Ask HN: Do you automate most of your routine tasks?
27 comments
I get bored easily with repetitive work. Generally about the third time I find myself doing something, is on the short-list to get some level of automation applied to it.
BCC is in maintenance mode, which means that my only interaction with the site is customer support and moving money around once a week to cover bills and my paycheck. Everything else is automated or outsourced. Bookkeeping for sales, for example, is 99.96% automatic. (Chargeback? Bah, you always miss one case...) Bookkeeping for expenses is about 90 ~ 95% static from month to month modulo dates and numbers, so I have my VA do it for me. Order fulfillment is automatic. Routine server administration is automatic. Customer services problems are, to the extent practical, automatic.
See generally: http://www.kalzumeus.com/2010/03/20/running-a-software-busin...
I should probably find a week or so to give BCC in the near future to make the ghost ship a wee bit less rudderless. (Like restarting content creation for it -- that was 99% automatic when I had a freelancer actively working on it, but she quit and I haven't bothered replacing her for... far too long now, it seems.)
AR and my consulting work are both less mature and, by nature, require more hands on attention. Still, you could reasonably expect that as I get more of a hang on what the business needs and where I add the most value to it, more work will get pushed off onto systems or processes.
BCC is in maintenance mode, which means that my only interaction with the site is customer support and moving money around once a week to cover bills and my paycheck. Everything else is automated or outsourced. Bookkeeping for sales, for example, is 99.96% automatic. (Chargeback? Bah, you always miss one case...) Bookkeeping for expenses is about 90 ~ 95% static from month to month modulo dates and numbers, so I have my VA do it for me. Order fulfillment is automatic. Routine server administration is automatic. Customer services problems are, to the extent practical, automatic.
See generally: http://www.kalzumeus.com/2010/03/20/running-a-software-busin...
I should probably find a week or so to give BCC in the near future to make the ghost ship a wee bit less rudderless. (Like restarting content creation for it -- that was 99% automatic when I had a freelancer actively working on it, but she quit and I haven't bothered replacing her for... far too long now, it seems.)
AR and my consulting work are both less mature and, by nature, require more hands on attention. Still, you could reasonably expect that as I get more of a hang on what the business needs and where I add the most value to it, more work will get pushed off onto systems or processes.
I use Moonshine for automating server configuration and Rails deployment, git aliases for shortening commonly used git commands, and zsh aliases for shortening things like "cd ~/wiki; soywiki" to "wiki".
I also spent a few bucks on an 8 GB RAM upgrade so I could just set MySQL, Postgres, MongoDB, and Redis to run on startup and not have to stop and start them whenever I am working on different apps, and I still have plenty of RAM left over for Photoshop, Starcraft II, etc.
I also spend time practicing vim tricks for things that commonly require a bunch of keystrokes, since practicing shortcuts until they become automatic is effectively the same thing as automating the task itself.
There are also a lot of things I don't automate, but it basically boils down to whether or not I use them enough to make it annoying enough to make me want to automate them. For example, I know that it's possible to set up a Rails template that creates a new Rails app with all of the gems, plugins, and configuration that you commonly use, but since I only create new Rails apps once every two weeks or so it's not enough of a pain that I want to spend the time creating the template.
I also spent a few bucks on an 8 GB RAM upgrade so I could just set MySQL, Postgres, MongoDB, and Redis to run on startup and not have to stop and start them whenever I am working on different apps, and I still have plenty of RAM left over for Photoshop, Starcraft II, etc.
I also spend time practicing vim tricks for things that commonly require a bunch of keystrokes, since practicing shortcuts until they become automatic is effectively the same thing as automating the task itself.
There are also a lot of things I don't automate, but it basically boils down to whether or not I use them enough to make it annoying enough to make me want to automate them. For example, I know that it's possible to set up a Rails template that creates a new Rails app with all of the gems, plugins, and configuration that you commonly use, but since I only create new Rails apps once every two weeks or so it's not enough of a pain that I want to spend the time creating the template.
I can't automate most of my routine tasks, and I'm not sure I would if I could: eating, sleeping, taking the bike to work, talking to people, ...
I do automate many of my computer related tasks. Since I'm a fan of the command line, small perl scripts, shell scripts and aliases do a lot of "work" for me.
I do automate many of my computer related tasks. Since I'm a fan of the command line, small perl scripts, shell scripts and aliases do a lot of "work" for me.
How long do you plan on being there? How much time will you spend on these tasks in the long run?
If the cost of performing these small tasks over some specified period of time outweighs the cost automating them, and they /are/ in fact easily automated, then I think the answer here is clear: AUTOMATE.
