Replies: 53 comments 111 replies
-
Can I enable the AI in an unofficial way now in my instance?The answer is yes. Steps for those using provided compose file:1. create a JSON config at
{
"$schema": "https://github.com/toeverything/affine/releases/latest/download/config.schema.json",
"copilot": {
"enabled": true,
"providers.openai": {
"apiKey": "your key",
"baseUrl": "open-ai-compatitable.example.com"
}
}
}2. restart your host docker compose up -dSteps for custom deployment:1. create a JSON config {
"$schema": "https://github.com/toeverything/affine/releases/latest/download/config.schema.json",
"copilot": {
"enabled": true,
"providers.openai": {
"apiKey": "your key",
}
}
}2. Mount or Copy the config file to your instance 3. import the config node --import ./scripts/register.js ./dist/data/index.js import-config ./path/to/config.json4. restart your host |
Beta Was this translation helpful? Give feedback.
-
|
I try to set up today but get import-config not found issue. any ideas? |
Beta Was this translation helpful? Give feedback.
-
|
pull the latest image |
Beta Was this translation helpful? Give feedback.
-
|
I tried this configuration to connect to my ollama server but it didn't work {
"$schema": "https://github.com/toeverything/affine/releases/latest/download/config.schema.json",
"copilot": {
"enabled": true,
"providers.openai": {
"apiKey": "ollama",
"baseUrl": "http://IP:11434/v1"
}
}
} |
Beta Was this translation helpful? Give feedback.
-
How can I configure DeepSeek? How to set up the model. Moreover, in the 0.21.2 management panel, the AI function cannot be activated.
|
Beta Was this translation helpful? Give feedback.
-
|
That why I called the solution in #11691 (comment) is "unofficial". to configure AI functionality in Admin panel is still work in progress. Regarding using DeepSeek, it's concurrently not supported yet, but I think it will be available with AI admin panel together |
Beta Was this translation helpful? Give feedback.
-
|
Hello! Thanks for the tips for selfhosted instance Will there be AI support for local workspace? |
Beta Was this translation helpful? Give feedback.
-
|
You can only use simple actions in local workspace like random chat, translation. The actions with context awareness, for example embedding, can not be used with local workspaces, because the server doesn't know the content of your local workspace |
Beta Was this translation helpful? Give feedback.
-
|
好像仍然有10次的使用限制，我使用自己的 API key 进行配置，并在openrouter后台看到了使用记录，但是affine中的AI仍然只能使用10次 |
Beta Was this translation helpful? Give feedback.
-
|
Wow, it's really possible! In the account settings, click on modify, and you can see the options, thank you very much 🙏 |
Beta Was this translation helpful? Give feedback.
-
|
You're welcome :) |
Beta Was this translation helpful? Give feedback.
-
你使用的是什么模型，我这边这个管理员面板的 AI 按钮无法开启，是禁用状态 |
Beta Was this translation helpful? Give feedback.
-
|
@ssly 使用OpenAI兼容的API以及正确的apikey没有问题，服务启动时会检测API的可用性，可以查看日志。如果没有办法开启AI可能是配置有误。理论上还支持其他的AI provider，具有可用性可查看对应的源代码。 |
Beta Was this translation helpful? Give feedback.
-
|
请问，怎么更换掉系统原来的提示词。每次提问都带一大串自我介绍提示词 |
Beta Was this translation helpful? Give feedback.
-
|
I created for myself a small server between Affine and Gemini or OpenRouter, who provides free API keys. Not sure if this allowed here in this chat. |
Beta Was this translation helpful? Give feedback.
-
|
Excellent, I used your method. Successfully used the model of Gemini and openrouter on AFFiNE 0.21.4 |
Beta Was this translation helpful? Give feedback.
