Not insurmountable, but not seeing this working as a market in current form; too much missing trust information to be a value add.
For instance:
- Who is behind (and liable) for each agent?
- How do people know it and/or each of the agents are not just a data harvesting operation?
- What data privacy guarantees are there for site/agent.
- Is site actually anything like a multi-vendor fair market?
With all that's going on in the United States, I'd be hesitant to include 47 in the name of a service you plan on offering. I was expecting this to be some MAGA job board or something, not AI Fiverr.
b. bug report: 1st visit to the site, i choose the coding persona, then the back button, i get a "resubmit form" alert
c. I have a detailed coding assignment prompt and a github link. What happens if the task is only 80% complete? How many iterations before one side cancels this?
I can see this becomming useful as the kinds of tasks AI can do get more complicated.
Just look at the complicated workflows people are making in comfyui for image/video generation. Making these workflows takes a lot of work and knowledge about the latest models, so I can see the use-case for monetizing these kind of multi-step, multi-model workflows.
Although I think the examples on the web site right now are a bit too simple, these look like things you could achieve with out-of-the-box solutions.
EDIT: to clarify: the value add from services that "connect" customers to suppliers (like uber, fiverr, whatever) is nominally there in that a shared marketplace can be used to extend protections to both sides of a transaction while making networking easier.
Agents neither require protections, nor do they really need networking; they're a commodity.
> What types of jobs would you want AI agents to handle first?
This would probably depend on the models available, compute available, and pricing for both.
EDIT: to be more concrete; what capabilities are the agents on offer exposing?
> Any UX or trust issues you’d expect with this model?
You tell me -- are the models ever guaranteed to run in an environment approaching confidential computing? Is any of the initial (query, files, whatever) stored or logged persistently beyond the lifetime of the agent tasked with solving the issue? Are the models run in an environment that's vulnerable to common attacks that could compromise the data provided to the model by the customer?
I agree. Agents today are basically light wrappers around models, which are known quantities, and so few in number that one can rattle off their names. A marketplace might make sense if there was an abundance of specialized agents with significant performance and price variation among tasks, and you would provide value through task-verified ratings and price discovery. But this is not the case. Models are static, and already benchmarked, so what do you bring to the table? You need to think hard about your value proposition; people are asking you why they would not simply use Claude or ChatGPT.
> Any UX or trust issues you’d expect with this model?
Yes, why I should trust those agents? How do they work? GCP have/planning (I'm sure Azure and AWS also working on something similar) to have agent marketplace, you should think about how you would integrate yourself there so you get big name recognizing your agents.
This is an excellent discussion on building future-proof products. In my experience, the biggest challenge isn't just the technology itself, but maintaining an architectural flexibility that allows for the seamless integration of new models and features without a full rewrite. During our work at ion301, we tested various approaches and found that a modular, API-first design from day one was a game-changer for long-term scalability. Have you also found that focusing on the foundational architecture early on is the key to avoiding future technical debt?
For instance: - Who is behind (and liable) for each agent? - How do people know it and/or each of the agents are not just a data harvesting operation? - What data privacy guarantees are there for site/agent. - Is site actually anything like a multi-vendor fair market?
curl: (7) Failed to connect to 47jobs.com port 443 after 125 ms: Couldn't connect to server
IMO no, as the tasks could be done cheaper and maybe with equal quality by interacting directly with a consumer LLM (eg. ChatGPT)
> Any UX or trust issues you’d expect with this model?
Currently yes: as for real freelancers, you'd expect a portfolio, with examples of projects done for clients
b. bug report: 1st visit to the site, i choose the coding persona, then the back button, i get a "resubmit form" alert
c. I have a detailed coding assignment prompt and a github link. What happens if the task is only 80% complete? How many iterations before one side cancels this?
c ) you get completed work , if no completed work the the issue is escalated
Just look at the complicated workflows people are making in comfyui for image/video generation. Making these workflows takes a lot of work and knowledge about the latest models, so I can see the use-case for monetizing these kind of multi-step, multi-model workflows.
Although I think the examples on the web site right now are a bit too simple, these look like things you could achieve with out-of-the-box solutions.
EDIT: to clarify: the value add from services that "connect" customers to suppliers (like uber, fiverr, whatever) is nominally there in that a shared marketplace can be used to extend protections to both sides of a transaction while making networking easier.
Agents neither require protections, nor do they really need networking; they're a commodity.
This would probably depend on the models available, compute available, and pricing for both.EDIT: to be more concrete; what capabilities are the agents on offer exposing?
You tell me -- are the models ever guaranteed to run in an environment approaching confidential computing? Is any of the initial (query, files, whatever) stored or logged persistently beyond the lifetime of the agent tasked with solving the issue? Are the models run in an environment that's vulnerable to common attacks that could compromise the data provided to the model by the customer?I am curious though -- why _47_ jobs?
I agree. Agents today are basically light wrappers around models, which are known quantities, and so few in number that one can rattle off their names. A marketplace might make sense if there was an abundance of specialized agents with significant performance and price variation among tasks, and you would provide value through task-verified ratings and price discovery. But this is not the case. Models are static, and already benchmarked, so what do you bring to the table? You need to think hard about your value proposition; people are asking you why they would not simply use Claude or ChatGPT.
why are there people names attached to the services? i thought it was supposed to be all ai?
It does, however who is your target market?
> Any UX or trust issues you’d expect with this model?
Yes, why I should trust those agents? How do they work? GCP have/planning (I'm sure Azure and AWS also working on something similar) to have agent marketplace, you should think about how you would integrate yourself there so you get big name recognizing your agents.
incase it happens its an indicator its a good market