so that the end users/builders can get the data into use with a “one-click” option or for some nominal fee. To gain access to IMDB or TMDB (app idea described below under the following headline), I had to create accounts on their platform or create an AWS account first and then request API access. The tool I’m imagining could be a merge.dev or kombo.dev for public consumer websites that don’t like to be scraped and could imagine monetizing their data…. In saying this, I found RapidAPI. Seems like a viable partner for lovable.dev.
Generally, the problem I wanted to talk about was that requesting and waiting a long time to get access seems to deter people from building apps, given the speed at which they can do so.
Prompting when building
Another aspect of using these app-building tools is mastering the art of prompting. Not that I’m a master, but many people out there, like Reuven Cohen, with his tool for generating symbolic math prompts, apparently mastered Open AI’s O3 level reasoning with neuro-symbolic prompting two weeks before the release of the O3 model. Regardless of whether that is true, the point is that what and how you prompt can save or cost you a lot of computing.
Before prompting lovable.dev to create specific tools, I asked perplexity (arguably other models are as helpful) to generate a prompt to use in lovable to develop the tool. It arguably gives you better results, and after testing, you get a bit further than you otherwise would, using only the first blurbs that come to your mind.
Some simple App ideas I tried to build
The ideas listed aren’t complicated or revolutionizing products. Nor do they produce outputs you couldn’t get to with the help of any large model (chatGPT, etc.).
However, for any repetitive task, it seems sensible to build a simple app that consistently gives the output in the same format (after figuring out the optimal one). Using app-building software, it seems relatively easy to deploy scalable products that can quickly capture value, even if that value doesn’t last very long.
This only makes me revert to the House of RadicL's vision. What sort of organization could last for decades in an increasingly fast-paced world?
Without iterating on that thought, here are a few app ideas. Maybe you want to continue jamming on them? I’ve included the links to lovable.dev.
YouTube Channel Directory Builder
This idea was based on my thoughts about improving Slush’s media content on YouTube. That channel likely has a year's content but is not effectively organized based on the target audience's interests. There is also a wealth of knowledge that, if organized correctly, could be used as a learning path for specific subjects.
The YouTube Channel Directory has working API access to YouTube, and when you drop the channel, it will fetch and try to organize it according to themes it recognizes at the channel. I am drawing some inspiration from the work done by indie makers like
here on Substack.I spent a few hours trying to get the tool to use algorithms to analyze the Slush channel, reorganizing the clips according to domain themes like “scaling,” “fundraising,” etc. It found a hugging face algorithm that could crawl the data, but going through 2000+ videos was a two-hour-long processing endeavor, and I lost patience.
Nonetheless, I bet many channels are improperly organized, and their video metadata is not in order. There is so much content that it will take years for someone to watch it all and manage it effectively. In other words, there is some arbitrage for a job ideally suited for an AI agent.
Movie Review Summarizer
I wanted a tool that would allow me to create an automated version of the Just Culture Movie Reviews, the “Just Culture Movie Review Summarizer.” You’d type the movie, and you’d get;
Ratings from all significant rating platforms IMDB & Rotten Tomatoes etc.
Trailer
Where you can stream/watch the movie
Plot Summary
Quotes from different sources like trending Reddit or X discussion about the movie
A custom-generated “review” based on Just Culture writing / tone of voice / some hypothetical guesses of what I would think of the movie that I could edit based on actual thoughts.
All in a nice layout format. The tool, of course, does none of that. As explained earlier, I get stuck with API access hurdles to some platforms. I, however, found TMDB, seemingly rivaling IMDB. It would be nice if they had a YouTube channel to go along with the platform one could subscribe to for the latest trailers.
White Paper Generator
The White Paper Generator sparked out of the need from Peik doing Corle (highlighted in Just Products #5: Ventures that should be backed). Content-wise, it’s nothing you couldn’t generate in ChatGPT. However, getting chatGPT to go and fetch data from different specific platforms could be a use case. Having a whitepaper builder that integrates with varying research databases is a specific enough use case for someone to generate some passive income.
The generator is not integrated with any LLM, so you won’t get anything but a design of what you input yourself. Quick market benchmarking, and it seems Storydoc is your preferred tool for the use-case, but they have all kinds of document generation, not just white papers.
Finally, it’s worth noting someone should come up with a better name than “white papers.” It just has an unnecessary racial vibe to it.
Summary
Give this another 3-5 years of development; you can imagine tools from scratch. In a b2b context, I wonder if companies will prefer building their systems instead of buying them.
With an internal team of 3-5 individuals who know what they are doing, you could likely completely substitute some software on which you’re spending a lot of money if you’re a company with 10M+ ARR and a few hundred employees.
It kind of makes the whole next section feel like a void.
Building in the HR space: Part 1
In Just Thoughts #36, I promised to discuss the HR tooling space. I also gave some thoughts to Taito.ai and In Parallel, building at the intersection of the “people space” and company operational tools. I’ve been drafting this section for quite some time but have mainly lost intrinsic traction on the subject. Nonetheless, I can use this section as an example to bridge into a more high-level thematics about what sort of space your building is in and continue with the actual space later on in “Part 2”.
I spent five years within the People team function, working on system onboarding, company culture development, internal training, recruitment, analytics, leadership coaching, and other projects. I have worked with multiple tools. Hence, I am interested in writing this down and trying to figure out if there is something I’d like to build in this space. If you’re a founder or entrepreneur trying to figure out what you should be thinking about when it comes to scaling an organization and the people aspect of it, you can read some of the first writing I have on Substack;
The Hr space represents a 10B+ opportunity, but this b2b space is crowded, and buying cycles are long.