-
|
Edit: Figured it out: I am following your guide from your repo to be able to use OpenAI with the AFFINE version of 0.21.4 which is the latest stable version and the connection is not working for some reason. Any help would be appreciated. Below are all my files:
{
"$schema": "https://github.com/toeverything/affine/releases/latest/download/config.schema.json",
"server": {
"name": "Affine"
},
"copilot": {
"enabled": true,
"providers.openai": {
"apiKey": "my-not-exist-api-key",
"baseUrl": "http://affine-copilot-fix:5000"
}
}
}
name: affine
services:
affine-copilot-fix:
container_name: affine-copilot-fix
user: "1000:1000"
build:
context: ./affine-copilot-fix
dockerfile: Dockerfile
volumes:
- ./logs:/app/src/logs
env_file:
- .env
environment:
- CREATE_LOG=True
- OPENAI_API_KEY=${OPENAI_API_KEY}
- OPENAI_BASE_URL=https://openrouter.ai/api/v1 # https://openrouter.ai/api/v1/
- OPENAI_MODEL=openai/o4-mini # Or any openrouter model
ports:
- 5000:5000
networks:
- proxy
affine:
image: ghcr.io/toeverything/affine-graphql:${AFFINE_REVISION:-stable}
container_name: affine_server
ports:
- '${PORT:-3010}:3010'
depends_on:
redis:
condition: service_healthy
postgres:
condition: service_healthy
affine_migration:
condition: service_completed_successfully
volumes:
# custom configurations
- /opt/docker/affine/.affine/self-host/storage:/root/.affine/storage
- /opt/docker/affine/.affine/self-host/config:/root/.affine/config
env_file:
- .env
environment:
- REDIS_SERVER_HOST=redis
- DATABASE_URL=postgresql://${DB_USERNAME}:${DB_PASSWORD}@postgres:5432/${DB_DATABASE:-affine}
restart: unless-stopped
networks:
- proxy
affine_migration:
image: ghcr.io/toeverything/affine-graphql:${AFFINE_REVISION:-stable}
container_name: affine_migration_job
volumes:
# custom configurations
- /opt/docker/affine/.affine/self-host/storage:/root/.affine/storage
- /opt/docker/affine/.affine/self-host/config:/root/.affine/config
command: ['sh', '-c', 'node ./scripts/self-host-predeploy.js']
env_file:
- .env
environment:
- REDIS_SERVER_HOST=redis
- DATABASE_URL=postgresql://${DB_USERNAME}:${DB_PASSWORD}@postgres:5432/${DB_DATABASE:-affine}
depends_on:
postgres:
condition: service_healthy
redis:
condition: service_healthy
networks:
- proxy
redis:
image: redis
container_name: affine_redis
healthcheck:
test: ['CMD', 'redis-cli', '--raw', 'incr', 'ping']
interval: 10s
timeout: 5s
retries: 5
restart: unless-stopped
networks:
- proxy
postgres:
image: postgres:16
container_name: affine_postgres
volumes:
- /opt/docker/affine/.affine/self-host/postgres/pgdata:/var/lib/postgresql/data
environment:
POSTGRES_USER: ${DB_USERNAME}
POSTGRES_PASSWORD: ${DB_PASSWORD}
POSTGRES_DB: ${DB_DATABASE:-affine}
POSTGRES_INITDB_ARGS: '--data-checksums'
# you better set a password for you database
# or you may add 'POSTGRES_HOST_AUTH_METHOD=trust' to ignore postgres security policy
POSTGRES_HOST_AUTH_METHOD: trust
healthcheck:
test:
['CMD', 'pg_isready', '-U', "${DB_USERNAME}", '-d', "${DB_DATABASE:-affine}"]
interval: 10s
timeout: 5s
retries: 5
restart: unless-stopped
networks:
- proxy
networks:
proxy:
driver: bridge
external: true |
Beta Was this translation helpful? Give feedback.
-
|
any issues or questions, please open in my repo. |
Beta Was this translation helpful? Give feedback.
-
|
@boom404 Hi, how do I customize the prompt word. Otherwise every request comes with a long string of prompt words: |
Beta Was this translation helpful? Give feedback.
