Overview
Vertex AI is Google Cloud’s model service. Loom signs in with a service-account key (a JSON file) and renews its short-lived tokens for you, so you only paste the file once.
Loom has two entries here: Google Vertex AI (Gemini and open partner models) and Claude on Vertex AI (Claude models). The setup is the same.
Get a key
- 1In the Google Cloud console, pick a project with billing turned on, and enable the Vertex AI API (APIs & Services → Library → Vertex AI API → Enable).
- 2Go to IAM & Admin → Service Accounts and create a service account. Give it the Vertex AI User role.
- 3Open the service account, go to Keys → Add key → Create new key, choose JSON and create. A file downloads.
- 4For Claude only: open Model Garden, find the Claude model you want, and Enable it (access can take a little while to be approved). Quota is per region, so if you hit limits try the
globallocation.
Cost. Billed through your Google Cloud account. A service-account key is a long-lived secret: create a dedicated account for Loom, and delete the key when you stop using it.
Add it to Loom
- 1Open Loom, then Providers in the menu, and choose Add provider.
- 2Pick Google Vertex AI or Claude on Vertex AI (the search box finds them).
- 3Fill in the fields in the table below.
- 4Choose Connect. Loom checks the key and says how many models it found. If it cannot check, it says so, and you can Save anyway and try a message.
- 5Open the model chip below the message box and pick a model.
- Paste the whole contents of the JSON file into Service-account key (JSON). Loom reads the project ID from it.
- Location:
globalis the best start. Use a region such asus-central1if your model or data rules require it. - Gemini and partner models: the optional More model names box lets you add models that are not in the starter list.
Google Vertex AI
| Field in Loom | What to put | Needed |
|---|---|---|
| API key (optional) | Leave empty when you paste a service-account key. Or paste a short-lived access token here | Optional |
| Service-account key (JSON) | { "type": "service_account", … } In Google Cloud: IAM → Service accounts → Keys → Add key → JSON. Give the account the “Vertex AI User” role. It stays on this device. | Optional |
| Project ID | my-project-123 Filled in from the key if you leave it empty. | Optional |
| Location | Pick from the list, or choose Other… to type one. global Global works for most models. Pick a region if your project needs data to stay in one place. | Yes |
| More model names (optional) | google/gemini-3.5-pro Separated by commas. | Optional |
| Endpoint | https://{vertexHost}/v1/projects/{project}/locations/{location}/endpoints/openapifilled in from the fields above | Pre-filled |
| Name | Anything you like. It is shown in the model picker | Optional |
Claude on Vertex AI
| Field in Loom | What to put | Needed |
|---|---|---|
| API key (optional) | Leave empty when you paste a service-account key. Or paste a short-lived access token here | Optional |
| Service-account key (JSON) | { "type": "service_account", … } In Google Cloud: IAM → Service accounts → Keys → Add key → JSON. Give the account the “Vertex AI User” role. It stays on this device. | Optional |
| Project ID | my-project-123 Filled in from the key if you leave it empty. | Optional |
| Location | Pick from the list, or choose Other… to type one. global Newer Claude models are on Global or the US/Europe multi-regions, not every single region. | Yes |
| Endpoint | https://{vertexHost}filled in from the fields above | Pre-filled |
| Name | Anything you like. It is shown in the model picker | Optional |
At a glance
| Appears in Loom as | Google Vertex AI |
| Sign-in | Service-account JSON (Loom fetches and renews access tokens), or a pasted access token that lasts about an hour |
| Models | You type or pick model names; this service has no list call Loom can use |
| Tools (connectors, search) | Yes |
| Photos | Yes |
| PDFs | Documented by Vertex |
If something goes wrong
| What you see | What to do |
|---|---|
| 403, “permission denied” | The service account is missing the Vertex AI User role, or the API is not enabled in that project. |
| 404, “model not found” | The model is not available in that location, or (for Claude) not enabled in Model Garden. |
| 429 | Regional quota. Try the global location, or request more quota. |