Self-Hosting on Hyteck

Trying Twenty: How does an Open Source CRM work?

2025-08-03T06:10:10+02:00

I spend my day working with Salesforce, a very, very feature-rich CRM that you pay big money to use. Salesforce is the opposite of OpenSource and the many features are expensive. Salesforce business model is based on this and on the lock-in effect. If your company invested in implementing Salesforce, they’ll likely pay a lot to keep it.

So what does an alternative look like? Let’s have a look at Twenty, an OpenSource CRM that recently reached the magic 1.0 version.

Getting started

There are two options of getting started: Register at app.twenty.com and start right away on the devs instance or self-host Twenty on your own server. I did the ladder, so let’s discuss how that. The basic steps I took were

point twenty.hyteck.de to a server
Install traefik on the server (I cheated, traefik was already installed)
Deploy this docker-compose.yml with this env file

Then visit the domain and set up the first user.

Features

Twenty offers an initial datamodel that you should be familiar from other CRMs. the standards objects are

Persons A individual person. You can attach notes, E-Mails, etc..
Companies The same for organizations. Organization websites must be unique
Opportunities The classic opportunity with customizable stages
Notes They can be attached to any of the objects above
Tasks Items to work on
Workflows Automations similar to Salesforce flows. E.g. you can create a task every time an Opportunity is created.

The basic datamodel can be extended in the GUI. Here is how my “Company” model looks like

You can add any of the following fields to an object.

Workflows

Workflows are Twenty’s way of allowing users to build automations. You can start a Workflow when a Record is created, updated or deleted. In addition, they can be started manually, on a schedule and via Webhook (yeah!).

You can then add nodes that trigger actions. Available right now are

Creating, updating or deleting a record
Searching records
Sending E-Mails This is the only option to trigger e-mails so far
Code Serverless Javascript functions
Form The form will pop up on the user’s screen when the workflow is launched from a manual trigger. For other types of triggers, it will be displayed in the Workflow run record page.
HTTP request Although possible via Code, this is a handy shortcut to trigger HTTP requests

What is currently completely missing are Foreach-loops and conditions. I can not say “If Opportunity stage is updated to X do Y else, do Z”. Without this, Workflows are really limited in their power.

What already seems quite mature though is the code option. It allows to put in arbitrary code and output a result.

I did not try a lot, but I assume most basic Javascript works. I successfully built an http request that send data to a server.

If what you’re doing is straightforward enough to not use loops and conditions or if you are okay with doing all of them in the Code node, you can do basically anything.

API

Twenty offers an extensive API that allows you to basically do everything. It’s well documented and easy to use.

Here is an example of me, syncing Rescue Organizations from notfellchen.org to Twenty.

import requests

from fellchensammlung.models import RescueOrganization


def sync_rescue_org_to_twenty(rescue_org: RescueOrganization, base_url, token: str):
 if rescue_org.twenty_id:
 update = True
 else:
 update = False

 payload = {
 "eMails": {
 "primaryEmail": rescue_org.email,
 "additionalEmails": None
 },
 "domainName": {
 "primaryLinkLabel": rescue_org.website,
 "primaryLinkUrl": rescue_org.website,
 "additionalLinks": []
 },
 "name": rescue_org.name,
 }

 if rescue_org.location:
 payload["address"] = {
 "addressStreet1": f"{rescue_org.location.street} {rescue_org.location.housenumber}",
 "addressCity": rescue_org.location.city,
 "addressPostcode": rescue_org.location.postcode,
 "addressCountry": rescue_org.location.countrycode,
 "addressLat": rescue_org.location.latitude,
 "addressLng": rescue_org.location.longitude,
 }
 headers = {
 "Content-Type": "application/json",
 "Authorization": f"Bearer {token}"
 }

 if update:
 url = f"{base_url}/rest/companies/{rescue_org.twenty_id}"
 response = requests.patch(url, json=payload, headers=headers)
 assert response.status_code == 200
 else:
 url = f"{base_url}/rest/companies"
 response = requests.post(url, json=payload, headers=headers)
 assert response.status_code == 201
 rescue_org.twenty_id = response.json()["data"]["createCompany"]["id"]
 rescue_org.save()

The Company, Business Model and Paid Features

The company behind Twenty is called “Twenty.com PBC” and mostly seems to consist of former AirBnB employees in Paris. The company is probably backed by Venture Capital. The current business model is to charge for using the company’s instance of Twenty. It starts at 9$/user/month without enterprise features. SSO and support will cost you 19$/user/month.

