We arrived at a similar concept, though ours it the negation of this: we have a series of deterministic checks that can potentially add the `needs-human-review` tag to a PR. If that tag isn't present, you can merge your own PR; otherwise you need at least one human.
I think we're probably around 50/50 needing human review vs not.
Our rules for whether something needs a human review are currently mostly filepath-based: are you touching the api, or an important internal contract, or adding db migrations? Though I'd like to make it a bit smarter as we're definitely getting false positives.
I thought for sure you might be one of my coworkers based on this comment, we do pretty much the same thing with our “LLM Review” tool. It tries its best to determine what needs human eyes (mostly through file paths and the model’s own “discretion” for changes outside the “important” paths).
In San Francisco, there's a bit of a push right now to re-allow single-stair design ("Single Stair Reform") for certain buildings (eg up to 4 or up to 6 stories). Requiring two stairways makes sense from a fire safety perspective, but it's made it incredibly difficult to cost-effectively build medium-size buildings on oddly-shaped lots.
Thanks! Several people on our team are actually exactly like you: they don't really use videos themselves for learning, but see the utility for others, and enjoy the technical challenges around building a high-performant video format.
I can’t stand this Silicon Valley trend of having launch videos for features of 2 people sitting in front of a camera but it also shows the camera in the beginning to trick you into thinking “oh they do this so casually haha so fun, look how relatable they are”
I hate all of this - OpenAI, Claude, Cursor, Windsurf etc
It screams detached Silicon Valley tech company to me, and I will never watch a single of them
Indeed. In much the same vein, I simultaneously accept that:
1. I loathe video with the white-hot glow of a thousand angry suns, and loathe LLM slop-digests of things more still.
2. I spend tremendous amounts of time in situations where I can listen but not read, whether driving or on my bike, and this is a legitimately useful way to catch up on HN in those situations.
I hate this too, but I acknowledge some may find it useful.
This should not be judged as a way to read hn, where users reading, it's more of a demo for technical users of their core startup tech.
I still don't like it, I feel it's a very small script with like ffmpeg and an LLM HTTP call, but that same thing was said about dropbox. It's also like 2 years late, so the timing is off.
I'm often negative in the comments of people launching their startup ideas, but I think that's fair, I don't like the atmospheres where we are supposed to pep talk each other and tell ourselves that it's really cool.. and then their project dies while everyone pats them in the back, I think yc is explicitly about that type of nuanced feedback, this needs a pivot as-is.
I'll second that. I tried the trial for a few months then started paying money for it. At $100/year I've found it useful for modeling and gaming out different scenarios for retirement thresholds.
Nothing's certain, but if you like to play with data and you don't want to build the whole damn thing yourself, it's useful.
Still waiting for them to get support for options.
Well, a coworker recently rigged up eleven's api to a coffee machine sensor. It now sings, loudly, an original composition to the effect of "make a new pot you asshole" whenever it gets low. We've gotten country, symphonic metal, southern hip hop, smooth british singing, accusatory broadway musical, 20's-style jazz singer, and more - it's different every time.
While the advisability of this particular project is up for debate, I do think it's something that could only exist with an API like eleven's.
A friend recently pushed me to try out their coding platform, Devin, after I decided to move away from Cursor. I had the same reaction: "What, the con artists from like 2024?" But after some cajoling, I gave it a shot and was pleasantly surprised. I guess they learned their lessons, grew up, and are doing good work now, maybe?
Devin CLI is easily the worst-in-class coding harness I've ever been subjected to using, rife with bugs (up to and including dropping answers to the question tool), so I can't say I'm exactly inspired to try anything else the company produces.
Had the same experience on it when it was going by Windsurf. Had to wire up a skill hooked to terminal runs or else it would hang or freeze and never finish literally every single time
I see issues with other harnesses too but not with the regularity I was getting from this. And the one moat they had with the better UI for per-project multi-agent orch disappeared and now is standardized
Can't say I've tried the CLI. I've mostly focused on the cloud agents, which I was explicitly looking for. I compared against cursor's cloud agents, ampcode, and hoplite, and came out surprisingly enjoying devin.
I will say that the lack of parity between Devin cloud and Devin desktop is downright embarrassing. It's very clear that the latter is a thinly-reskinned Windsurf. A visually similar UI with vastly different capabilities. Definitely a black mark on the whole thing.
Not all that surprising. The original Devin really was just an early attempt at agentic coding before models were really even trained for it. Now that it's a well established pattern and we've figured out what works, I'm not surprised they've morphed into something reasonble.
When they launched Devin it was supposedly at the performance of an engineering intern. Friends who used it found the bad parts of an intern (tons of handholding, review required) but it didn’t learn from mistakes or add throughput.
I think the evolution of the harness and ability to preserve loop context outside the context window has made running these kinds of agentic experiences easier.
sorry so many buzzwords to say, the capabilities to do this kind of work are more accessible and easier to manage, so now it works!
Good to see, and agree they were severely overhyping their product back then.
I quite like the idea of a tablet I can keep in my pocket. If there was some sort of first-party attachable keyboard (a la the iPad magic keyboard), I'd be strongly considering getting this. In the age of AI and increasingly-mature cloud agents, I can get a lot of my work done from a setup like that.
I think we're probably around 50/50 needing human review vs not.
Our rules for whether something needs a human review are currently mostly filepath-based: are you touching the api, or an important internal contract, or adding db migrations? Though I'd like to make it a bit smarter as we're definitely getting false positives.
reply