dazzeloid

very cool. do you think of this primarily for engineering work or more generally collaborating with AI?

mid-level output, clearly had more capability than his output suggested from his ideas and some particularly strong contributions

trust. he was not forthcoming when confronted with the "this other company says you are full-time and just went to their offsite - is that true?"

funny thing was he had other places on his linkedin under "active employment" but we never really dug into it (until we learned he was full-time there) because he just seemed like the kind of person who wouldn't keep his LinkedIn up to date.

He was a solid middle of the pack contributor on net, but it was clear he was way stronger than his net output just from interactions with him (he had great products ideas, clear technical understanding of many areas, etc).

he's a really talented engineer, crushed our interviews. the funny thing was that he actually had multiple companies on his linkedin at the same time, including ours. we just thought they must have been internships or something and he never updated them (he felt a bit chaotic). but then it turned out he was working at all of them simultaneously.

worked for us for almost a year and did a solid job (we also let him go when we discovered the multiple jobs)

Cosign, Spenser and I are aligned on this :)

<3 thank you!

I both think this comment is spot on (acquisition theater / cope-core is very real) and what I said was true :)

Obviously, the better financial outcome is to grow huge independently and go public, etc, but there are a lot of good outcomes that are not that.

Command AI co-founder here. What I'll add to Spenser's nice comment is that there was this strange tension when we were deciding to do this between: (1) Feeling like the only right outcome for a growing startup with runway is to keep going and try to build a massive independent company, (2) Realizing that we can probably grow faster as part of Amplitude (obv with less upside capture)

Obviously if your growth is so epic and you're a top 1% company, the choice is obvious. But there are a lot of companies that are doing well outside of that group that I don't think allow themselves to consider the acquisition route. We were like that until Amplitude reached out and we got good counsel to seriously consider their offer.

I'm going to be following very closely how Arc approaches monetization. Don't think they've talked much about this yet. Right now they are towing the line between art project and business extremely well so far, but I wonder how this will work at scale once they're through the honeymoon early adoption phase.

Love the concept of prompt velocity. Although it doesn't capture the initial quality of the prompt or whether the changes are effective.

A frustration I have with langchain is the number of primitives and terms -- insane how long this blog post needs to be to explain everything.

It is completely mind blowing to me that the first version of Gmail was built in a day. YC mindset before YC.

Here for the "I could build this in a weekend" comment so I can bookmark for the IPO :)

I'm not super familiar with chatbase (which looks awesome in its own right). Some things that I haven't seen before: -Semantic search added -Ability to open source docs in the widget -Recommendation sets -Personalized suggested question

Also we're pretty focused on embedded use cases and looks like chatbase is more generically usable (e.g. on discord, via API).

But honestly, I expect these products (including our own) to grow so much over the next few months than this answer could totally change.

Right now it's hitting OpenAI, but we want to try out other LLMs in the future.

it's a main quest now!!

Wrote about a few things here: https://www.commandbar.com/blog/why-we-built-helphub

There are so many low-hanging fruit: -Edit the system prompt -Shape common answers -Give bot access to actions so it can act like ChatGPT with plugins

Not currently but would be pretty easy with our API. Just a different UI.

Lotties :sweat

I guess my concern is being a system of adjusted record on top of a system of record. Wouldn't it be a cluster if changes are being made to docs in CommandBar but not [Zendesk]?

That said maybe there's room to store bot-specific stuff in CB. For example, tagging passages with "exclude this from training data" if they're causing bad answers for some reason.

Wahhh. Strava example?

For CMS's we built custom integrations (not just a generic crawler) that strips out obviously sensitive info like internal notes to support people.

Nothing revolutionary to report on the prompt injection stuff. Most people using HH are using it for public documentation so there really isn't any info in the source content that couldn't be surfaced in an answer.

If we're doing our job right then HH should only be answering based on its source content (and not background knowledge). So bad answers coming from incorrect source content would need to be corrected in the source content. Citations should help with this, e.g. as a human admit, review answer -> notice it's bad -> click citation -> locate incorrect part -> change it.

Also thinking about ways to ensure the bot answers common questions correctly while still being able to personalize responses. Working on something called "answer shaping" where an admin can write out a response and tag with the question it responds to. Then the bot would first check to see if the human question matches a cached question, and if so would prioritize using info in the cached answer in its response. Seems like this can give the bot freedom to personalize the answer but make sure it includes the right stuff.

Oops looking into it. Do you remember which bot you used and what question you asked?

Not yet but this has come up a lot so thinking about it. We probably don't want to build our own human ticketing system so the clearest path would be to have an entrypoint to kick the user over to something like intercom and maybe provide the context of the AI conversation as history in that interface. Not ideal to have two chat interfaces tho :/

Ideally we could be the UI layer for intercom, zendesk, etc. We already do that for docs search / exploration.

HN has slowed it down a bit but speed is definitely not where I want it to be. A lot of that comes from the openai side atm. Just turned on caching for the suggested questions so at least those are extremely fast.

That pricing is for our whole product, which includes more than just Helphub.

For HH only we have a free tier with 2k messages / mo and a $50 tier for 10k messages a month.

https://www.commandbar.com/pricing-helphub

I feel like a lot of companies put effort into their changelog in the early days when they're trying to cultivate their first users and know every user personally. But then I see them inevitably fall into the "bug fixes and performance improvements" laziness.

Would be a cool (but painful to do) analysis to look at whether this happens and at what scale companies throw in the towel.