Comments
Auto-DM everyone who comments
Someone comments “PRICE” on your Reel and your link lands in their DMs seconds later. This is the thing most creators install it for.
Build the same flows you'd build in ManyChat: keywords, comment-to-DM, broadcasts, sequences, a shared inbox. Then run them on your own server, where nothing counts your contacts.
Six channels. Self-host and nothing is metered.
What it does
The jobs creators and marketers actually hire a tool like this for.
How you build it
Pick what starts the conversation, then stack up what happens next by dragging blocks into order. Send a message, hold for a day, check whether they wrote back, and send the two groups down different paths. There is nothing to code.
Comments
Someone comments “PRICE” on your Reel and your link lands in their DMs seconds later. This is the thing most creators install it for.
Replies
Shipping, pricing, booking, where to buy. Write the answer one time and everyone gets it straight away, including at 2am.
List building
Ask for an email in the middle of the conversation and it saves onto that person. No landing page, no form to build.
Broadcasts
Send an offer to everyone, or only to the people you tagged as buyers. You choose how fast it goes out.
Follow-ups
A reminder tomorrow, a testimonial on Thursday, a last call on Sunday. Anyone who already replied drops out of the run.
Your team
Every channel in one list of conversations. Assign one to your assistant, set a reminder, or take over from the bot yourself. Separate workspaces if you run more than one brand.
It also plugs into Zapier, Make and n8n, so a new lead can land in your CRM or a spreadsheet without you copying anything across.
Channels
ManyChat spreads these across four paid tiers. Here they all ship in the open-source build. Bring your own credentials; they're encrypted at rest.
Bot API. Fastest to set up, fewest messaging restrictions.
DMs, story mentions, comment-to-DM, follow triggers. Inside Meta's messaging window.
Page inbox, ref URLs, keyword triggers. Same window rules, enforced per platform.
Cloud API. ManyChat keeps this on paid plans. Here it's in the free build.
Twilio. STOP and HELP handled for you.
SMTP, Resend or SES. One-click unsubscribe on every send.
The app enforces each platform's messaging window, so a flow can't quietly break the channel's rules.
Pricing
ManyChat runs four tiers priced by contact band, then charges $29 a month more for AI. We run one tier, and it counts nothing.
Self-hosted
$0/month, forever
You pay for a server, plus whatever Twilio or Meta bill you directly.
Get the codeHosted · Free
Coming soon$0/month
Matches ManyChat's free plan, limit for limit. Enough to build a flow and watch it work. Self-hosting has no caps at all.
Coming soonHosted · Unlimited
Coming soon$12/month, billed yearly
or $15 month to month
One price, whether you have 500 contacts or 500,000.
Coming soonThe hosted version isn't open yet. Self-hosting works today, so start there.
Comparison
ManyChat is a good product and the reason this category exists. The point of this table is how much of it you already get: the first eleven rows are identical. The six below them are where the two part company, and two of those go ManyChat's way.
| Feature | BrightBean Chat | ManyChat |
|---|---|---|
| Visual flow builder | Yes | Yes |
| Keyword triggers | Yes | Yes |
| Comment-to-DM | Yes | Yes |
| Story mentions and replies | Yes | Yes |
| Ref URLs and QR codes | Yes | Yes |
| Shared team inbox | Yes | Yes |
| Broadcasts | Yes | Yes |
| Drip sequences | Yes | Yes |
| Tags, custom fields, segments | Yes | Yes |
| REST API and webhooks | Yes | Yes |
| Instagram, Messenger, Telegram | Yes | Yes |
| WhatsApp, SMS and email | Every plan | Paid plans only |
| AI reply generation | Not in v1 | $29/month add-on |
| Template and agency ecosystem | No | Large |
| Self-hosting | Free and unlimited | Not available |
| Source code | Public, AGPL-3.0 | Closed |
| Priced by contact count | Never | Every paid tier |
ManyChat details as published on manychat.com/pricing, September 2026, following their pricing change of 2 March 2026. AI reply generation, TikTok, website chat widgets and e-commerce catalogues are on BrightBean Chat's own list of things deliberately left out of v1.
Self-hosting
Docker Compose brings up the app and PostgreSQL. No Redis, no message broker.
$ git clone https://github.com/brightbeanxyz/brightbean-chat.git
$ cd brightbean-chat
$ docker compose up
Then open http://localhost:8000.
Production compose runs behind Caddy with automatic HTTPS. One-click buttons for Heroku and Render if you'd rather not touch a server.
$ cp deploy/env.prod.example .env && make prod-secrets
$ docker compose -f docker-compose.prod.yml up -d
Django 5, HTMX, Tailwind 4, PostgreSQL. Credentials encrypted at rest, SSRF guards on outbound requests, CSP with nonces. The repo has guides for self-hosting, channel setup, the API and flow templates.
The flow builder, all six channel adapters, broadcasts, sequences, the shared inbox and the API are on GitHub under AGPL-3.0. It is the same code we would run for you on the hosted plan, so nothing useful is held back for a paid edition.
That matters most for the one channel where the audience belongs to you. A contact list you cannot export, running on flows you cannot read, is only yours until the pricing changes.
Read the sourceQuestions
It covers the same ground: keywords, comment-to-DM, flows, broadcasts, sequences, a shared inbox. ManyChat has years of polish, a template library and an agency network that we don't. What you get instead is the source code and a bill that doesn't move when your list does.
It matches ManyChat's on purpose, so the two are easy to compare. 25 active contacts is enough to build a flow and watch it run. If you want real headroom for nothing, self-host instead. That build has no caps.
No feature held back, no contact band. You still pay Twilio, Meta or your email provider directly for messages sent, same as any other tool. The $12 covers running the app.
Yes. No open-core split: all six channels, broadcasts, sequences, the inbox and the API are in the AGPL-3.0 repo. Hosted is the same code with the servers handled.
The concepts map closely. Keywords, flows, sequences, broadcasts, tags and custom fields all exist under the same names. You'll have to rebuild the flows by hand, though. There's no importer yet.
Run it for your own business, modify it, publish nothing. The licence only bites if you offer a modified copy to other people as a network service. Then those users get your source. Marketing your own products doesn't count.
A server with Docker and PostgreSQL. No Redis, no message broker. Then your own credentials for whichever channels you switch on.
It's pre-1.0, so treat it that way. The core is complete and the security baseline is documented. Read SECURITY.md, start on a channel where a mistake is cheap, and check the roadmap before you move anything critical.
Public code, no limits when you self-host. Clone it and see how far it gets you.
AGPL-3.0 License · Django 5 + HTMX + Tailwind 4 + PostgreSQL
We use optional cookies to understand how this site gets used and whether our ads work. Nothing optional loads until you choose, and declining changes nothing about how the site works.
There are exactly two optional things on this website. Here they are, in full.
One entry: your answer to this question. We keep it in your browser's local storage under
bb_consent for 180 days so we don't ask
again. It never leaves your device and identifies nothing about you. That is the entire
list.
| Cookie | Purpose | Duration |
|---|---|---|
| _ga | Random ID so repeat visits count as one person | 180 days |
| _ga_* | Keeps the analytics session state | 180 days |
| Cookie | Purpose | Duration |
|---|---|---|
| __oppref | Which OpenAI ad brought you here, if any | 30 days |
| __obref | Recognises your browser so a later sign-up can be matched to an ad | 1 year, deleted when you withdraw consent |
Full detail in our Privacy Policy and Legal Notice.