I am a big fan of pickaxe having been a long time user. I personally have a lifetime account and then I got my entire organization onto pickaxe using a PRO account. We use it extensively with almost 1500+ users. However, recent developments in consumer facing LLM’s have had a huge (negative) effect on our pickaxe platform adoption.
In the last quarter, there have been major upgrades to every platform. Deepseek-R1 started it, followed by Mistral’s redesign (which includes a canvas, web search and rudimentary agent creation). Meanwhile GPT-4oMini and Gemini have really leveled up in terms of functionality. User file uploads are now available in all these models to all users across tiers including free ones (this was previously a differentiator for pickaxe) and even Gemini Flash now performs very well for routine tasks. Due to this, our usage of pickaxe has been dropping. When we spoke to our users on why they chose free Gemini Flash 2.0 over Pickaxe (where we offer GPT-4o, o3 mini and Claude Sonnet), the number one complain they have is that they don’t like the interface. It’s slow and clunky compared to others. For example, people are so used to viewing the upload in (or around) the chatbox, they don’t even notice that the upload is happening on the top. Similarly, we use pickaxe for creating lesson plans. Now, Microsoft has included Copilot for free with office subscriptions which makes it very, very easy to edit things. (unlike pickaxe where copying and pasting tables/Latex is a hassle and we had to create our own chrome extension to solve this issue). As a result, People now prefer creating a draft on Gemini flash and then copy it in Word where they can just right click to format it properly.
Similarly, we have been trying to use o3-mini in some complex use cases (SQL query generation, evaluating CV’s against a set of guidelines). The current interface doesn’t show reasoning which means there’s a delay of 30 seconds even though the model is working in the background. By then, most users lose their patience. The interface is completely unoptimized for working with reasoning models.
As reasoning models become popular, it is likely that in 6 months most current models would be upgraded to reasoning based equivalents. ChatGPT-4o now also shows its reasoning steps. While I appreciate the hard work the pickaxe team has been putting into the studio redesign, I feel the platform desperately needs work on the UI/UX front to remain competitive, especially when you factor in the progress the free versions of Gemini, ChatGPT, etc.
I am sorry if this comes across as impolite. But I think it’s important for me to raise this flag at the right time. I am not suggesting that pickaxe have each and every functionality that OpenAI offers (like editable canvas, etc.). I understand it’s a small team who is just getting started. But the least we can have is something like OpenWebUI which looks far more modern and responsive enough to give the same feeling as base ChatGPT/Gemini to the user.
And there in lies an even bigger issue in my opinion. Us Pro level users are promised “BETA access”. Many of us have have been waiting for this big “Redesign” for months (Was told it would be January for months and we don’t find out until Jan 31 that it has been pushed back) while paying for a platform that ultimately is going to have a massive change.
I have brought up wanting to know more multiple times and get the response to email ian@pickaxeproject.com to request beta access. I’ve done this multiple times and I know he opens my emails, but I hear nothing back.
This is just one of the broken promises presented by their pricing page. The other promised “Custom User Roles” has been there from the start, but has never shown up. Not to mention “White-labeling”… Now we are being told that the back-end of Studios will not be white-labelled??? So basically we can’t bring in clients (which was the whole reason for wanting these user roles), because they will know we use pickaxe. Why would anyone hire us or even continue to use our services, when they can see that they can just use pickaxe?
I also don’t want to sound like I am whining, but it is honestly frustrating
Agreed, the user interface is subpar and could use some basic fixes – e.g., allow the user to resize the left panel, make the response window full width (vs a narrow box).
This is the first I’ve heard that white labeling might be going away. That’s the reason I chose Pickaxe as a solution. I would be leaving the platform if that happens. I sure hope it does not!
Just to clarify about the White-labelling. The new update will allow us to bring clients into the backend of a studio. Basically we can sell them a Studio now, instead of a singular pickaxe. The issue is that the backend of the studio cannot be white-labelled. As far as I know, the front end will still be able to be white-labelled.
Agreed on all (most) counts. We’re well-aware of the state-of-the-art in chat interfaces regarding file uploads, reasoning models, and other in-chat interactivity (e.g. citations, conversation-based buttons, etc.). Part of our redesign will address cleaning up how our forms and chatbots look to end-users.
@ecomprocess I’ve followed up with an invitation to test the beta. Sorry for the delay. The testing groups have been full—but let’s talk! To clarify, “Custom User Roles” is now “User Memory,” which will be wrapped in a new CRM system for monitoring your tools and users in a way that no other product does—we promise you that.
