What is the future of GPTS and the corresponding marketplace? As nobody can have a definite answer, I wrote down my thoughts to refine them based on the discussion here.
Thoughts (further explanations at the end of the post):
Assuming a revenue-share model, the upside per user from a GPT is strongly limited and will likely lie below 1$ per user/month or a few dollars in case of massive ChatGPT Plus price increases (e.g., $100/month) – this already assumes only a few GPTS per user, which is optimistic
Given the ease of creation and value-potential for users (simplified experience & multimodal use-cases), especially for less tech-savvy users, thousands of GPTS will be widely adopted
Yet, due to the limits in revenue potential per user (and the development options), only very few high-effort, advanced GPTS with actions based on purpose-build APIs and massive proprietary datasets will remain over time
GPTS will enable rapid prototyping of new ideas and the adoption of GPTS on the marketplace (own and third-party GPTS) will result in ideas being integrated into existing products or be launched as dedicated solutions, utilizing functionalities not provided by GPTS (UI/UX, offline usage, etc.) and dedicated monetization
With people (professional and private) using dedicated apps & SaaS solutions for their main tasks, GPTS will be used to enhance longtail activities (e.g., the LinkedIn post optimizer not for the marketing professional, but the individual posting now and then)
In addition, the cost of using GPTS, dictated by the cost of ChatGPT Plus, will serve as a new anchor for App/SaaS purchase decisions, as many light-touch solutions might be available cheaply via GPTS and dedicated solutions need to be able to provide substantial add. value
Beyond that, I see two main expansions that shift the frontier of what will sustainably remain as GPTS/on the OpenAI Platform (excl. API usage):
A) Additional monetization via developers (e.g., via payment mechanism integrated as “action”, need for paid APIs by the respective provider, transition point to dedicated solution): This will enable more professionally created and managed GPTs, yet limited by the development options provided by OpenAI within ChatGPT
B) Official payment mechanisms for premium GPTs in the marketplace + expansion of development options: Shift of ChatGPT to being the Operating System/Platform for AI-centric applications and thus even more aiming for the revenue-pool of current dedicated apps and SaaS solutions
Many factors will likely impact the future monetarization and strategic positioning:
- Other vendors providing offers comparable to custom GPTS
- Speed with which app/SaaS vendors adopt AI-based features
- OpenAI’s strategic focus regarding AGI vs. SaaS platform provider
- Potential cannibalization of Microsoft revenue streams
- Ability of OpenAI to have developers use OpenAI APIs in their dedicated solutions after piloting them via GPTS or seeing other GPTS (this way OpenAI keeps the revenue)
Additional background thoughts:
Ingoing assumptions: Focus on the public marketplace for GPTS, not GPTS being created within individual companies (here other mechanisms like company-funded developments might influence creation); assuming economic incentives will determine long-term direction; looking at mid-term implications, as adoption/behavior change will take time
On 1): +ChatGPT Plus price – OpenAI cost - % standard GPT usage - % other GPTS usage – OpenAI take-rate on your specific GPT // Example with fully made-up numbers: 20$ - 10$ (50% cost) – 5$ (50% of standard ChatGPT usage) – 4.5$ (avg. usage of 10 GPTS) – 0.15 (30% take-rate on your GPT) → 0.35$
On 2) GPTS substantially reduce the effort for users. They shift the experience from one of exploring and having to think through the right prompts, context, and flow to one of only providing the information specific to the user (e.g., what they eat, where they go on vacation, a picture of a bill). This is particularly important, as the majority of users will not have the same sophistication as the current “early adopters” thinking about GPTS (incl. selection bias of those active in this community). In addition, a group of people developing a GPT can put more time into thinking it through than each user when trying to solve a task via ChatGPT (especially for longtail tasks). On top of that, the native multimodality of ChatGPT can unlock use cases, previously considered quite advanced without major coding effort.
On 3) <1$ per user will not allow for particularly deep and technologically sophisticated solutions that target a specific audience of power users, but rather solutions with a broad user base
On 4) The extremely fast time to launch and first feedback on usage allows for prototyping. On top, existing apps/solutions might look at the top emerging GPTS in their field to copy functionality. As long as those solutions are then built using OpenAI APIs, there might be limited efforts from OpenAI to manage this
On 5) While GPTS might not replace the professional marketing or sales tools for the respective professionals or the gym app for the ambitious athlete, they will provide opportunities for less heavy users. Like me creating a LinkedIn post or somebody wanting to find new HIT workouts. On top, it might allow for simple niche applications in fields that typically did not have any tool support, as it was not economically sustainable (e.g. if there are too few users).
On 6) As an economic actor before getting a new app or SaaS solution in the enterprise setting, I would check what I can do with the existing GPTS. Thus, I will only pay for the value beyond that, which creates a new reference point for purchasing decisions.
Respect for whoever made it until here