akumatus
95cf2e047f
feat(core): cite source documents in the AI answer ( #9863 )
...
Support issue [BS-2424](https://linear.app/affine-design/issue/BS-2424 ).
### What changed?
- Add relevant document prompt templates.
- Add citation rules in system prompts.
- Change message `params` type to `Record<string, any>`
- Add unit test.
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/ec24e664-039e-4fab-bd26-b3312f011daf.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/ec24e664-039e-4fab-bd26-b3312f011daf.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/ec24e664-039e-4fab-bd26-b3312f011daf.mov ">录屏2025-01-23 10.40.38.mov</video>
2025-01-24 04:04:01 +00:00
akumatus
862a9d0bc4
feat: use footnote for perplexity search results ( #9851 )
...
Support issue [BS-2475](https://linear.app/affine-design/issue/BS-2475 ).
2025-01-22 10:54:01 +00:00
L-Sun
29995e498a
feat(editor): add start-with-ai button for empty doc ( #9836 )
...
Close [BS-2391](https://linear.app/affine-design/issue/BS-2391/bs-ai-toolbar-空状态下添加-actions-列表 )
https://github.com/user-attachments/assets/cbded517-2d3d-4a75-b144-644e2b03f68a
2025-01-21 16:00:49 +00:00
akumatus
d048ac6c91
feat(core): support chat panel chips and suggest current doc for embedding ( #9747 )
...
Support issue [BS-2347](https://linear.app/affine-design/issue/BS-2347 ).
## What Changed?
- Add chat panel components
- `chat-panel-chips`
- `chat-panel-doc-chip`
- `chat-panel-file-chip`
- `chat-panel-chip`
- Add `chips` and `docs` field in `ChatContextValue`
- Add `extractMarkdownFromDoc` function to extract markdown content of a doc
- Add e2e test
Click a candidate card to add it into AI chat context:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/4e6b11ef-f993-4e6a-9f40-b2826af1990c.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/4e6b11ef-f993-4e6a-9f40-b2826af1990c.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/4e6b11ef-f993-4e6a-9f40-b2826af1990c.mov ">录屏2025-01-17 01.02.04.mov</video>
2025-01-18 08:35:19 +00:00
akumatus
c8e550138d
fix(core): add ai send button test ( #9685 )
2025-01-14 10:06:31 +00:00
akumatus
2f49c064eb
feat(editor): remove chat-cards and Open AI Chat menu item ( #9626 )
...
Support issue [BS-2353](https://linear.app/affine-design/issue/BS-2353 ).
### What changed?
- Remove `Open AI Chat` menu item.
- Remove `chat-cards` web component.
- Add `extractAll` function for page and edgeless doc full content extract.
2025-01-10 03:53:11 +00:00
akumatus
58ce86533e
feat(core): support ai network search ( #9357 )
...
### What Changed?
- Add `PerplexityProvider` in backend.
- Update session prompt name if user toggle network search mode in chat panel.
- Add experimental flag for AI network search feature.
- Add unit tests and e2e tests.
Search results are streamed and appear word for word:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/56f6ec7b-4b21-405f-9612-43e083f6fb84.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/56f6ec7b-4b21-405f-9612-43e083f6fb84.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/56f6ec7b-4b21-405f-9612-43e083f6fb84.mov ">录屏2024-12-27 18.58.40.mov</video>
Click the little globe icon to manually turn on/off Internet search:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/778f1406-bf29-498e-a90d-7dad813392d1.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/778f1406-bf29-498e-a90d-7dad813392d1.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/778f1406-bf29-498e-a90d-7dad813392d1.mov ">录屏2024-12-27 19.01.16.mov</video>
When there is an image, it will automatically switch to the openai model:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/56431d8e-75e1-4d84-ab4a-b6636042cc6a.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/56431d8e-75e1-4d84-ab4a-b6636042cc6a.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/56431d8e-75e1-4d84-ab4a-b6636042cc6a.mov ">录屏2024-12-27 19.02.13.mov</video>
2025-01-09 04:00:59 +00:00
darkskygit
475e3d80b2
fix(server): vague initial chat content in test ( #9267 )
2024-12-24 04:26:52 +00:00
akumatus
2f79104bdb
feat(core): support ai insert image, mindmap, slides and make it real in page mode ( #9164 )
...
Support issue [BS-2085](https://linear.app/affine-design/issue/BS-2085 ).
### What changed?
- Refactor the `actionToAnswerRenderer` function to support reuse in both page mode and edgeless mode.
- Add a new `page-response.ts` module to handle AI-generated answers in page mode.
- Remove the redundant `edgelessHandler` function from `_common/config.ts`.
- Introduce the `AIContext` class along with the `ctx` TypeScript type to standardize context management.
- Implement the `createTemplateJob` function to enable AI slide insertion in both page mode and edgeless mode.
Insert mindmap on page mode:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/30630d3e-ebd9-416b-9bb9-5f27086e48a3.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/30630d3e-ebd9-416b-9bb9-5f27086e48a3.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/30630d3e-ebd9-416b-9bb9-5f27086e48a3.mov ">mindmap.mov</video>
Insert image on edgeless note
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/b850ee5a-a06b-4ae7-8b68-ed5929a6e81a.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/b850ee5a-a06b-4ae7-8b68-ed5929a6e81a.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/b850ee5a-a06b-4ae7-8b68-ed5929a6e81a.mov ">image3.mov</video>
Insert image on page mode:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/c4f98e2d-0b15-4310-b3e0-0725e330302b.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/c4f98e2d-0b15-4310-b3e0-0725e330302b.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/c4f98e2d-0b15-4310-b3e0-0725e330302b.mov ">image.mov</video>
Generate image from image:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/2776a55f-cbb7-47ce-8e7d-7cae243fa3e9.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/2776a55f-cbb7-47ce-8e7d-7cae243fa3e9.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/2776a55f-cbb7-47ce-8e7d-7cae243fa3e9.mov ">image2.mov</video>
Insert presentation on page mode:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/4e228fa5-88f4-478c-8b79-647612d5515c.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/4e228fa5-88f4-478c-8b79-647612d5515c.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/4e228fa5-88f4-478c-8b79-647612d5515c.mov ">ppt.mov</video>
Insert make it real on page mode:
<div class='graphite__hidden'>
<div>🎥 Video uploaded on Graphite:</div>
<a href="https://app.graphite.dev/media/video/sJGviKxfE3Ap685cl5bj/c71139b2-fb55-4d89-84e2-d52eeb905b57.mov ">
<img src="https://app.graphite.dev/api/v1/graphite/video/thumbnail/sJGviKxfE3Ap685cl5bj/c71139b2-fb55-4d89-84e2-d52eeb905b57.mov ">
</a>
</div>
<video src="https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/sJGviKxfE3Ap685cl5bj/c71139b2-fb55-4d89-84e2-d52eeb905b57.mov ">make it real.mov</video>
2024-12-16 10:04:16 +00:00
darkskygit
4d44bcc145
fix(server): image action behavior in ci ( #9077 )
2024-12-10 02:54:31 +00:00
darkskygit
31146e5213
fix: copilot ci ( #9066 )
2024-12-09 11:11:04 +00:00
darkskygit
2e7a7d5909
fix(core): copilot test condition & retry ( #8665 )
2024-11-04 06:17:06 +00:00
darkskygit
7a201984e9
feat(server): add real-world prompt test for copilot apis ( #8629 )
...
fix AF-1432, PD-1176
2024-11-01 07:05:12 +00:00
