OP here - I’m a dev, pediatrician and father. I’ve been thinking about this idea for a long time. I recently sold my last company and this seems like a good time to try something a little more risky but that I’m passionate about.
In the distant past most kids worked on farms from a young age (I live in upstate NY and families with farms still have their kids work from a young age). With industrialization we got away from child labor because children were being exploited and exposed to dangerous factory environments.
With Emerald, I’d like to challenge the notion that “child labor” is necessarily a bad thing. As a pediatrician and parent I can tell you that American children could benefit from a dose of the real world. There are many important benefits to having the responsibilities of a job at a young age. I outlined several of these below.
The idea behind Emerald is to use the loophole that kids can work for their parents. We are going to create LLCs and bank accounts (will probably use Stipe Atlas and Mercury bank) for each family and then contract work through that LLC. This way kids can work on projects, earn money and gain experience.
Emerald will be a marketplace similar to Fiverr or Upwork. Companies or people who need work done can post the spec and then will be matched with a workforce who can complete the work.
The main difference will be that Emerald will be much more hands on during the work completion. Instead of individual freelancers bidding for work Emerald will construct teams of youth (prob 2-5 per team - maybe more?) with an adult mentor. The mentor will be paid and will oversee the work. The youth team will then collaborate to complete work. Lots of unknowns here and we will have to iterate on the best process.
There will also be extensive safety tooling and oversight. We plan to require all conversations to occur on-platform (Slack like chat), set time limits and give parents dashboards/controls.
The obvious first type of work to be done are things that teens are good at - content editing, content creation and software dev. Although I’d like to hear other ideas for work that teens might be well suited for.
Benefits:
Earn money
Learn time management
Learn how to work with a team
Learn real world software development or tech skills
Start creating a resume and list of references
Risks:
Time commitment
Increase screen time. Kids already have too much screen time
Shorten childhood - we should let “kids be kids” for longer
Parents masquerading as their kids and doing the work (I don’t know why they’d do this but figure I’d list it)
Questions:
What are some additional risks?
What are additional benefits?
Are there examples of programs like this out there? If so, what are the learnings?
Side note on AI
I’d like to avoid the “AI will take all these jobs anyway” rabbit hole. I’d like to assume we can find something productive for kids (or any of us) to do.
Who I’m looking for
Educators or people who have worked with teens on longer-term group projects - learnings from these projects
People interested in being mentors for teens on Emerald - will be paid positions
People with children who are interested in the initial pilot - no dates or commitment are set yet. Just if you are remotely interested.
Companies/people who are interested in submitting projects that they need completed. We can help you scope the project so it can be amorphous.
This is VERY early for Emerald. We haven’t built any tech yet (aside from the very basic landing page). If anyone has interest in this idea feel free to reach out - my HN handle at gmail or here: https://form.typeform.com/to/Pf6EscNP
I've used Supabase with pgvector to do this. You can even store the position next to the embedding in the db so you could jump to the content in the UI. Ping me at my hn handle at gmail if you want more specifics :).
So you can add credentials by clicking on the agent name and getting to the agent config. The agent in the video that had the credentials was a different private agent I have. Credentials were just added as a feature and I haven't built the functionality yet to fork them (need to fork only the keys not the value). But you can add credentials to any agent. If you can't find them, ping me at my HN handle at gmail and I can give you a quick walk through. Thanks again for checking Fabrk out!
Yep! I'm using OpenAI's assistants api. Regarding the tool forking - I have the canonical function hosted on the server. When someone forks an agent it grabs the agent context and all the tools and creates a new `assistant` with Open AI. When the tool is called during a thread it just points to the canonical function. I'm really excited about the next step - allowing folks to create their own tools and add them to the canonical function list. This will allow everyone to build on each other's work and use tools as building blocks for complex agents. Thanks for checking it out! Lmk if you have other questions!
I built something very similar to your "documentation interaction" agent for a client recently. I used https://fabrk.ai/ and it took a couple days to get to a proof of concept. Happy to discuss the details if you'd like. Ping me at my handle at gmail.
We have a client who's using fabrk.ai to let non-technical folks on the team run sql queries for looker and run analysis of csv/spreadsheet files. Pretty easy for non-tech folks to use.
Great point. We are seeing ongoing consolidation among hospitals and large physician groups. However, patients still deserve to have easy access to their own data. This is largely handled with patient portals that give patients access to some of your data. PatientBank gives you full control of your complete record, including information not necessarily located in your patient portals and information located in separate systems that do not communicate.
This is a great point. As a MD I don't think the process of asking a patient his/her past medical history will every go away. It clarifies how the patient thinks about their medical problems and develops a good rapport.
But as you allude to, having a summary of pertinent medical information, compiled from patient records located in various hospitals, would be enormously valuable for a doctor seeing a new patient. For example, your emergency room doctor having a one page summary of your medical information to be reviewed quickly in an emergency.
It can be hard for patients manage all their health information. That is the problem we are trying to solve! Obviously, this starts with patients having more control over their data. Imagine the next patient you transport pulling up their PatientBank account and it has an up to date medication list, allergies and PMH. Instead of relying on memory they have all their information in one place - when they need it.
Hi there - this is Kevin Grassi, MD with the PB team. You are absolutely correct that many primary care doctors rely on the patient to provide a medical history and rarely need the specific information located in the medical record. However, if you have complex medical issues, your primary care doctor may want to see specific data - especially lab or imaging reports.
Many specialist doctors need to see previous medical records before evaluating and treating the patient. This is especially true in oncology. We work with the Smilow Cancer Hospital, part of the Yale-New Haven Hospital, to help ensure that all oncology patients present for their first visit with a complete medical record.
One final note that is my assessment of how physicians operate in the current system - doctors are accustom to working with incomplete information. PatientBank is striving to make previous medical information more accessible to your next doctor. My hope here is that increased access to information will cause doctors to pay more attention to your data and lead to better care.
Thanks for your question! I hope this clarifies things.