Selfhosting is free but SSO is locked behind an enterprise badge with seemingly no way to pay for activating it. I suspect that in the future more features will become “Enterprise only” even when self-hosting. All contributors must agree to a Contributor License Agreement (CLA), therefore I believe they could change the License in the future, including switching away from Open Source.

AI Usage

The repo contains a .cursor directory and CLAUDE.md so I assume the devs make (heavy?) use of LLM generated code. The ethical and legal problems with this are for you to decide. I don’t know what effect this has on code quality, for now I’d say things are sometimes buggy (failed upgrades) and UX could be better tested (looking at the e-mail integration) - if this is due to AI slop I don’t know.

Conclusion

Twenty is a really promising start of building a good CRM. The ease of customizing the datamodel, using the API and a solid beginning to Flows allows users to get a lot of value from it already. Flows need some more work to become as powerful as they should be and the E-Mail integration needs to get better.

Stating the obvious: This is not something that could ever replace Salesforce. However, there are many organizations that would benefit a lot from a CRM like Twenty, they simply don’t need, can’t handle or don’t want to pay for all the features other CRMs like Salesforce offer.

If Twenty continues to focus on small to medium companies and the right mix of standard features vs. custom development options I see a path where it becomes a solid choice for these companies. On the other hand there are the usual problems of VC-backed OSS development, and we shall see how it goes for them. Unless there is a strong userbase that credibly threatens a hard fork, enshittification could start soon.

Addendum: Important Features

Here is a short list of features I missed and their place on the roadmap if they have one

Compose & Send E-Mails Planned Q4 2025
Foreach loops in Workflows Q3 2025
Conditions in Flows Q4 2025

Disappearing messages with matrix

2021-12-31T20:00:00+02:00

Introduction

I am a HUGE fan of matrix. It allows me to organize my chats in a sensible way, it works with multiple identities and completly anonymous if I want it to. Spaces made Matrix my favourite messenger by far. Yet, there is one feature I have been missing: Disappearing messages!

Regarding the security and usability, only Signal is comparable to matrix. But: Signal offers the possibilty to define disappearing messages for groups and direct messages ranging from 30 seconds to 4 weeks. No Matrix client (to my knowledge) offers this functionality. Nevertheless, it is possible to configure matrix rooms to have the same feature. This needs a special server configuration and the sending of a special event in the room. This post tries to show both steps. If you do not administer a server you can probably skip to Room configuration

Be aware that this blogpost was written at the end of 2021 - Matrix develops fast and this could be subejct to changes.

Instance configuration

To make disappearing messages possible you need to enable retention on your matrix instance. Retention allows server and room admins to configure how long messages should be kept in the instances database before being purged from it. It is not part of the matrix specification, yet it is supported by synapse.

A client SHOULD not display these messages anymore after the max_lifetime is exceeded. This was NOT true for element web an desktop while staying logged in. Nevertheless, a newly logged in client did not have access to the messages.

To configure Synapse to make use of retention you will need to enable it in your homeserver.yaml

retention:
enabled: true # enables the retention, is enough to enforce it once per day
purge_jobs: # configures a job that delete the events from the database after some tome
- longest_max_lifetime: 3d
interval: 1h
- shortest_max_lifetime: 3d
interval: 1d

The example configuration creates two jobs that delete messages from the database. One only focuses on events that should be deleted after three days or less. These events will be deleted every hour. It is therefore possible for a message that was send in a room with a max_lifetime=7200000 (equals 2h) to be deleted one hour after the maximum lifetime.

Ansible

If you use the Ansible/Docker setup to deploy your server you can add the following to inventory/host_vars/matrix.example.com/vars.yml

matrix_synapse_configuration_extension_yaml: |
retention:
enabled: true
purge_jobs:
- longest_max_lifetime: 1d
interval: 2h
- shortest_max_lifetime: 1d
interval: 1d

Room configuration

If you are a user on a server that has retention enabled, you can enable disappearing messages yourself for each room. Sadly, this is still experimental - but managable! You have to craft a m.room.retention event that defines the maximum lifetime of a message. You will need to access the rooms settings in order to do this.

First you need to open the developer tools in the rooms settings.

Then click “Send custom event” to create your event

And fill the event with the appropriate max_liftime. The time is an integer in milliseconds. X hours is therefore a value of X*3 600 000. Make sure to click the red event button. The State Key can be left empty nevertheless.