Tasks I've automated:
* Server/OS/software deployment
* Daily calculation of linear regression of email queues on my mail servers
* Webcam snapshots and automatic uploads - http://www.charleshooper.net/blog/automating-webcam-snapshot...
* Most tasks requiring data entry - http://news.ycombinator.com/item?id=2359287
* And lots and lots of other things
If the cost of performing these small tasks over some specified period of time outweighs the cost automating them, and they /are/ in fact easily automated, then I think the answer here is clear: AUTOMATE.
Tasks I've automated:
* Server/OS/software deployment
* Daily calculation of linear regression of email queues on my mail servers
* Webcam snapshots and automatic uploads - http://www.charleshooper.net/blog/automating-webcam-snapshot...
* Most tasks requiring data entry - http://news.ycombinator.com/item?id=2359287
* And lots and lots of other things
Depends on the definition of "some time" compared to the 2 weeks (solid?) it would take the automate. And how long you expect to continue doing said tasks.
That said, I would always err on the side of automating - at least some of the tasks. Take the low hanging fruit and tackle those first.
For example, I send out a weekly report containing a bunch of metrics sourced from our website monitoring tool. It took me a couple of days to automate that, but it saves me a good hour every week (and means I don't make mistakes copying and pasting figures which I used to with alarming regularity). It also means that my Monday mornings can start immediately without pounding through a repetitive task.
That said, I would always err on the side of automating - at least some of the tasks. Take the low hanging fruit and tackle those first.
For example, I send out a weekly report containing a bunch of metrics sourced from our website monitoring tool. It took me a couple of days to automate that, but it saves me a good hour every week (and means I don't make mistakes copying and pasting figures which I used to with alarming regularity). It also means that my Monday mornings can start immediately without pounding through a repetitive task.
Exactly. Unless you have what "some time" is, how can you possibly justify something. It all boils down to ROI and risk (including mental sanity).
practice, practice, practice
The more you automate routine tasks, the faster you'll become. Don't just do a calculation of:
# of times to do a menial task * amount of time to do menial task <> amount of time to automate
That oversimplifies and doesn't take into account how much faster you'll be at scripting etc. in 20 years if you take every opportunity to hone your skills. Think of it like touch typing; might slow you down at first, but you're going to more than make up for it by the end of your career.
The more you automate routine tasks, the faster you'll become. Don't just do a calculation of:
# of times to do a menial task * amount of time to do menial task <> amount of time to automate
That oversimplifies and doesn't take into account how much faster you'll be at scripting etc. in 20 years if you take every opportunity to hone your skills. Think of it like touch typing; might slow you down at first, but you're going to more than make up for it by the end of your career.
Yes I do. For example, I have a little script to install Drupal modules, which is nothing more than downloading the tarball from the Drupal website, and untarring it in the proper directory.
Likewise, I've got scripts to extract strings to translate (with gettext()) using xgettext from various project directories, and merging it with older, already existing translations; and for installing edited translations afterwards (with msgfmt).
I also have a script to upload files that were changed locally to a remote server.
No, it didn't take me 2 weeks to code them.
Most of the time I just do the task once, and store the commands in a file. Next I edit that file replacing values with script arguments. Tada! Instant script. For lists of values that depend on the project, I create presets (one argument determines what set of values to use).
Likewise, I've got scripts to extract strings to translate (with gettext()) using xgettext from various project directories, and merging it with older, already existing translations; and for installing edited translations afterwards (with msgfmt).
I also have a script to upload files that were changed locally to a remote server.
No, it didn't take me 2 weeks to code them.
Most of the time I just do the task once, and store the commands in a file. Next I edit that file replacing values with script arguments. Tada! Instant script. For lists of values that depend on the project, I create presets (one argument determines what set of values to use).
You should check out drush - "drush is a command line shell and scripting interface for Drupal":
http://drupal.org/project/drush
http://drupal.org/project/drush
I automate as much as I can. Sometimes I'll even "automate" a one-off task, simply because writing steps down in a shell script makes it easier to review what is about to happen and avoid trashing, say, a big chunk of my digital photo album... :-)
Automation doesn't just save time; it also avoids many mistakes caused by fat-fingering things or forgetting step 6 of an 8 stage process.
Also, once you have automated something, you've (at least potentially) created a tool you can share with someone else, which is good for teamwork, delegation and continuity in a work-place.
Automation doesn't just save time; it also avoids many mistakes caused by fat-fingering things or forgetting step 6 of an 8 stage process.
Also, once you have automated something, you've (at least potentially) created a tool you can share with someone else, which is good for teamwork, delegation and continuity in a work-place.
