kc10

Can you please expand on this?

The sensitivity to prompts and response quality are related to an agent's functionality, A2A is only addressing the communication aspects between agents and not the content within.

We built similar animation part of a product few years ago. The flow was something like - when a new user signs up and users our product for the first time and creates certain artifact, the confetti animation would be displayed. Product managers loved it and they would show it off to execs as playful, refreshing, etc. But later on after UX reviews, accessibility testing, the feature was ultimately removed from the product.

It was fun to present it in demos, but it can also be annoying to users.

Over the weekend, I saw a Model S with number plate "WOZ" in Oakridge mall, San Jose. I wondered if it is him and sure enough it was him driving the car to the supercharger stations.

I was pleasantly surprised to see him driving and trying to charge the car in public locations.

Congrats on the launch!!

I am previously the founder of a synthetic monitoring startup, devraven.io.

Just sharing my experience - monitoring is brutally competitive. From my conversations most large enterprises have very little synthetic monitoring, they use DDOG or other APM tools and do not want to try any new tools for few thousand dollar savings. And in a lot of cases they are comfortable with their custom test frameworks that use Selenium. Some are even worried that setting up synthetic monitoring will bring down their environment or trash their database with junk data ::sigh::

Most smaller companies we spoke to are not mature enough to have monitoring and did not have resources who can setup monitoring. They used to ask us for help to build tests for them. Asks for discounts on $29.99/mo price point were not uncommon.

After few months of operating the product, we did find few angels who were interested in investing in us (not the product). But in the end, we did not feel that we can make good use of investor money and provide a decent return to them, so we ended up backing out of the investment and chose to shutdown the product.

Only javascript is supported for building transformations and destinations. But you can use any npm library in your script, platform will automatically download it at runtime and executes your script.

Thank you.

Fair feedback. I definitely need to anchor myself to specific niche instead of saying it can work for anything and everything.

The entire backend is available via REST interface. I am looking to support finer API permissions via auth tokens.

But the APIs are accessible today with basic auth.

There are a couple of ways to address this:

1. Create a Source for each customer and send customer specific data to each source. Configured pipelines for each source will receive the event data and processes it.

2. Create one source with multiple pipelines connected to it. In each pipeline, you can drop processing the data based on any conditions.

#1 is preferred so you are sure each customer's data is only reaching the pipelines configured for that customer data.

Not sure I understood the use case. Are you looking into integrate into SN platform? You can add a pipeline to integrate with SN platform using OOTB rest api or add scripted rest end points for custom integrations.

A ServiceNow integration is on my radar. But if you can provide more details I can definitely try to address it when I build the app. Please feel free to log an issue in github if you can provide any details or send an email to me (email in my profile).

There is no throttling today based on outbound failures. But I think it is a pretty common use case for handling destination failures.

Something like - if x or x% events are failing in a pipeline, pause the pipeline for y minutes or until some one unpauses it.

I can look at adding it.

Thanks. It's not v1 yet.

It's about 3 months in the making and needs improvements around ops, tests. I am hoping to make progress with adding tests in the coming days and provide monitoring capabilities.

It is self-hostable today. Just clone the repository, update the .env file and run "docker compose up" to get the instance running.

Thanks.

I think there are a number of platforms for building workflows and integrations today. Some of the platforms serving specific niche (e.g. RudderStack chose to serve CDP, airplane for building internal dev tools). But in the end all of them act on a trigger to transform inbound event data and send it to a destination end point.

Windmill looks cool with support for multiple languages, no-code editor and all the OOTB integrations.

cptn is still evolving - it needs to have more integrations, UX flows need to be improved, it has to get better from Ops perspective.

I haven't used n8n before, but n8n looks much more mature, feature rich and no code capabilities. You can build any workflow in cptn as well, but most of the workflow is built via scripting.

Licensing "seems" to be restrictive if you want to use for commercial purposes (I am not quite sure about Sustainable Use License).