robbiemitchell

This appears to be an excuse to rep your own AI startup.

I get a better first pass at code by asking it to write code at the level of a "staff level" or "principal" engineer.

For any task, whether code or a legal document, immediately asking "What can be done to make it better?" and/or "Are there any problems with this?" typically leads to improvement.

One helpful addition would be Requests Per Minute (RPM), which varies wildly and is critical for streaming use cases -- especially with Bedrock where the quota is account wide.

For training, would it be useful to stabilize the footage first?

some gimmick

"key differentiator" and not necessarily easy to pull off or pay for

It wasn't even set up for success at selling.

After years of raising 3 kids, you would think if I ask to add diapers to the cart, it would know something. But no, it would just go with whatever is the top recommended, or first in a search, or something like that. Nothing using the brand or most recent sizes we purchased.

There was no serious attempt to drive real commerce. Instead, Alexa became full of recommendation slots that PMs would battle over. "I set that timer for you. Do you want to try the Yoga skill?"

On the other hand, they have taken on messy problems and solved them well, but not using technology, and for no real financial gain. For example, if you ask for the score of the Tigers game, Alexa has to reconcile which "Tigers" sports team you mean among both your own geography and the worldwide teams, at all levels from worldwide to local, across all sports, might have had games of interest. People worked behind the scenes to manage this manually, tracking teams of interest and filling intent slots daily.

Asking even a top-notch LLM to output well formed JSON simply fails sometimes. And when you’re running LLMs at high volume in the background, you can’t use the best available until the last mile.

You work around it with post-processing and retries. But it’s still a bit brittle given how much stuff happens downstream without supervision.

Processing high volumes of unstructured data (text)… we’re using a STAG architecture.

- Generate targeted LLM micro summaries of every record (ticket, call, etc.) continually

- Use layers of regex, semantic embeddings, and scoring enrichments to identify report rows (pivots on aggregates) worth attention, running on a schedule

- Proactively explain each report row by identifying what’s unusual about it and LLM summarizing a subset of the microsummaries.

- Push the result to webhook

Lack of JSON schema restriction is a significant barrier to entry on hooking LLMs up to a multi step process.

Another is preventing LLMs from adding intro or conclusion text.

It is workable. It’s like $6/month.

The real problem is that most people don’t want to pay for things.

As a Zapier user looking to go with something private for new use cases, how does this compare to an on-prem n8n.io?

This is a huge bummer! I wonder why they removed the playlist station feature.

You can do this in Spotify now by creating a playlist and then playing the station driven from that playlist.

The context is a for-profit company. If a third-party using the API is doing so as a substitute for someone using the ad-supported site, the API will be priced to equal the revenue lost from using the site. If the people switching from UI to API are more valuable than usual, the API will cost more than usual.

If the API is being used to support something new -- i.e., not a replacement for a user on the site, but as a new revenue source -- it will be priced to maximize overall API revenue, or overall revenue as an indirect result of growth supported by the API. (Maybe it's one massive customer using it or 100,000 freemium developers.)

The idea that a company "must" do something doesn't make any sense. And the assertion that the company is harming itself is ultimately something the Board, acting on behalf of shareholders, is responsible for.

(FWIW, I agree Reddit is shooting itself in the foot.)

That’s pretty much how insurance works. This is a 20% fee for DDoS insurance.

The letter is a demand for action that assumes a conflict exists and the client is correct. It is standard. The recipient can acknowledge and agree or acknowledge and disagree.

How does this fit in with Bard? I see no mention of Jack Krawczyk here, who is listed as its product lead.

Same! I had a directory of info coded horribly by some CRM into HTML DIVs (not a table) and copy/pasting it into a text file resulted in a single column of names, contact info, kids, grades, etc.

I asked ChatGPT to reformat it into a CSV, noted the useful breaks, requested some transformations and filtering, and specified a delimiter. After 5 minutes of experimentation, it worked like a charm. Absolutely amazing.

Executives has larger, permanent desks (or offices) that don’t get bothered, and never get kicked out of a meeting room — but their assistants will kick you out of yours. The problem is invisible to them.

This is a project I've always dreamed of building! Already pretty excited about some of the recommendations I'm getting in some niche genres.

Feature request: allow me to auth Spotify and click a button next to each track to add a track to a "Maroofy Recs" playlist

You can connect code steps to other actions. For example, every Python code step has a reserved `output` variable which allows you to pass a list out to the next steps. It gets passed as stringified JSON: if it contains dicts, those values are made available automatically as inputs to other steps. You can also pass more complex, nested JSON and just json.loads it later.