I would, and I do. Coding is much more fun than doing repetitive tasks, and once it's done you are just so much quicker in completing that particular task. This is true even for very simple tasks. For example I have to copy some files to a network share once a month or so. It does not take much time to do manually, but I had to open a file explorer, search the source folder, ctrl+a, ctrl+c, search the destination folder and ctrl+v.
Now it's just a doubleclick on a .cmd file on my desktop.
Now it's just a doubleclick on a .cmd file on my desktop.
I try to automate as much as I can. I also use these little programs/scripts as opportunities to learn something new. Examples:
1. We had a support portal at work which I had to check regularly during the day. I wrote a small program that monitors the relevant page and pops up a taskbar notification when something new arrives. Used the opportunity to learn about taskbar notifications and a little WPF.
2. I sometimes get database backups from customers and have to restore them on my machine, which means opening SQL Management studio, clicking and right clicking a bunch of things. Then I need to add myself as user, change a couple of settings before I can use the database for anything. I wrote a small utility so I can right click on any .bak file in explorer, choose Restore database and update the relevant tables so the database is ready to use right away with the connection string on the clipboard. Also used it as an opportunity to learn about windows 7 taskbar progress bars.
Basically I like doing small projects, but I never finish them unless they really do something. So, automating stuff I do makes much more sense than learning by following a tutorial that builds something useless.
1. We had a support portal at work which I had to check regularly during the day. I wrote a small program that monitors the relevant page and pops up a taskbar notification when something new arrives. Used the opportunity to learn about taskbar notifications and a little WPF.
2. I sometimes get database backups from customers and have to restore them on my machine, which means opening SQL Management studio, clicking and right clicking a bunch of things. Then I need to add myself as user, change a couple of settings before I can use the database for anything. I wrote a small utility so I can right click on any .bak file in explorer, choose Restore database and update the relevant tables so the database is ready to use right away with the connection string on the clipboard. Also used it as an opportunity to learn about windows 7 taskbar progress bars.
Basically I like doing small projects, but I never finish them unless they really do something. So, automating stuff I do makes much more sense than learning by following a tutorial that builds something useless.
I used to, until I spend a whole day on a vim script that, realistically, only would save me significant amounts of time by the time the work I was automating would bring in 1000's of dollars a day.
Nowadays I only do it after it's been proven (or I'm 100% certain up front) that a certain task will come up more often, and that letting someone else do it is not feasible or cheaper.
Nowadays I only do it after it's been proven (or I'm 100% certain up front) that a certain task will come up more often, and that letting someone else do it is not feasible or cheaper.
I have started automating a lot of the smaller, repetetive tasks. It doesn't save a huge amount of time, just removes small annoyances and friction. One example is that I'm new to OSX, so I was playing with Automator and have used it to create a command which brings up my entire development environment. This saves only 1.5 to 2 minutes each time, but it removes that friction when I get home from work I can just put my laptop on the sofa and have the dev environment load while I make a drink.
At work we have to create performance reports monthly for a particular client. The procedure has always been to do it manually, which takes around 4 to 5 hours. Writing the code to automate this in C# took less than 4 hours and now the reports take 2 to 3 minutes each month. Time well spent, money definitely saved.
At work we have to create performance reports monthly for a particular client. The procedure has always been to do it manually, which takes around 4 to 5 hours. Writing the code to automate this in C# took less than 4 hours and now the reports take 2 to 3 minutes each month. Time well spent, money definitely saved.
"We will encourage you to develop the three great virtues of a programmer: laziness, impatience, and hubris."
-- Larry Wall
Task automation encompasses all three virtues:
Laziness - "I don't want to do this task every (other) day"
Impatience - "I don't want to spend time doing this task"
Hubris - "Of course this will work on the first try"
-- Larry Wall
Task automation encompasses all three virtues:
Laziness - "I don't want to do this task every (other) day"
Impatience - "I don't want to spend time doing this task"
Hubris - "Of course this will work on the first try"
I automate all the small, repetitive tasks I can. Even when I'm heavy into coding (where I'm working with JS/php/db all day long every day), I'll have a script which opens my terminal and editors and the db etc. so that my day starts with a ./startEd.sh command.
Right now I'm setting up a staging server for my new webapp (I'm frantically working towards a private beta). This is new learning for me- and I'm writing a script for everything I do (that I can). This not only sets up a way for me to do setup/config tasks automatically, so I don't make typos, etc.), but also documents what I'm doing. I hadn't considered that aspect of it before I started writing these scripts.
