Hi everyone,

I’m a 15yo builder, founder of Promptyx, a platform for organizing, versioning, and testing AI prompts. I’m planning our first online competition (promptathon) and looking for a few partners to help shape it. It is still on the idea stages, so looking for initial validation

The idea is to have participants turn a recurring task into a reusable set of prompts and context that someone else can run successfully. Think turning a client brief into a proposal, preparing a weekly report, or creating content that follows a brand’s guidelines.

The challenge would focus on useful results, consistency across different inputs, and clear instructions for reuse. We’re still deciding the exact scope, so input from experienced organizers would be helpful too.

The tentative schedule is three weeks of promotion, one week for participants to build and submit (probably the first week of November), and one week for judging and results.

I’m looking for:

• Prize partners: Tool licenses, API credits, or small monetary contributions.
• Judges or mentors: People with experience using AI for real tasks who could review submissions or run a short session.
• Community partners: AI communities, newsletters, or educators interested in sharing or co-hosting the event.

For partners, possible benefits include attribution on the event page, mentions during the event, an optional workshop, and inclusion in the winning-project recap. We can agree on a contribution and scope that makes sense for both sides.

This is our first edition, so I don’t have confirmed attendance numbers or sponsorships to claim. I’m starting small and would welcome practical advice as well as partnership interest.

If this sounds relevant, comment or DM me with what you do and how you’d like to get involved. Happy to share the draft format and judging rubric. As a teen who wants to build his own things and get into the AI world, this would be great for me!

Not a marketing - only idea
TL;DR: A concept for a reporting tool that connects business data and delivers recurring, decision-ready briefs. Instead of showing only charts, it explains what changed, why it may have changed, what evidence supports the conclusion, and what action to take next. Users can ask follow-up questions, while uncertainty and alternative explanations remain visible. The idea still needs validation with SaaS teams, agencies, and e-commerce operators.

Businesses rarely lack data. A founder can open analytics, payments, advertising, and support tools and still not know what deserves attention first.

I have been thinking about a reporting tool built around that gap. It would connect to a company's existing sources and produce a report every day, week, or month. Instead of stopping at charts, it would explain what changed, suggest why, recommend concrete actions, and accept follow-up questions.

Every important conclusion would show its source, metric, comparison period, and confidence. The report would separate measured facts from interpretations, missing information, and alternative explanations. Recommendations would include their expected effect, cost, downside, and a safe way to test them.

The same engine could support different report structures for SaaS teams, agencies, and e-commerce operators. I would still begin with one group, probably small SaaS teams, rather than build a shallow tool for everyone.

For SaaS, that could mean connecting revenue, activation, product usage, and support. Agency reports might explain campaign performance and delivery risks. E-commerce reports could relate advertising, orders, inventory, fulfilment, returns, and repeat purchases.

AI could help connect patterns and answer questions, but it would remain an implementation detail. The promise is better decisions, not an “AI analyst.”

The system should recommend actions without executing consequential ones automatically. Its value would depend on whether people repeatedly discover something useful and make better decisions. Otherwise, it is only a prettier dashboard.

If you are a BI practitioner or run a SaaS company, agency, or e-commerce business, I would like to know: what decision did your dashboards fail to help you make recently?

Not a marketing - only idea
TL;DR: A concept for a reporting tool that connects business data and delivers recurring, decision-ready briefs. Instead of showing only charts, it explains what changed, why it may have changed, what evidence supports the conclusion, and what action to take next. Users can ask follow-up questions, while uncertainty and alternative explanations remain visible. The idea still needs validation with SaaS teams, agencies, and e-commerce operators.

Businesses rarely lack data. A founder can open analytics, payments, advertising, and support tools and still not know what deserves attention first.

I have been thinking about a reporting tool built around that gap. It would connect to a company's existing sources and produce a report every day, week, or month. Instead of stopping at charts, it would explain what changed, suggest why, recommend concrete actions, and accept follow-up questions.

Every important conclusion would show its source, metric, comparison period, and confidence. The report would separate measured facts from interpretations, missing information, and alternative explanations. Recommendations would include their expected effect, cost, downside, and a safe way to test them.

The same engine could support different report structures for SaaS teams, agencies, and e-commerce operators. I would still begin with one group, probably small SaaS teams, rather than build a shallow tool for everyone.

For SaaS, that could mean connecting revenue, activation, product usage, and support. Agency reports might explain campaign performance and delivery risks. E-commerce reports could relate advertising, orders, inventory, fulfilment, returns, and repeat purchases.

AI could help connect patterns and answer questions, but it would remain an implementation detail. The promise is better decisions, not an “AI analyst.”

The system should recommend actions without executing consequential ones automatically. Its value would depend on whether people repeatedly discover something useful and make better decisions. Otherwise, it is only a prettier dashboard.

If you are a BI practitioner or run a SaaS company, agency, or e-commerce business, I would like to know: what decision did your dashboards fail to help you make recently?

I have been building Promptyx as a prompt management tool, mainly focused on storing prompts, versioning them, testing across models, and seeing what actually performs better.

But the more I work with LLMs, the more I feel that a prompt alone is rarely the full system.

A good result often depends on things around the prompt:

• system instructions
• attached knowledge or files
• retrieved context
• model settings
• tool access
• output format
• conversation history
• the specific task or user state

You can have a well-written prompt and still get weak results if the model has the wrong context, too much context, or no clear instructions about what to prioritize.

That is making me think Promptyx should eventually expand beyond "prompt engineering" into a broader workspace for context engineering and AI workflow design.

Not in a way where it becomes another complicated platform with endless configuration. I still want it to stay simple for someone who just wants to save and improve a useful prompt. But for more serious workflows, it seems important to track the full setup that made a result work.

For example, instead of saving only:

>

you may need to save:

• the prompt version
• the model and parameters
• the brand voice context
• example outputs
• source documents used
• evaluation criteria
• the workflow steps before and after it

At that point, you are not really managing prompts anymore. You are managing repeatable AI behavior.

I am curious how other people see this.

Do you think prompt management is still a useful category by itself, or will it naturally become context/workflow management over time? And if you build with LLMs regularly, what do you currently save besides the prompt itself?

I have been building Promptyx as a prompt management tool, mainly focused on storing prompts, versioning them, testing across models, and seeing what actually performs better.

But the more I work with LLMs, the more I feel that a prompt alone is rarely the full system.

A good result often depends on things around the prompt (content and harness):

• system instructions
• attached knowledge or files
• retrieved context
• model settings
• tool access
• output format
• conversation history
• the specific task or user state

You can have a well-written prompt and still get weak results if the model has the wrong context, too much context, or no clear instructions about what to prioritize.

That is making me think Promptyx should eventually expand beyond "prompt engineering" into a broader workspace for context engineering and AI workflow design.

Not in a way where it becomes another complicated platform with endless configuration. I still want it to stay simple for someone who just wants to save and improve a useful prompt. But for more serious workflows, it seems important to track the full setup that made a result work.

For example, instead of saving only:

>

you may need to save:

• the prompt version
• the model and parameters
• the brand voice context
• example outputs
• source documents used
• evaluation criteria
• the workflow steps before and after it

At that point, you are not really managing prompts anymore. You are managing repeatable AI behavior.

Do you think this would be good? Will you use something like this? Does a solution already exist?
Should Promptyx also expand into Graph Engineering too? What do you guys say?