elawler24

7 ms·

61.
Show HN: Warehouse OpenAI requests to your own database(usevelvet.com)
18 points by elawler24 2y ago|6 comments
62.
by elawler24 2y ago
Studying existing component systems is a great starting point. You have to imitate and respect the systems that already exist before you can innovate on new things from scratch. I'd recommend starting by reading, building with, and imi
63.
by elawler24 2y ago
I always thought I was bad at math. Then I decided to learn it again from the ground up when studying for the GMAT. I hired a tutor who completely re-taught me the basics and got me excited about it for the first time. I was amazed by how q
64.
by elawler24 2y ago
I think it largely depends on the culture of the org (maybe share stage, eng vs sales led, what kind of product). It sounds to me like you don't have buy-in from engineering or they're not engaged with your process. What if you as
65.
by elawler24 2y ago
PMs are always tasked with implementing a feedback loop to improve features after launch. With LLMs, it's faster than ever before. You have a new production dependency that reasons and interacts with end users. In this article, I share
66.
How to optimize LLM features after launch(usevelvet.com)
2 points by elawler24 2y ago|1 comments
67.
by elawler24 2y ago
Velvet (usevelvet.com) | Founding engineer | Remote (EST) | Full-time or contract Velvet is a gateway for warehousing LLM requests, responses, and parameters. Our platform helps engineers analyze, optimize, and fine-tune their AI features.
68.
by elawler24 2y ago
Agree, temp 0 would be interesting to compare for this usecase where there's a clear right and wrong answer based on historical data. We experimented with temperature for our AI SQL editor and found .3 to be ideal so it can still self
69.
by elawler24 2y ago
Late to this conversation, but I'm a designer and very excited about these new AI features. They address tedious work that most designers don't want to deal with. It reminds me of why Figma won love from the design community in th
70.
by elawler24 2y ago
Agreed, you should make a decision on who is the CEO. Eventually this will become clear anyways, so just deal with the awkward situation now. When (if) you go out to raise money, the CEO should be the person doing outreach, pitching the vis
71.
Tactics for going beyond 0-1 with LLMs?(usevelvet.com)
1 points by elawler24 2y ago|1 comments
72.
by elawler24 2y ago
I’ve spent my career as a founder / designer / PM. Getting past 0-1 to product-market fit is really rare with any product. When you introduce LLMs, you’re adding non-deterministic elements to your experiments. Sharing some of our
73.
We use OpenAI to automate our data copilot(usevelvet.com)
3 points by elawler24 2y ago|0 comments
74.
by elawler24 2y ago
VC incentives are the real problem here. I was a mentor at another bootcamp and also the founder of a high-skill developer marketplace in 2017. A valuation of $1B for this business is crazy. Investors were simply underwriting students payin
75.
by elawler24 3y ago
Bessemer's second lesson learned on vertical software in the past 10 years: "The best vertical software companies build a layer cake of new products that drive continuous growth" - https://www.bvp.com/atlas&#x
76.
by elawler24 3y ago
The real problem is for non-technical users who don't know about all the settings, and view their notifications as "tasks" they must complete. This isn't dissimilar from unsubscribing from emails. The two solution propos
77.
by elawler24 3y ago
You also have to buy all apps through Apple's app store to natively download to a device. The Digital Markets Act addressed something similar, requesting that developers can sell through alternate marketplaces. Apple came back with a p
78.
by elawler24 3y ago
I've worked as a founder or employee a variety of startups from conception seed to series C to pre-IPO to post-IPO. This has been true in every case - urgency is all that matters. The startup or project dies as soon as it becomes busin
79.
by elawler24 3y ago
This is what Keynes predicted might happen - https://onlinelibrary.wiley.com/doi/full/10.1111/ecca.12439#... That article notes most people can't make enough money from 15 hours of work to save up for re
80.
by elawler24 3y ago
For me, mornings are my most productive time - I can get a whole day's work done from 6-9 if I really focus. So I use it for whatever I need more time on. Right now I'm trying to build a 0-1 startup, so it's almost always abo
81.
by elawler24 3y ago
Velvet | Founding engineer | Remote-first US EST | Full-time Velvet is a database service to unify your data and ship personalized features with a simple API. Our mission is to make every product more human (usevelvet.com). Tech stack: Nex
82.
by elawler24 3y ago
Velvet | Remote (NYC/EST preferred) | Full-time | Fullstack software engineer | $100-200K USD + equity Our database service captures real-time events from any source. Product teams use it to unify customer data and identify segments wi
83.
by elawler24 4y ago
Velvet | Founding Engineer | NYC / Remote (US) Hey all - I'm the founder of Velvet (velvet.cash), looking to bring on founding engineer(s). Velvet enables seamless token-based registration, payments, and access through a no-code p