# Description of Changes
Adds storage in the database for full document content alongside the RAG
content (and changes the service to `DocumentService` instead of
`RagService`). Then adds a generic capability that should be usable by
any agent (currently just used by the Question Agent) which allows the
agent to pull out the full contents of the doc, chunks it into various
sections that will fit in the context window, and then processes them in
parallel to create an intermediate result, and then processes the
intermediate result into a final answer. It will re-chunk as many times
as necessary to get the content small enough for the actual answer to be
analysed (I've tested on PDFs ~3500 pages long, which is well above the
context limit and requires maybe 3 rounds of compression to get an
answer).
The new full doc analysis stuff is heavier than the RAG lookup so both
remain. The agents should use RAG for targeted info and the chunked
reasoner for info that requires reading the full doc.
# Description of Changes
Add an extra parameter to every agent to receive the conversation
history in addition to the current message. This will make it possible
to answer followup questions from the AI without needing to give full
context in your message.
# Description of Changes
Redesign AI engine so that it autogenerates the `tool_models.py` file
from the OpenAPI spec so the Python has access to the Java API
parameters and the full list of Java tools that it can run. CI ensures
that whenever someone modifies a tool endpoint that the AI enigne tool
models get updated as well (the dev gets told to run `task
engine:tool-models`).
There's loads of advantages to having the Java be the one that actually
executes the tools, rather than the frontend as it was previously set up
to theoretically use:
- The AI gets much better descriptions of the params from the API docs
- It'll be usable headless in the future so a Java daemon could run to
execute ops on files in a folder without the need for the UI to run
- The Java already has all the logic it needs to execute the tools
- We don't need to parse the TypeScript to find the API (which is hard
because the TS wasn't designed to be computer-read to extract the API)
I've also hooked up the prototype frontend to ensure it's working
properly, and have built it in a way that all the tool names can be
translated properly, which was always an issue with previous prototypes
of this.
---------
Co-authored-by: Anthony Stirling <[email protected]>
Co-authored-by: EthanHealy01 <[email protected]>
# Description of Changes
When I added Prettier formatting in #6052, my aim was to use just the
default settings in Prettier. Turns out, Prettier looks _really hard_
for any config files if it's not explicitly given one, which means that
if a developer has some sort of Prettier config file lying around on
their system, Prettier might find it and use it. Also, Prettier changes
its defaults based on stuff in `.editorconfig` without any good way of
disabling that behaviour explicitly in its config file.
To solve both of these issues, I've introduced a `.prettierrc` file
which sets Prettier's defaults explicitly, and then reformatted all our
code _again_ in Prettier's actual default settings. This should achieve
the aim of #6052 and remove the possibility for it breaking on different
dev computers.
# Description of Changes
Adds a streaming endpoint to the Java AI orchestrator
(`/api/v1/ai/orchestrate/stream` in addition to the existing
`/api/v1/ai/orchestrate`). This allows the caller to get updates of what
stage of orchestration is being run at the time so UIs can give the user
feedback.
Also contains some dubious Gradle changes to suppress errors coming from
Spotless, when it crashes in Google stuff. I'm not sure if that's
appropriate to add, feel free to ask for changes in review.
# Description of Changes
Changes the strategy for autoformatting to reject PRs if they are not
formatted correctly instead of allowing them to merge and then spawning
a new PR to fix the formatting. The old strategy just caused more work
for us because we'd have to manually approve the followup PR and get it
merged, which required 2 reviewers so in practice it rarely got done and
just meant everyone's PRs ended up containing reformatting for unrelated
files, which makes code review unnecessarily difficult. If the PR's code
is not formatted correctly after this PR, a comment will be added
automatically to tell the author how to run the formatter script to fix
their code so it can go in.
This also enables autoformatting for the frontend code, using Prettier.
I've enabled it for pretty much everything in the frontend folder, other
than 3rd party files and files it doesn't make sense for. I also
excluded Markdown because it sounds likely to be more annoying to have
to autoformat the Markdown in the frontend folder but nowhere else. Open
to changing this though if people disagree.
> [!note]
>
> Advice to reviewers: The first commit contains all of the actual logic
I've introduced (CI changes, Prettier config, etc.)
> The second commit is just the reformatting of the entire frontend
folder.
> The first commit needs proper review, the second one just give it a
spot-check that it's doing what you'd expect.
# Description of Changes
Add prototypes folder to test new functionality in. This build of the
app is spawnable with `npm run dev:prototypes`.
Currently just contains a very developer-y chat interface to help us
develop & explore the AI backend before we make the frontend for it for
real.
James BruntonandGitHub