Depending on your choosen lifetime the client should not show the messages anymore.

Limitations

The process of deleting messages can not be enforced. A malicious server or chat partner could ignore the request to delete the messages or they could have saved them elsewere. You should not rely on a deletion actually happening. Nevertheless I think this is a good step to take to improve your security in real life.

Further Information

Relevant part of the Synapse configuration file: https://github.com/matrix-org/synapse/blob/v1.36.0/docs/sample_config.yaml#L451-L518
Synapse documenation on message retention policies https://matrix-org.github.io/synapse/v1.41/message_retention_policies.html

What else?

Thanks to Tastytea for helping me get this to work!

Comments

If you have questions, corrections or want to leave something else, please feel free to use the comments!

Cactus comments via Matrix

2021-08-25T11:08:55+02:00

Integration

Following the quickstart documentation I tried to add cactus comments to this blog. I currently rely on infrastructure by cactus.chat as I do not host a private synapse server.

I currently implemented this as a shortcode with hard-coded site title and a variable room name.

Quickstart with HUGO

Register your site

There is a registration system, that ensures that you are moderater in your comment section(s). I order to register your site you have to send a message to @cactusbot:cactus.chat . First try help to ensure that the bot answers you, then register your site e.g. register hyteck. The bot should inform you of success and add you to a moderation room.

Embedd comment section via Shortcode

HUGO, the static site generator I use, has an option to use shortcodes that provide a nice interface to hide some HTML, CSS and JavaScript.

The shortcode chat.html must be added to layouts/shortcodes/ and looks like this

<script type="text/javascript" src="https://latest.cactus.chat/cactus.js"></script>
<link rel="stylesheet" href="https://latest.cactus.chat/style.css" type="text/css">
<div id="comment-section"></div>
<script>
initComments({
node: document.getElementById("comment-section"),
defaultHomeserverUrl: "https://matrix.cactus.chat:8448",
serverName: "cactus.chat",
siteName: "hyteck",
commentSectionId: "{{ index .Params 0 }}"
})
</script>

If you want to use this, replace the site name with the one you registered in the previous step.

You can then use it as simple as

{{< chat cactus-comments >}}

where cactus-comments is the name of the chatroom. You can decide if you want to create a new one for each post (change the name to somthing like the post title) or if you want to use only one (keep the same name).

Organisation

I try to make a comment with my main matrix account, so that I am automatically joined to the room (tbh. I don’t know if this is necessary). Then I organize the rooms by adding them to a private space.

Community

CW: Homophobia, Slur

The community in the official chat seemed helpful and nice. Nevertheless I want to mention, that the first time I tried the demo page a user (that does not seem to be associated with the develoipment or involved in the community) insulted me with a homophobic slur. The “reason” behind this were the prounouns I had in my name. I reported the comment and it was removed by a moderator. The moderator made it clear that this behaviour is agains the Code of Conduct and that the user is not active in the community. For me this is a very good indication of a functioning community, cheers!

Comments

Owncast & Streaming a talk

2021-05-20T16:08:55+02:00

I recently installed an Owncast server and wanted to share my experience. Here it is:

What is owncast?

Owncast is a streaming server that you can selfhost, a Twicht in a box as the developers call it. You host owncast on your server (a small VM with good downlink is enough) and can stream your own own content like you would do on Twicht, YouTube etc…

It has a chat, a admin panel for customization and thats it! You don’t need more to e.g. stream while you are playing minecraft or want to share a talk.

Getting started

Get the latest release on GitHub by using

$ mkdir owncast
$ cd owncast
$ wget https://github.com/owncast/owncast/releases/download/v0.0.7/owncast-0.0.7-linux-64bit.zip
$ unzip owncast-0.0.7-linux-64bit.zip
$ rm owncast-0.0.7-linux-64bit.zip

And move the webroot to your document root and make sure the permissions fit

$ cd ..
$ mv owncast /var/www/owncast
$ cd /var/www/
$ chown -R www-data:www-data owncast