-
this is hard coded in Affine. |
Beta Was this translation helpful? Give feedback.
-
|
Since recently, I get this error message when trying to use it: My relevant {
"$schema": "https://github.com/toeverything/affine/releases/latest/download/config.schema.json",
"server": {
"name": "myname",
"host": "affine.myname.de",
"https": true
},
"copilot": {
"enabled": true,
"providers.openai": {
"apiKey": "sk-proj-...",
}
}
}Any ideas? |
Beta Was this translation helpful? Give feedback.
-
|
I experienced the same problem. The AI chat in the sidebar wasn't working, but the normal "Ask AI" in documents was fine. I've turned off the "world icon" in the chat sidebar. But, i'm using my own Affine Copilot Fix, because I want to use Gemini - nevertheless, the problem should be maybe the same.
|
Beta Was this translation helpful? Give feedback.
-
|
Same issue here using the latest stable self hosted docker version. |
Beta Was this translation helpful? Give feedback.
-
|
if you enable the websearch feature, by toggling the |
Beta Was this translation helpful? Give feedback.
-
|
Which docker version does your solution work with? I have used stable and canary version and none work with your configuration and would like to know the version that works to see the difference in code from before to now to know how to fix this and possibly have a Github PR made with a solution. |
Beta Was this translation helpful? Give feedback.
-
|
the latest stable should work fine |
Beta Was this translation helpful? Give feedback.
-
|
Is there a way to define which model to use? |
Beta Was this translation helpful? Give feedback.
-
|
not available yet |
Beta Was this translation helpful? Give feedback.
-
|
I use openresty to redirect the requests to deepseek and the config.json |
Beta Was this translation helpful? Give feedback.
-
I think this method is good, but after trying it, it still gets |
Beta Was this translation helpful? Give feedback.
-
|
Copilot provider [openai] registered. Gemini might be because of the network (configured in the background). Openai is configured in the background and config.json. It still prompts that it is unavailable. |
Beta Was this translation helpful? Give feedback.
-
|
Is there a usage cap when using AI on self hosted even when using my own AI Token???? |
Beta Was this translation helpful? Give feedback.
-
|
nope |
Beta Was this translation helpful? Give feedback.
-
|
Even tho I set the config.json (validated if its set with cat config.json in root/.affine/config) and set unlimited ai usage, I get this error when trying to send a message: |
Beta Was this translation helpful? Give feedback.
-
|
Same here, but I think it's because of the unlimited_Copilot Feature. When I set my config.json and imported it into node, everything was working until the quota cap. Once I set unlimited_copilot, the quota was gone, but I get this error. So i think they have hardcoded something that tries to use a specific ai provider once somebody has purchesed the ai feature. I mean it makes total seance cause once you purchased, affine is responsible for providing the ai.. |
Beta Was this translation helpful? Give feedback.
-
|
currently AFFiNE binds actions to specific provider or models, because the models capabilities distinguish, like some model only process text and return text, some can read from pictures, some can generate pictures. You may also hear GPT-4o, GPT-o1, gemini-2.0, etc. They are all for different purposes. And what's even worse, some advanced models was considered to be used in AFFiNE are not available for all, like openai accounts. the error this is the main reason we didn't tell AFFiNE AI as self-host ready and can only be used as experimental feature. |
Beta Was this translation helpful? Give feedback.
-
|
Coudn't we allow users to congigure which model to use instead of hard coding, eg: openai/openai-4.1-mini, etc... and just mention that the LLM endpoint must be openai compatible. This helps, eg: when using liteLLM that give an openai endpoint for whichever model is configured behind it. ofcourse for a selfhosted instance the user/hoster will need to know if they should use a text only or a multimodal model. There are cheaper and even free models, that can be used this way, eg: gpt4.1-mini instead of gpt-4.1 (which is hardcoded and more expensive) or Mistral-small which is free and multimodal. |
Beta Was this translation helpful? Give feedback.
