The Robots are Coming
/
Greetings everyone,
I have something very exciting that I want to tell you about this week. Something that will finally fulfill a promise that I made several years ago. The promise was that I would re-organize my YouTube and website content to make it searchable in a way that would allow anyone to find exactly the content that they need, quickly and accurately. Doing this manually, which required reviewing every scrap of content, indexing the best bits by subject, and building a practical database proved to be an impossibly complicated task that, at the end of the day, would result in a final product that was about as accessible as the work already was, in my YouTube back catalog - which is to say, not at all accessible.
I have recognized all along that having a way to retrieve specific content from my extensive collections on YouTube and my website was going to be a mandatory condition for the continued growth of the channel. However, the best I could come up with was adding a search function to my blog page, which turned out to be a much needed improvement, but still not able to find the content without considerable effort and some prior knowledge of the appropriate search terms. We limped along like this for a long time, though it quickly became apparent that there was another factor that was preventing useful extraction of information from my videos and articles. To successfully find specific content through the search function on my blog page (https://www.allanwallsphotography.com/blog) relied heavily on my provision of complete and precise video documentation, along with accurate tagging and key-wording (a fact I only learned much later). As I had been doing none of this, it rendered the search tool frustrating and unhelpful.
Accepting help from outside the organization, an idea that seemed reasonable at the time, proved instead to be a huge mistake. I was so frustrated with the situation, having hundreds of hours of content but no way for viewers to take advantage of it on demand, that my judgement was impaired. Despite overwhelming misgivings, I allowed a slick-talking and self-proclaimed tech-master to come onboard to lead a restructuring of the channel’s operations, including the development of a searchable content database. Many months later, with no sign of the promised content library, it gradually dawned on me that I had been right to doubt this person’s integrity. Eventually it became clear, through official documents that this individual had a hidden agenda and a plan to repackage and commercialize my content through a proposed new company, controlled solely by a new majority owner. Needless to say, my naiveté in engaging with this person was a huge mistake, and while this mistake may still pose a threat to the viability of my company, it also provided me with a lesson that needed to be learned. So, here I was, another year older, and no closer to the searchable content database that had become something of a millstone around my neck
That was such an unpleasant experience, leaving me angry and embarrassed, that I did not look any further for outside assistance with this chronic issue. I did ask for, and received, help from a wonderful group of supporters who volunteered to help me with my next idea for building the database. I am eternally grateful to this group of supporters who took the thankless task of reviewing all my content, organizing it into a catalog, assigning relevance scores to the content blocks, and facilitating the building a searchable data base. These wonderful people were all experienced in business and/or data management, and were led by two long time supporters of the channel. Though this awesome group worked hard to produce a huge amount of of data characterizing the content in a great many of my videos, the data proved to be too granular and fragmented for me to work into a tool that could offer practical assistance to a viewer looking for specific content. I must quickly add that it was my own limited abilities and resources in this area that brought this project to a halt. The data is still here and it may one day be part of the solution. I would point out that this effort was made recently enough that some of the Artificial Intelligence language tools were already appearing on the scene and had been involved in this unsuccessful attempt at organizing my mess.
The men and women who were part of this initiative have my gratitude and respect for all they did. Thank you Amy, Susan, Mike, Jeff, Jillian, and everyone else who offered assistance. To each of these friends I will reiterate that the eventual failure of this strategy was entirely my fault. Overconfidence, underestimation of the size and scope of the project, and my complete lack of time-management skills all conspired to slow and eventually halt all forward momentum. And with that I found myself right back at square one - making content faster than I could document a way to find it once it was made. I thought that would be it, nothing would change, and I would keep on doing the same thing that I had been doing for the last seven years. It turns out that even in this assessment, I was wrong.
This past weekend, during the first Pzoom meeting of December, something happened that, in an instant, transformed this elephant in the room into a real opportunity to reimagine this channel and give my audience what they have been asking for all along. This epiphany had nothing to do with me. It was a result of an idea proposed by my dear friend and long-time supporter, Patrick Stahel. Patrick demonstrated a project he had been working on that used an agentic AI resource (NotebookLM) in the Gemini family to assimilate the transcribed content of most of my back catalog into a fully functional interactive data repository, accessible to anyone with a computer and internet connection. In plain English, Patrick had fed a huge part of my accumulated content into a computer program that used this content, and only this content, to build something very similar to Chat GPT but on a very limited scale. Patrick was able to take yeas of my transcribed content and turn it into an AI powered interface that any of my viewers can use to find exactly what they want from my back catalog, in precisely the format they prefer, in an instant.
A dragonfly smile
Let this sink in for a moment. I am not giving you a list of video titles in which I talk about focus stacking, or bee photography. This is vastly more powerful. You can ask this agent to scan through every word of my content, since the first day of the channel, and give you answers to questions like this:
Q) please recommend the best three pieces of video content related to insect cleaning techniques. Prioritize videos with detailed demonstrations. Exclude videos longer than 45 minutes in duration.
Q) Please find every piece of video and written content in which Allan gives gift suggestions for macro photographers and list the gift suggestions from lowest to highest price. Include a brief summary of each gift’s key specifications.
Q) Please list the top ten field-technique videos based on the number of viewer comments posted on YouTube. Summarize the viewer comments and order the videos based on how helpful the commenters found each video.