Now create a new NGINX site e.g. /etc/nginx/sites-enabled/owncast with the following content

map $http_upgrade $connection_upgrade {
default upgrade;
'' close;
}
server {
listen [::]:443 ssl ipv6only=on; # managed by Certbot
listen 443 ssl; # managed by Certbot
ssl_certificate /etc/letsencrypt/live/stream.hyteck.de/fullchain.pem; # managed by Certbot
ssl_certificate_key /etc/letsencrypt/live/stream.hyteck.de/privkey.pem; # managed by Certbot
include /etc/letsencrypt/options-ssl-nginx.conf; # managed by Certbot
ssl_dhparam /etc/letsencrypt/ssl-dhparams.pem; # managed by Certbot
server_name stream.hyteck.de;
# Set header
add_header X-Clacks-Overhead "GNU Terry Pratchett";
add_header Permissions-Policy interest-cohort=(); #Anti FLoC
location / {
proxy_set_header Host $host;
proxy_set_header X-Forwarded-Host $host;
proxy_set_header X-Forwarded-Server $host;
proxy_set_header X-Forwarded-Proto $scheme;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection $connection_upgrade;
proxy_pass http://127.0.0.1:8080;
}
}
server {
listen 80;
listen [::]:80;
server_name stream.hyteck.de;
return 301 https://$server_name$request_uri;
}

Make sure to adjust the server name and SSL certificats (I will not go into detail on how to obtain them, but feel free to ask me!).

Start server

Now start owncast to test

$ cd /var/www/owncast
$ ./owncast/owncast

and visit https://yourdomain.org! If everything works you should see your site now. By visiting https://yourdomain.org/admin you can configure your server. The default credentials are admin and your stream key which is abc123. Change this immediately!

Before you configure, let’s make sure this runs whenever your server starts. Goback in the terminal and cancel with Ctrl+C.

Run as system service

You want to install owncast as a system service. Therfore create /etc/systemd/system/owncast.service with the following content:

[Unit]
Description=Owncast
After=network.target
[Service]
Type=simple
Restart=always
RestartSec=1
WorkingDirectory=/var/www/owncast
ExecStart=/var/www/owncast/owncast
[Install]
WantedBy=multi-user.target

Update the daemon with systemctl daemon-reload enable systemctl enable owncast and start with systemctl start owncast. Make sure everything is correct with systemctl status owncast.

Configuration

You can now change back to yourdoiman.org/admin and configure, title, logo and more.

Streaming

You can now use OBS or similar software to start streaming. Got to settings and configure your server.

Concept for talks

To livestream we have a few important things to take care of:

The speaker
The presentation
The streaming software
The streaming server
The chat

I suggest the following setup. Moderation and Speaker are in a BigBlueButton conference. The meet 15 minutes in advance, enough time to upload a presentation. Also in this room is a user called stream. This user does only listen and has OBS configured to record audio and the (fullscreen) window of the BBB room.

Before the talk starts, it is a good opportunity to play some videos via the stream, or simply put on a picture, saing the talk will start soon.

When the speaker an moderation are ready they tell the streamer and the scene is switched to the BBB room. The recording is started (if needed).

A weakness of the setup is the streaner. If they loose their network connection, the stream stops.

There is a neat project for live streaming directly from a BBB room, but it is quite complicated to set up and a crash of the BBB server would make stop the stream. Also the start of the stream is not as beautiful.

Performance

It can be hard to estimate the needed server capacities for a livestream that is why I want to share my experiences. All of this was done on a Strato Linux V40 with 16 GB of RAM, 8 vCPUs and 500 MBit/s.

On the following screenshots you see the configuration and the monitoring data of a talk with 40 participants. We pplayed high-quality videos from 17:43 to 18:05. At 18:05 we changed to the presentation. You see the decrease in CPU load, as the compression of the videos was easier with a steady picture. At 19:05 we invited the participents to join us in the BBB room, some left the stream.

The network load scales linear with the participants that are watching. You can simply calculate the needed downlink capacity with

Downlink capacity = Num. participants x Outbound Video Stream Rate

Here this was

40 participants x 1200 kbps = 48 Mbps

which matches the observed network traffic quite well.

The CPU load is determined by the compression the server has to do. The more different stream formats the server has to put out, the higher the load becomes. On the given Screenshot you see that only one downlink format was advertised, so the inbound stream had to be compresse only once. Matching imbound and outbound stream formats help reducing the server load.

If your bottleneck is the bandwidth you can try to add a second low streamrate, maybe ome clients will take that one.

Final notes

Owncast is a great project. Be aware that it is still not in version 1.0 so expect some limitations. This is e.g. the unability to block a user from chat. If you have any questions let me know in the comments below.