Use Cases

Your consultants, their infrastructure

On-premise deploy

An infrastructure consultancy deploys a Spaces server on Acme Corp's internal network. Three consultants connect remotely over VPN, each with their own login. Sarah spins up Claude Code against Acme's monorepo to audit their API layer. Meanwhile, James runs Gemini CLI for a security review of the auth module, and a shell pane monitors the client's CI pipeline. Every session is logged. Every credential stays on Acme's hardware. When the engagement ends, the server is wiped — nothing leaves the building.

Spaces — acme-corp — sarah@apex-consulting — 4 panes

Claude Code ~/acme-corp

Claude Reviewing API endpoints for v2 migration...

Found 14 endpoints using deprecated auth pattern:
src/api/users.ts
src/api/billing.ts
src/api/teams.ts

Refactoring to OAuth2 middleware...
> show affected routes_

Gemini CLI ~/acme-corp

Gemini Security audit on auth module...

PASS bcrypt rounds: 12 ✓
PASS JWT expiry: 15min ✓
WARN API key rotation: never ⚠
WARN No rate limit on /login ⚠

2 issues found, 0 critical

Shell ~/acme-corp

$ kubectl get pods -n acme-prod

api-server-7d8f Ready 1/1
worker-pool-3c Running 3/3
redis-cache-01 Ready 1/1

$ _

Aider ~/acme-corp

Aider Applied edit to 3 files.

src/middleware/rateLimit.ts (new)
src/api/auth.ts
src/config/security.ts

Added rate limiter: 100 req/min per IP
aider> _

[audit] session: sarah@apex → acme-corp | duration: 2h14m | files touched: 12 | logged: ✓

One workspace, every engineer

Team rollout

The platform engineering team at a mid-size company rolls out Spaces as the standard AI development environment. Every engineer gets their own login with isolated credentials and session history. Mike on the backend team is building a new payment service with Claude Code while running integration tests in a shell pane. Next to him (figuratively), Lisa uses Gemini CLI to generate docs for the API she shipped last week. The platform team monitors usage in the admin dashboard — active sessions, login events, and agent activity — without ever touching an engineer's machine.

Spaces — payments-v2 — mike@backend-team — 4 panes

Claude Code ~/payments-v2

Claude I'll implement the Stripe webhook handler.

Created 3 files:
src/webhooks/stripe.ts
src/webhooks/handlers.ts
src/types/payment.ts

> add idempotency key check_

Gemini CLI ~/payments-v2

Gemini Generating OpenAPI spec...

Documented 8 endpoints:
POST /payments/create
POST /payments/confirm
GET /payments/:id
POST /webhooks/stripe

openapi.yaml written ✓

Shell ~/payments-v2

$ npm run test:integration

PASS stripe-webhook.test.ts
PASS payment-create.test.ts
FAIL refund-flow.test.ts

Tests: 1 failed, 11 passed, 12 total
$ _

Shell ~/payments-v2

$ docker logs spaces --tail 8

mike logged in (2FA) ✓
lisa logged in (2FA) ✓
sync: 14 new sessions indexed
terminal: 4 active PTYs
$ _

[admin] backend-team: 4 active sessions | agents: claude, gemini | logins today: 6

Every client, fully isolated

Multi-tenant

A 12-person dev agency manages five active client projects. Each client gets its own Spaces workspace with isolated credentials, separate Git configs, and independent session history. Right now, two developers are on the MegaCorp engagement — one building a data pipeline with Claude Code, the other running load tests. After standup, one of them will switch to the StartupXYZ workspace in one click to review a PR. No credential bleed, no context confusion, no "wrong repo" mistakes. Session logs double as delivery documentation.

Spaces — megacorp-data — dev02@brightpath-agency — 4 panes

Claude Code ~/megacorp-data

Claude Building the ETL pipeline for the warehouse.

Created pipeline stages:
extract/salesforce.ts
transform/normalize.ts
load/bigquery.ts

> add retry logic for API rate limits_

Codex CLI ~/megacorp-data

Codex Analyzing query performance...

Slow queries identified:
SELECT * FROM orders JOIN... (2.3s)
GROUP BY with no index (1.8s)

Suggested indexes generated.
Applied 2 optimizations ✓

Shell ~/megacorp-data

$ k6 run load-test.js --vus 200

scenarios: default
200 VUs, 5m duration

http_reqs: 48,291 (161/s)
http_req_duration: avg=124ms p95=340ms
$ _

Shell ~/megacorp-data

$ git log --oneline -5

a3f8e21 add retry logic for ETL
7b2c4d9 optimize BigQuery loader
1e5f8a3 add Salesforce extractor
9c4d2b1 init pipeline scaffold
f82a1c7 project setup
$ _

[workspace] megacorp-data: 2 consultants active | sessions today: 14 | files touched: 38

Built for teams that ship with AI

Your consultants, their infrastructure

One workspace, every engineer

Every client, fully isolated

Ready to get started?