Right now I'm setting up a staging server for my new webapp (I'm frantically working towards a private beta). This is new learning for me- and I'm writing a script for everything I do (that I can). This not only sets up a way for me to do setup/config tasks automatically, so I don't make typos, etc.), but also documents what I'm doing. I hadn't considered that aspect of it before I started writing these scripts.
I'd automate everything if given enough time. To me it's the essence of running an efficient business (and indeed scaling). Some examples:
1) I automated the creation of Google Adwords CPC keywords for my ecommerce website. Based on attributes of the product, the algorithm creates 500 unique keyword combinations. Over the coming days I will automate the creation of entire campaigns.
2) I automated gathering information about 200 universities by hiring a smart virtual assistant.
3) I automate my customer support emails by using canned responses.
4) I automate the running of my unit tests so that they run whenever I save relevant files.
5)I continue to automate my code base by writing new layers of abstraction which call lower layers intelligently.
1) I automated the creation of Google Adwords CPC keywords for my ecommerce website. Based on attributes of the product, the algorithm creates 500 unique keyword combinations. Over the coming days I will automate the creation of entire campaigns.
2) I automated gathering information about 200 universities by hiring a smart virtual assistant.
3) I automate my customer support emails by using canned responses.
4) I automate the running of my unit tests so that they run whenever I save relevant files.
5)I continue to automate my code base by writing new layers of abstraction which call lower layers intelligently.
I would love you know how you automate the creation of Adwords keywords.
I try to automate tasks in a hierarchical, re-usable fashion, based on log file output, and always in a pragmatic, least-cost approach. No-one is paying me to make my life easier.
For example, I'll find myself performing tasks X, Y, and Z quite a lot. I'll find out that I have 20 minutes free and I'll figure out the most common denominator of the tasks, let's call it A, and then implement it. I'll quickly re-write the high level processes X, Y, Z and to use A and then move on with my life. The up-shot is that future tasks B, C, D, E, F can all use the now-battle-hardened procedure A, and maybe others in my company will find A useful.
For example, I'll find myself performing tasks X, Y, and Z quite a lot. I'll find out that I have 20 minutes free and I'll figure out the most common denominator of the tasks, let's call it A, and then implement it. I'll quickly re-write the high level processes X, Y, Z and to use A and then move on with my life. The up-shot is that future tasks B, C, D, E, F can all use the now-battle-hardened procedure A, and maybe others in my company will find A useful.
Two weeks are quite some time for an automation. But depending on how much time you spend every day, you might want to go for it. It's not only about saving time daily, it's even more about scaling things up.
If the process is automated, it should be quite easy to use modified versions for similar problems. And you can quickly run it again and again without having to worry about making mistakes.
If the process is automated, it should be quite easy to use modified versions for similar problems. And you can quickly run it again and again without having to worry about making mistakes.
Yes, server tasks, deployment tasks. My daily routine 'feels' automated to a large extent and I'm still trying to optimize for output.
When I say my daily routine feels automated I mean, when I have two boring tasks I usually combine them to make them less inane. So an automata type scenario comes into play. I don't have to 'think' about the two crappy but necessary tasks.
When I say my daily routine feels automated I mean, when I have two boring tasks I usually combine them to make them less inane. So an automata type scenario comes into play. I don't have to 'think' about the two crappy but necessary tasks.
While I automate some computer related tasks, I can't automate house chores, working out, etc.
So I wrote an app to remind me: http://i.imgur.com/exzb0.png
I find this sort of thing most useful if you have chores that you want to do regularly, but not on a specific day (i.e. clean every Tuesday).
So I wrote an app to remind me: http://i.imgur.com/exzb0.png
I find this sort of thing most useful if you have chores that you want to do regularly, but not on a specific day (i.e. clean every Tuesday).
Using Automator on my mac I have created a few simple folder workflows. For example I have one that everything I drag into it gets printed. This seems simple but its great when you constantly get a few emails a day with multiple attachments that need printing. Saves a few seconds every time.
I have similar experiences with mopoke.
I've automated a lot of small tasks lately and as a result I have a small library of stuff which makes it easier and faster to automate yet more small tasks. It definitely is worth the invested time.
I've automated a lot of small tasks lately and as a result I have a small library of stuff which makes it easier and faster to automate yet more small tasks. It definitely is worth the invested time.
Check out http://sikuli.org/. It's visual based programming for automating task. I use it all the time and it's super fast to set up.
It is sad though, that we can automate software hassles with such ease, but we can't automate real world tasks yet.
I automate a lot of excel and outlook related work
Also, what other tasks do you currently automate?