HN user

patman81

91 karma
Posts0
Comments50
View on HN
No posts found.

Once you decide to "fire a customer", it may not be easy to get thru that process. It can be tricky if the cancellation started on the side of the software provider.

We run an enterprise saas business and at one point decided we can't work with a particular customer anymore. We informed the customer that we will terminate his service and cancel all his outstanding (unpaid) invoices.

Soon after that, the customer started to take legal action against our company. It was quite an ordeal for about 12 months. At least it helped us improve our contracts and legal processes going forward.

However, since than we are careful to vet new potential customers before we offer our software. If we feel the software isn't a good fit, we will tell the customer and work thru our concerns before signing a contract. Even if we may lose some potential customers in the process, it builds a more healthy and sustainable business.

"Avoid toxic customers." Is now part of our handbook.

Tesla Cybertruck 7 years ago

Based on the stated ranges of 250, 300 and 500miles (!), I would guess those battery sizes are 75, 100 and 150kWh. I hope the upcoming triple motor model S, will get that 150 battery pack next summer. That would mean 600miles of range for the new model S.

There was another Tesla software update announced as "almost done": a major update to Autopilot 2, with an all new neural net and what not.

This all new navigation system might include functionality to help, not only the human driver, but that new Autopilot as well.

At some point Autopilot and Navigation will have to merge. Right now, both systems operate completely separate.

But if those upcoming updates will allow Autopilot to follow the navigation system changing lanes thru highway intersections, than that would indeed be 'light-years ahead'.

On the Tesla earnings call last week, Elon Musk stated that a software update coming in 3-6 months will allow Tesla cars with Autopilot HW2 to drive fully autonomous. At least on highways.

He's predictions are always late, so it may not be ready in August. But his predictions seem to come thru in end. If he can deliver by the end of the year or even next year, he would still beat all other car manufactures schedules.

German highways are not ideal for a Tesla. The right lane is usually to slow, with numerous trucks going at about 90kph.

You can drive fast on the left lane, a Tesla goes up to 220kph super fast, but there is usually traffic and you frequenty have to slow down to 120kph for a while and than accelarate back to 220kph.

That frequent accelaration from 120kph to 220kph, is what kills range.

Ideally, if there is no traffic, you engage Autopilot and go with a constant 150kph for great range and good enough speed.

However you can't keep 150kph on the left lane, since BMWs and Audis will push you off when they come with 250kph or more.

Uploading of collected data is indeed deactivated by default on Tesla's cars in the EU. But you can opt-in by checking a box in the cars setting. You can also opt-in to share driving videos from the eight cameras sourounding the car, to help improve the selfdriving system.

Tesla Autopilot Hardware 2 is now up to par with version 1. (The most recent update increased the Autopilot highway speed to 150kph, same as version 1).

The next software update, scheduled for june, should bring automatic perpeticular parking, automatic windshield wipers and an improved "smother" autopilot.

Although perhaps disturbing for some, I enjoy the monthly software updates. Like a little surprise every month.

Segway robot 11 years ago

That may be a cool thing, but the site is almost impossible to read. At least on my iPad. It might be complex to program that robot, but navigating their site shouldn't be.

Introducing AWS WAF 11 years ago

Now the only pice missing are AWS SSL certificates, for a one stop shop on the whole security stack. Perhaps tomorrow at the AWS keynote. Fingers crossed.

For existing software providers, the book recommends a transition to an recurring revenue model ala SaaS. Short term business incentives and technical limitations in an existing code base, may make this a tuff sell to existing software providers.

But the long term advantages are well argued in the book (less support costs, higher long term revenue per customer) and perhaps worth the costs of transition.

The recurring revenue model is particularly advantages for products with a strong lock-in.

I would love to use this to share my Macbook screen on an iPad with a client. One screen seems to small for more than two people to share. Perhaps I could even connect multiple iPads for small presentations without a projector. All running in mirror mode.

The concept of "intrinsic blind spots" sounds like from a spy thriller. How can I very slightly alter my face to mess up face recognition software? The paper describes how to calculate "blind spots" that should hold for various different neural networks (even if they are trained on different data sets).