What makes them long is that once you get yourself a vendor, you’re looked in primarily because of two reasons;
The products don’t distinguish themselves much in core functionality (not entirely true, but more accurate in mature markets)
HR tools are highly integrated into a company's operations, and the entire workforce needs to be educated about using them.
What adds to the complexity is the stack's fragmentation and the roles associated with using the b2b people software. A few software vendors sell systems that cover the whole stack, if any, depending on how you want to scope things.
You essentially have four categories, with a fifth I call ”other” of complementary.
HR - Software
Recruiting - Software
Engagement or Performance Software
Learning and development software
Given the ever-growing trend of well-being struggles at every level of society, you could add ” well-being software” as its category. Otherwise, you can categorize everything in the abovementioned categories, with one extra for ” miscellaneous complementary software” or ”other.” Depending on your domain and size, you’ll also have financial elements for your product, like payroll management. Still, these are better viewed as features of the HR tooling than a stand-alone category, mainly because the tooling a CFO wants is not the same as what the CPeO wants.
An example of "miscellaneous” could be Kombo.dev, which started by solving a problem caused by the fragmentation of the stack: system compatibility. Few vendors provide the whole stack, and the mid- and long-tail market rarely demands it. Therefore, different systems need to communicate effectively. Maintaining APIs and ensuring their work can be outsourced, and Kombo.dev is your provider.
When building in this space, you need to understand the order in which people buy the different software and the ”default” capabilities required. Additionally, we must remember that no one wants more than three tools to use when they work, even though people usually have many.
Finally, unlike other SaaS tools, it's harder to prove the tool's bottom-line or top-line effects, even if the impact on a company, such as attrition and employee well-being, can be quantified. However, as with most tools, it’s about making the customer believe in the promises it can deliver.
I’ll continue with this in HR later, but the main takeaways are as follows:
“People Software” should be automated and obscured behind the organization’s primary chat interfaces (like Slack), regardless of category.
There aren’t any other than red ocean opportunities in the people space unless;
you regard AI agents that substitute the need for HR professionals as a blue ocean opportunity
Organizations completely change how they operate due to AI, e.g., you’ll have In Parallel tools that become your primary interface for executing work, and everything integrates into that / those platforms.
Creating a tool with better data integrity, fewer bugs, and a more enjoyable interface is a viable strategy (What Linear did to product development).
Thoughts on Red vs. Blue Ocean Ideas
The HR space is a red ocean market. It’s as red as it gets. If you’re not familiar with the differences between red-ocean and blue-ocean strategies, here’s a perplexity generated answer;
Red ocean and blue ocean markets differ significantly in their competitive landscape and strategic approach:
Market Space: Red oceans represent existing, known markets with defined boundaries, while blue oceans are untapped, new market spaces (1, 3).
Competition: Red oceans are characterized by fierce competition and saturation, where companies fight for market share. In contrast, blue oceans have little to no competition, as they focus on creating new demand (2,5).
Strategy: Red Ocean's strategies aim to outperform rivals within existing market boundaries, often through price wars or product improvements. Blue ocean strategies seek to create innovative value propositions, rendering competition irrelevant (6, 8).
Growth Potential: Red oceans often have limited growth due to market saturation, while blue oceans offer more significant potential for profitable growth by tapping into uncontested market space (3, 8).
Value-Cost Trade-off: Red ocean strategies typically involve choosing between differentiation and low cost. Blue ocean strategies aim to pursue both simultaneously, breaking the value-cost trade-off (8).
What will AI do to the Red Ocean Markets? It will probably render many providers obsolete, as customers can build their solutions at a fraction of the cost. It is kind of the way OpenAI’s ChatGPT updates have been obliterating entire markets every quarter.
What will AI do to Blue Ocean Markets? It will make them grow ridiculously fast and lower market entry rates, speeding up the cycle from blue to red to obsolete. How do you hedge yourself? Probably quite a few ways, but I increasingly feel brand loyalty and super fandoms can shed a few learnings on strategies to employ.
I hope you enjoyed these product thoughts! Are you working on a blue ocean or red ocean market idea? If so, why would you place it in that market?
I’m hoping into the blue ocean to see if we could filter the world from crappy content before AI can generate so much that we get lost in a sea of “average.”
Subscriber teaser answer
In the email header, I share some exclusive thoughts and post-engagement reasons for subscribers. Here’s the answer to this week’s teaser about my New Year’s resolution last year to quit nicotine until “further notice.”
What made me quit? The thought was that if someone would hold me at gunpoint and tell me that if I used nicotine, I’d die, I could resist using it. However, if I’m held at gunpoint and I’m asked to control mental health issues or, e.g., cry, I can’t necessarily prevent my emotions or mental health issues from manifesting.
Being able to make this singular choice is a testament to the difference between addiction issues and mental health issues. Someone trained in dealing with the situations described above may feel differently about being able to control yourself in life-threatening situations.
However, I don’t think a mentally ill person can control their symptoms (Even if fictitious, the new movie Joker gives a good example; in the end, he’s not able to control his reaction to what happens to him), and that’s the only thing that matters if the goal is to find a reason to quit a habit that is not good for you.
If you want exclusive subscriber thoughts and support for this publication, subscribe now!
Who should see or hear about these ideas? Share if you’re inspired!
Happy New Year!