I think for an intern, just the fundamentals will put you above 90% of your peers. Everything else is gravy. Ask intelligent questions at the right time. What is the right time? That's where EQ kicks in.
The opportunity to vibe code and reinforce with something like claud code for example will poach any application that thinks they have the market share and charge whatever the hell they want. I am very excited for projects like this where you leverage open source for a very specific user and offer alternatives to SAAS. Keep iterating and solidy you security and make sure you get direct feedback from experienced CAD users, which I am not. good luck
I like your thought process of dual-capital startup model. However the time portion I believe may overlook the valued contributions of employees. More time doesn't necessarily mean more value, so I think employees will manipulate the system to get more profit-share. I am a big believer in increasing the cut of the pie through profit-share and not the current model of stock options but a separate profit-share compensation segmented for contributors that are not owners and not investors in the VC way. The hard part is measuring contribution that leads to value, which leads to profit. For example, with software companies, is it just who wrote the most code? Or the most valuable line of code that transforms an application from mundane to transformative. I don't know the answer, but profit-sharing of employees outside of stock options should be a higher priority, especially when break-even is reached. Meaning, quarterly net profit after tax is not just rewarded to shareholders or founders. I know people will say, " Well, you're getting a salary", tax code tells you fu and your salary, thank you for our 50% cut, now get back to work.
Do you have any latency criteria as to the measurements you are seeing to define what is acceptable latency or have some sort of deviations matrix to show users where their mcp activity is falling within. Like a feedback loop? Just a thought.
In business communication of every kind, voice, text, emails, paper sticky-notes, doesn't matter. Assume someone is sharing your information with someone that you don't know is sharing it. Communicate as if you have an audience at all times. Also, I am not a lawyer, but the error rates are too high for anything transcribed to stand up in court with a good lawyer, unless corroborated by human witnesses.
Any project that requires a local model should always be the way to go on first attempts and if the functionality is acceptable should stay with local models. Token burn is a serious problem and will ultimately lead developers to ask one question "Do I really need Opus xyz?" For most requirements of standard applications the answer is no. So using open-source llm models that are integrating in practical use-cases to create a value-add not for 'hey look I have AI in my app, sign up please.' Open source models are competing well and is the way to go for the majority of projects and mindsets do have to change and I see them changing this way rapidly. You don't have to host your open-source llm locally but host it with a 3rd party, it is cost-effective and the token burn is not a barrier.