As for “white-labelling the back-end,” we’re still figuring out what that means ourselves, but there are some practical limitations. At the end of the day, we think your moat as a prompt engineer lies more in your ability to prompt engineer, rather than an ability to completely whitewash any software you use. Similar to how people hire Webflow experts to make Webflow sites. They’re not hiring them on some false premise that they engineer these sites themselves—they’re hiring them because of their web design taste and craft with the tool. Prompt engineering = the new web design.
In short, your problems are heard, and we’re working hard to fix them!
Following up on Ian’s message, I think it’s important to understand that these are companies with billions of dollars we’re talking about comparing Pickaxe’s interface to. A flood of companies have started building awesome consumer interfaces. There will always be flashier interfaces to use AI than through Pickaxe. The value add has to come from the unique (superior/differentiated) behavior that our builders bring to the table.
That being said, as we start to roll out the new version of the product, our bandwith for rapid improvement of other parts of the product will only increase over the next month or so. We’re looking forward to working together to show up all those other platforms!
Right, my point is I think we actually will keep in more lockstep with the cutting-edge of AI interfaces—we just needed to sort out the dashboard designs for how you’ll be able to manage the AI first. Now that work is done, so we can catch up to what’s considered best practice in front-end chat design for your end-users.
Thanks, @nathaniel. I’ve worked for startups/small companies before and I get the need to balance perfect vs MVP, priorities, urgency-strategy, etc. I respect the balance you are striking and the choices you are making. Unless you royally screw up something, I’m here to support your continued growth and development. We appreciate all you and your team do.
I do believe that White Labeling the Backend will be more beneficial to both the PicAxe Team and Us as your partners. As the AI landscape evolve, people who want to build a business with their own brand will keep on looking for this solution.
For example, last week I presented https://ailifecoach.io/ to a potential partner and when they ask me on how they can access the backend to manage the users, models, settings, etc., I showed them the PickAxe backed and they like it very much.
Now here’s why White Labeling the Backend is important! Last Tuesday, a couple of days after our meeting. My potential client called me saying “We have already registered at your company https://www.pickaxeproject.com/ we’ll upgrade to Gold Plan for now and we’ll see how it goes”. They taught that PickAxe is my company.
Anyway, the thing is I did not build AI Life Coach as a B2C SaaS product but a B2B Solution for Coaches and Mentors. And for it to work, they will need access to the backend. Is there a way for me to at least use my domain to access the backend of PickAxe? Something like http://ailifecoach.io/studio? and also rename the tools as “AI Tool” instead of “Pickaxes”?
If your tool is basic enough that your customers can easily rebuild it on our platform, there’s not much we can do to help you. You’ve clearly put a ton of effort into your products, and that work product is what customers of yours are truly buying, not just chatGPT. Because of that, you shouldn’t NEED to white label the backend. We don’t plan on white labeling the backend of our product in any way. That’s not what Pickaxe is for.
Nathaniel, I have the same issue as stated above with white-labeling. I would suggest the team look at “GoHighLevel” and the way they are able to provide surface level white-labeling for the Client facing side. While I understand the teams argument that the moat of “prompt engineering” is what sets our products apart, the fact is, prompt engineering is a very low-level and low barrier to entry skill. So this presents two types of customers for Pickaxe (as far as I can see):
People building tools for their own business, or are selling the tool themselves ie. through a studio
People selling a B2B or B2C service to build these tools.
I’m not trying to “grift” and do B2B with a subpar product, but it is undeniable that the moat of prompt engineering while time consuming, is very low. With much of the user base here such as the previous poster looking to do B2B, having an exposed backend presents a lot of challenges such as - having the client mistaking pickaxe for the brand, diminishes the work we’ve done in the setup (almost like a magician showing how the trick is done), or simply copying the prompts and setting it up themselves. After all, the selling point of Pickaxe is the ease of which one can setup.
With this in mind, I can certainly see many power users in the 1st group stay, but this will drive away the 2nd group of users over time as this isn’t viable nor profitable. I hope this conveys why the 2nd category of users would like to see at the very least white labeling of the backend and custom url integration possible in the near future, even if the source code can’t be white labeled. I understand it’s a small team working on a big project with big goals so thank you for listening and thank you to the Pickaxe team!