The first example on the homepage is "Create an email drip feed campaign in under 2 minutes"

This is not an advanced use case to be developed in house and owned by developers. This is bread and butter marketing or product engagement email flows using anything from Mailchimp to customer.io to HubSpot.

The developer side would involve capturing the identities and events from your system and sending them (typically via something like Segment) to the downstream tools, where business users will manage those campaigns: flows, templates, content, integration with other flows, and reporting.

The next example is "Sync GitHub issues to Linear". Again, this is a fairly simple Zapier use case, probably using built-in integrations, or falling back to Python if needed. Zapier would store the credentials to both security and use a trigger/action flow.

I can see trigger.dev being more useful for things like:

- Schedule-based tasks, or super high-volume tasks. (These are expensive on platforms)

- That are driven primarily by code, not pre-built integrations

- Using private data (such as authorization tokens) you don't want to expose in plain text

Given there is undoubtedly a market of developers who want to bring things back into their standard codebase and code release practices, I suggest targeting the examples to situations to those more typically owned by developers.

Speaking as a tech-forward "business user" who uses Zapier _a lot_: a key benefit of Zapier is that it enables business users to work independently. Moving triggers and actions behind "developer-friendly" (i.e., developer-required) tooling caps how fast an organization can move on anything outside the critical path for product and strategy.

- Core transactional email flow? Great.

- Alerting Slack for non-critical activities elsewhere? Ehh...

The important point here is that this puts a big decision in front of people before they get started, rather than treating this as something more like a post-signup engagement vehicle.

For general adoption, every ounce of friction has to be removed from signup.

Furthermore, reading the docs here[1]:

Before you sign up for a service, it is important to understand its policies and terms of use. A Mastodon website will usually have its policies listed on the /about/more page, which can be found by clicking “learn more” on the landing page while not logged in to that website.

Every server can have different terms?

Mastodon allows website administrators to set one of three different signup modes

:facepalm:

[1] https://docs.joinmastodon.org/user/signup/

The first thing you have to do when you sign up is choose a server

It doesn't hugely matter which one you are on because you will be able to follow users on all the others anyway

The server you choose becomes part of your user name

Unlike Twitter, Mastodon won't suggest followers you may be interested in

OK, so this shouldn't be the first step unless the "server" is a key part of the experience or has other benefits. A company laser focused on signups, like Twitter was at one point, would aim a growth team at this and remove all unnecessary signup steps to reduce friction. Steps like... first enter email, then auto-suggest a username, then suggest some interests (i.e., people) to follow, then (and only then) allow for the editing and selection of your username.

he had over 6,000 new joiners in 24 hours and had to pause registration

You don't turn off signups when things get too hot. You allow signup and then throttle the experience. Reach back out when things cool down and get people re-engaged. (This is difficult, but better than nothing.)

In any case, if Mastodon is a bit like email (as another comment suggested), what will make it "just work" is an upgraded experience to make it easy to sign up. Yes, you will be "picking your server" by selecting your upgraded experience -- such as AOL, Gmail, Yahoo, etc. -- but then they will also need to monetize it or track your usage for other reasons.

In fact, if it gets any traction, perhaps Google will spin up a Mastodon service as a companion to Gmail.

The US is not the only country raising interest rates. It's happening in developed countries worldwide.

https://www.reuters.com/markets/europe/central-banks-raise-r...

Figma turned ~$330M in investments into a $20B outcome and gave countless companies a better way to do their work. How is this a bad path for society?

Here's just a few seconds of searching re: increased mortality for non-vaccinated:

https://www.scientificamerican.com/article/how-to-compare-co...

https://www.cdc.gov/mmwr/volumes/71/wr/mm7104e2.htm

https://doh.wa.gov/sites/default/files/2022-02/421-010-Cases...

https://www.washingtonpost.com/health/2022/02/08/omicron-dea...

Legal, finance, HR, R&D, facilities, comms, support, endless vendors for all of the above…

And no, you can’t just automate support. Refunds are not the only issue.

Spaces, Communities, Topics, Creator tools, Safety controls

Interesting: as a casual Twitter user since 2008, I am completely unaware of Spaces, Communities, Topics, and Creator tools. Probably because I use Tweetdeck (desktop) and Tweetbot (mobile). I never see trending posts or ads, either.

I guess you can shell out a few hundred more for a proper drone.

But who flies it?

They do it because it is used to make footage (I own one).

But doesn’t that involve you flying the drone to make footage by looking at something else?

What’s interesting about this Snap drone is that it’s pitched as something that will just stare at you so you can make fancy selfies. I could be missing it, but I don’t think this is how most drones market themselves.