I could go on… How you will use this tool is entirely up to you. You get to ask the questions and the bot (for want of a better word) will do the heavy lifting. And it works. Much of the work has already been done, but I plan to start over, building the database with all of my previous work. It will include not only my videos but over 300 hours of livestream discussions and as many printed articles. I think this is at least as important as including video content as my livestreams have evolved into some of the densest and most thoroughly researched content available through my channel. And I have covered a great many topics not discussed in previous videos. I would say the same for some of my written content, which will also be included in the resource pool.
I am not going to rush this process and part of the reason for starting over from scratch is so that I can learn as much as possible from the process. Once the Agent is built it will need to be tested and probed for weaknesses. I will be asking a few people to stress the system to bring out any issues that need to be addressed prior to release. With testing complete, the Agent will get a catchy name and will jeavailable for you to use. I will post the link on my website, the Walls-app, Discord and Patreon as soon as it is launched. It will cost nothing for you to use and I will continue to keep the tool current, uploading each new content block as it is released. While I am obviously not the first person to use today’s amazing technology in this way, I must confess that I am unaware of anyone in the macro-phtography space who is using it in this way - and that is very exciting!
Let me curb my enthusiasm long enough to offer a few words of caution. I am enthusiastic about the promise of this tool for one primary reason - this is exactly the kind of thing that LLMs (Large Language Models) do extremely well. LLMs are made for this kind of advanced language processing, it is their superpower. BUT… (and you will hear this from me again) you must keep in mind that the interface I am giving you access to has a reality that consists solely of the words that I have strung together in hundreds of hours of content. It knows nothing about the world beyond my content (except what it may have picked up from my content) and so it will not be able to respond to prompts about other matters. Also, I make mistakes and there are likely to be inconsistencies in some of the material that I taught five years ago and the way I treated the same content last Tuesday. In other words, you will have access to a machine that can find and organize every word of my published content, but what you do with that information is up to you. Going forward I will be looking for ways to use this remarkable technology to correct earlier mistakes and make the output more up to date, coherent, and complete. But that is going to take time. As has been my goal since the first video was published, I want to produce content that is an accurate, complete, enjoyable, and as accessible as possible. This new development will move me a little closer to that goal. As we develop this tool, I will be counting on you to tell me what is working and what is not. Do not hesitate to tell me when there are improvements to be made - If this isn’t working for you, then there is no reason for me to be doing it.
*****
Before I go, let me briefly outline the week ahead. As usual, we start the week with Tuesday’s Macro Talk, at 8pm central time. This week I am planning to share with you the ultimate holiday gift list. It is actually a wish list for macro photographers, so unless you have one of these rare creatures in your life, you can use this list to guide your hint dropping activities over the next couple of weeks. Your link to the livestream is right here… https://youtube.com/live/i_CtzBlGoVE?feature=share
Thursday at 2PM is Macro Talk Too where we will expand on a discussion we had several weeks ago and talk about your photography plans for the coming off-season. Last time I visited this topic I talked mostly about interesting activities that could replace macro photography during the coldest months, but this time I want to talk about some actual macro activities that either take advantage of the season, or are at least accessible when the frost is on the pumpkin (is that a real saying?). Anyway, some of the ideas I want to discuss are unusual enough to come as a surprise to some of you. Your link to this livestream is here… https://youtube.com/live/a-127D-7Dug?feature=share
For more information about these streams, please visit my website at https://www.allanwallsphotography.com/blog/gift
I will be taking a few minutes at the beginning of both of these livestreams to talk about the new website feature described above, going into a little more detail and giving an update on my progress. This would be a great time to ask any questions you may have, or make any suggestions about features you would like to see included. I will also be able to give a better estimate of the timeline for getting this tested and into service.
Friday is Crystal Art, with Harold Hall and your’s truly. This is a live Zoom event for anyone interested in the fascinating world of birefringence photography. We talk about growing and imaging crystals to make stunning abstract art. If that sounds like fun to you, please join us on Friday, December 12, at 2PM. Here is your invitation to the event. Remember that we will record the meeting for release on YouTube - if you don’t wish to be in the video do not turn on your camera or microphone while in the meeting!
Allan Walls is inviting you to a scheduled Zoom meeting.
Topic: Allan Walls’ Crystal Art with Harold Hall
Time: Dec 12, 2025 02:00 PM Central Time (US and Canada)
Join Zoom Meeting
https://us02web.zoom.us/j/6916802815?pwd=TS9tZi9ZL1NXeVUvOUF4eTg5YjdlZz09&omn=83515007568
Meeting ID: 691 680 2815
Passcode: 678122
Join instructions
https://us02web.zoom.us/meetings/83515007568/invitations?signature=sUhBI8zwugIau4CwZK0BLhbSds787lqhjfJEe2tsM2w
And finally, Saturday sees Episode 37 of AfterStack - a discussion roundtable for macro photographers interested in post processing. A wonderful group of talented artists who love to play in Photoshop! Come learn with us. It is free and you are most welcome. Your invitation is right here -
Allan Walls is inviting you to a scheduled Zoom meeting.
Topic: Allan Walls’ AfterStack with Bud Perrott
Time: Dec 13, 2025 10:00 AM Central Time (US and Canada)
Join Zoom Meeting
https://us02web.zoom.us/j/6916802815?pwd=TS9tZi9ZL1NXeVUvOUF4eTg5YjdlZz09&omn=83925997556
Meeting ID: 691 680 2815
Passcode: 678122
Join instructions
https://us02web.zoom.us/meetings/83925997556/invitations?signature=svqe_uG7LH6yER1QyIELFkMZCG2Kr1xy-HnxxWF9bS8
And that is it for this week! It is a lot for one post and I hope to see you sometime this week.
Allan