aws / aws/graph-explorer

Kickoff: AI Model Integration

Open
#2,013 0 comments 0 reactions 0 assignees View on GitHub
internal wayfinder:kickoff
Dominant language
TypeScript
Stars
481
Forks
108
Avg merge
6d 8h
Merged PRs (30d)
5

Description

> [!NOTE]
> **Kickoff — Wayfinder not yet run, no chart yet.** Placeholder child of the **AI Model Integration** initiative (#2012). "Placeholder" means only that there's no chart — the content below is real. Run `/wayfinder` against the initiative to create the map; **this kickoff closes when that map is created**.

## Rough intent

Research and scope how AI models could integrate into Graph Explorer, and produce a concrete **list of features AI would unlock** — then identify the highest-value candidates.

## Notes

- **Skills to consult when charting:** `/research` (model options, hosting, cost), `/wayfinder`, `/prototype` for candidate features, `/domain-modeling`.

## Paths to explore

- **Feature catalog** *(grilling)* — enumerate what AI could unlock. Known candidates so far:
- **Natural-language querying** — translate plain English to Gremlin/SPARQL (relates to #690, #1945).
- **AI-generated schema theme** — describe a look and have a model produce a coordinated style theme for your schema (an AI *action* that passes schema info to a model and enacts the result; ties to the Styling System Overhaul #2005).
- **AI-chosen canvas layout** — have a model pick/tune a node & edge layout for the current graph (an AI *action*; ties to layout work).
- Others to surface: summarizing a subgraph, explaining a path, suggesting expansions.
- **Model hosting & cost** *(research)* — where do models run given Graph Explorer is self-hosted? Bring-your-own API key, Bedrock, local models? Who pays?
- **Prioritization** *(grilling)* — of the catalog, which features are highest value vs. effort?

## Potential blockers & pitfalls

- **Self-hosted deployment:** no central backend to call a model from — likely requires the user to supply their own model access (API key/Bedrock), which shapes every feature.
- **Cross-cut with styling (#2005):** the AI-theme action depends on the style model that initiative defines — sequence accordingly.
- **Trust/accuracy:** AI-generated queries/styles must fail safe and be reviewable; a wrong query against a user's DB is worse than no feature.

## Related issues

- Initiative: #2012
- Related: #690 / #1945 (natural language querying), #2005 (Styling System Overhaul — for the AI-theme action)

> [!IMPORTANT]
> Internal only — this issue is maintained by the core team and is not accepting external contributions.

Contributor guide

Open the contributing guide

Research direction

Start with initiative #2012 and run `/wayfinder` as requested; review related issues #690, #1945, and #2005 for existing context. Done means the initiative has a concrete map, an AI feature catalog, model-hosting and cost considerations, and prioritized candidates; this issue then closes when that map is created.

Written by the indexing model from the issue text.

Assessment

Tech stack
machine-learning, react, typescript
Domain
ai, web-dev
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.