-
|
definitely. We are on it but it will take some time to finish the refactory. Please be patient then. |
Beta Was this translation helpful? Give feedback.
-
|
I am trying to set up the latest stable and beta for AI running on a self-hosted cloud. After several tries I got this: Is it possible to resolve this? |
Beta Was this translation helpful? Give feedback.
-
|
Hope AFFiNE to support custom providers(ollama, deepseek, ...) |
Beta Was this translation helpful? Give feedback.
-
|
Hey guys, since ./scripts/register.js got removed, I can't import my config for the ai anymore with node --import ./scripts/register.js ./dist/data/index.js import-config ./path/to/config.json. Did anybody get ai to work on 0.22.2? |
Beta Was this translation helpful? Give feedback.
-
facing same issue |
Beta Was this translation helpful? Give feedback.
-
|
@forehalo |
Beta Was this translation helpful? Give feedback.
-
|
npm run cli import-config /root/.affine/config/config.json |
Beta Was this translation helpful? Give feedback.
-
|
Any updates? |
Beta Was this translation helpful? Give feedback.
-
|
I'd like to get some info as well. The deep AI Integration is one of the main features of affine and I would love to see that for self-hosted as well. |
Beta Was this translation helpful? Give feedback.
-
|
Still getting the same error as mentioned above in self hosted, after 7 days.
|
Beta Was this translation helpful? Give feedback.
-
|
Just wanted to drop by and report that I've finally managed to connect affine to my LLM server (currently Ollama, but does not matter, any OpenAI API compatible will work) using a solution proposed in https://github.com/axcode07/affine_ai_helper Except I took the idea and incorporated it into my own OpenAI proxy, and also forces a specific model name, just in case. So model name from Affine is completely ignored and replaced (not sure if this is something that the affine_ai_helper does). This still just means that Affine does not follow the OpenAI API spec correctly and it is something that has to be fixed. Writing adapters like that should not be required for software that claims to support local LLM servers. |
Beta Was this translation helpful? Give feedback.
-
|
Hey how do I connect to azure openai? |
Beta Was this translation helpful? Give feedback.
-
|
Whats the problem with it that affine helper does not solve? Frankly I did not use it. With my proxy it's just: Affine -> OpenAI Proxy -> any openai compatible server. I assume that you should be able to set it up the same way: Affine -> Affine AI Helper -> Azure OpenAI I am not sure why the author chose to include Caddy and LiteLLM in his example there. I assume it's just what he's using, but it should likely work without those extra links in the chain. The important thing is to translate the /responses to /v1/chat/completions and then adapt the returned data. This is what Affine AI Helper is supposed to be doing. LiteLLM might be there to force a name of the model. |
Beta Was this translation helpful? Give feedback.
-
|
Hi there, |
Beta Was this translation helpful? Give feedback.
-
Could you share your affine and maybe litellm config and what extra steps, if any, you took? Of course without the private bits. I think many have tried but failed to do it directly through litellm, without proxies, me included. It would help others. |
Beta Was this translation helpful? Give feedback.
-
|
After a long time and numerous updates to Affine, I'm attempting to solve this problem again. Now for v0.25.0, you could add a custom copilot in a self-hosted instance using the following method:
{
"$schema": "https://github.com/toeverything/AFFiNE/releases/latest/download/config.schema.json",
"copilot": {
"enabled": true,
"scenarios": {
"override_enabled": true,
"scenarios": {
"audio_transcribing": "gemini-2.5-flash",
"chat": "gemini-2.5-flash",
"embedding": "gemini-embedding-001",
"image": "gpt-image-1",
"rerank": "gpt-4.1",
"coding": "claude-sonnet-4-5@20250929",
"complex_text_generation": "gpt-4o-2024-08-06",
"quick_decision_making": "gpt-5-mini",
"quick_text_generation": "gemini-2.5-flash",
"polish_and_summarize": "gemini-2.5-flash"
}
},
"providers.openai": {
"apiKey": "sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
"baseUrl": "https://api.openai.com/v1"
},
"exa": {
"key": "xxxxxxxxxxx"
}
}
}Fill in your own ApiKey and baseUrl. Each model name used can be modified as needed. Ensure they are accessible through your providers. Providers can be OpenAI, Gemini, etc. Refer to the To enable the
For those using local LLMs, services like new-api can be used to translate requests to OpenAI-compatible format, although this inherently adds a certain level of complexity. |
Beta Was this translation helpful? Give feedback.
-
I'm not talking about the original trained model itself. I'm saying that at the level where Affine calls the API, some API services provide web search functionality, and you just need to handle the request and response in a special way, i.e gemini models with grounding with google search. Of course, many API services only provide the raw model chat, and for this part, it clearly requires additional implementation of the web search function, whether using MCP or something else. If that's the case, then this part of the implementation is clearly missing. I haven't examined Affine's source code carefully, so I'm not sure how the official team is handling this issue.
As for how Affine associates model names with providers, based on my experience and guess, the model name is not forcibly bound to a specific provider. Using a model name like "gemini-xxxx" will still use the provided OpenAI-compatible API request. As for whether there is some kind of fallback or request priority, that needs to be confirmed by looking at the code. |
Beta Was this translation helpful? Give feedback.
-
|
@Nixellion @NeverBeLazyG I have successfully enabled the As for the specific implementation details, you can also ask deepwiki to obtain them. |
Beta Was this translation helpful? Give feedback.
-
|
after investigating the source code of Affine i found out, that in the current state the web search (with the button in the affine ai UI)is hard coded to use an api from : https://exa.ai/ i dont know why exa.ai i am very confused. never heard of it earlier. and why its hard coded. so how i see is: its not possible to change the websearch without changing the source code, or do some other proxy level tricks. 
|
Beta Was this translation helpful? Give feedback.
-
|
Intercepting the network requests without modifying the source code would be quite complicated. I understand why Affine currently hard-codes the integration of exa's service. The web search relies on two function calls implemented through exa's API. Even if there are open-source projects that offer web search services, their APIs might not be completely consistent. This would require a lot of adaptation and decoupling work, which might not be a high priority for Affine's development team. Affine still needs to improve many things, including the self-hosting experience. My understanding is that Affine is at least currently a project that can be self-hosted (though not completely customized), but it's not focused on self-hosting or saying self-hosting first. |
Beta Was this translation helpful? Give feedback.
-
|
hey does this replace the model name in the affine UI under intelligence section,? |
Beta Was this translation helpful? Give feedback.
-
Although version 0.25.1 has been released, it still does not support custom providers such as Ollama and DeepSeek. Is there an estimated timeline for when this feature will be available? |
Beta Was this translation helpful? Give feedback.
-
|
Ollama provides an openAI compatible API endpoint no? setting the baseUrl to your ollama url doesn't work? |
Beta Was this translation helpful? Give feedback.
-
Thats the point, it does not handle openai api without a proxy that modifies requests and responses like the ones discussed in the thread |
Beta Was this translation helpful? Give feedback.
-
|
it does now, with the oldApiStyle:true parameter: |
Beta Was this translation helpful? Give feedback.
-
|
I have AI working. Am I correct in assuming the names in the UI will not change regardless of what I put in the AI config? They will always say gemini? |
Beta Was this translation helpful? Give feedback.
-
|
Web Search failed in Affine AI option and I am using gemini api key |
Beta Was this translation helpful? Give feedback.
-
|
ty |
Beta Was this translation helpful? Give feedback.
-
|
Does anyone know how to integrate openrouter in place of open ai? |
Beta Was this translation helpful? Give feedback.
-
|
chat and text generation was working for me the other day, but now nothing works. cursed application. anyway, at the very least make sure you are using the correct model names. open router needs the company prefixed. so gpt-4 would be openai/gpt-4 and so on in config settings. use the 'copy' button in the respective model's model page on open router |
Beta Was this translation helpful? Give feedback.
-
|
What's going on here? 
The pricing page for self-hosted AFFiNE doesn't even mention AI usage limits. 😢 
|
Beta Was this translation helpful? Give feedback.
-
|
You can switch off this restriction in the settings, then you can make infinite AI requests. It's typical nonsense from Affin again.
|
Beta Was this translation helpful? Give feedback.
-
|
I don't know if it's the same for you, but after struggling for a few days, I went to the admin settings > Accounts on the far right, click the 3 dots under "actions" for your username > edit > then select UnlimitedCopilot. Everything worked fine for me after that. I hope it's as easy for someone else as it was for me. |
Beta Was this translation helpful? Give feedback.
-
|
It worked, thank you so much! 🙌🏻 |
Beta Was this translation helpful? Give feedback.
-
|
Same, thanks a lot ! |
Beta Was this translation helpful? Give feedback.
-
|
thanks |
Beta Was this translation helpful? Give feedback.
-
Can people share their working configs?someone post a working open router config, a litellm, a ollama, etc. then we can put them at the top in the main message @forehalo this thread is getting long with lots of conflicting information |
Beta Was this translation helpful? Give feedback.
-
|
Here, I use LiteLLM to proxy third-party models, employing a reverse proxy method (cloudflare worker or caddy server) to proxy the Gemini API. I didn't cover all the scenarios. That can only solve the basic problem of using third-party models in Affine selfhost. |
Beta Was this translation helpful? Give feedback.
-
|
I wonder if everyone is still interested in this topic. I’ve made an important discovery (with the help of GitHub Copilot): by simply setting a single environment variable— The article has been updated, and I hope it will be helpful for everyone https://torchtree.com/en/post/affine-selfhost-ai-configuration/ 
|
Beta Was this translation helpful? Give feedback.
-
|
Setting This issue persists with self-hosted LLMs like vLLM and Ollama, where the logs from |
Beta Was this translation helpful? Give feedback.
-
|
Alright looks like you need a model with tools functionality. Switching to Qwen 3 30b works. |
Beta Was this translation helpful? Give feedback.
-
|
Correct, I'm using "qwen2.5:7b-instruct". Still looking for a working embedding model. |
Beta Was this translation helpful? Give feedback.
-
Are Image Inputs working for you? |
Beta Was this translation helpful? Give feedback.
-
|
Apparently not, AI thinks my images are white. |
Beta Was this translation helpful? Give feedback.
-
Images just give me an error :/ |
Beta Was this translation helpful? Give feedback.
-
|
Hi
I still have the error message about the limit call and what is strange is that in the provider selector I only have Gemini, no other element or model to select. I don't see any error in the logs What am I missing ? |
Beta Was this translation helpful? Give feedback.
-
|
Agreed however I will not complain. AFFiNE has great features in a free application so I can accept to run a SQL command from time to time |
Beta Was this translation helpful? Give feedback.
-
|
I think i activated 7 | unlimited_copilot in the database. |
Beta Was this translation helpful? Give feedback.
-
|
For deepseek-chat:
Works in v0.25.7 |
Beta Was this translation helpful? Give feedback.
-
|
Where to add 'self hosted=true' in Docker compose? Looking forward to your reply! @methodx |
Beta Was this translation helpful? Give feedback.
-
|
@auto-fly it's docker-compose.yaml. |
Beta Was this translation helpful? Give feedback.
-
|
I tried everything in this thread and Copilot still not working. Even using an OpenAI key after a fresh install I get I'm using LiteLLM and is working well with other tools, and the config for the openai provider set at I even used a proxied URL and AFFiNE is not calling it when I configure it as openai baseUrl... How can this be debugged to at least find the URL used? |
Beta Was this translation helpful? Give feedback.
-
|
same issue the only error i get is |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Progress
Beta Was this translation helpful? Give feedback.
All reactions