Use Cases - Till | Scoped API Keys for AI Agents

Agent Sandboxing

AI agents can enter infinite loops, make unexpected API calls, or burn through tokens faster than anticipated. Till lets you give each agent run a hard budget it cannot exceed.

Set activation limits to cap total API calls
Set dollar limits to prevent surprise bills
Get clear errors when limits are reached
Revoke keys instantly if something goes wrong

Read the docs

# Create a sandboxed key for an agent run
key = till.create_key(
    provider="openai",
    upstream_key=os.environ["OPENAI_KEY"],
    max_activations=50,
    max_spend_cents=500  # $5 max
)

# Agent uses the scoped key
agent.run(api_key=key.scoped_key)

# Key auto-disables at 50 calls or $5
      

Contractor & Freelancer Access

Need to give a contractor access to your AI APIs? Don't share your real keys. Create a scoped key with exactly the limits they need, and revoke it when the project ends.

Share access without exposing real credentials
Set project-specific budgets
Track usage per contractor
Instant revocation when contracts end

Get started

0

Keys exposed

$50

Budget cap

1,247

Calls tracked

1 click

To revoke

Demo & Trial Environments

Let prospects try your AI-powered product without worrying about abuse. Create demo keys with tight limits that showcase your features without exposing you to risk.

10 free API calls per trial user
Automatic cutoff prevents abuse
Track demo usage across all prospects
Convert high-usage demos to paid plans

View pricing

"We used to give demo users our real API key with rate limiting. One person ran a script and burned $800 in an hour. With Till, we cap demos at $2 and sleep at night."

— Engineering lead at a YC startup

Team Budget Allocation

Allocate AI spend across teams, projects, or environments. Each team gets their own scoped keys with independent limits. No more shared keys, no more surprise bills.

Per-team or per-project budgets
Separate dev/staging/prod limits
Usage visibility by key
Central admin dashboard

Open dashboard

# Allocate budget per team
teams = {
    "ml-team": 10000,    # $100
    "product": 5000,     # $50
    "support": 2000,     # $20
}

for team, budget in teams.items():
    key = till.create_key(
        provider="anthropic",
        label=f"{team}-april",
        max_spend_cents=budget
    )
      

Built for real-world AI operations

Agent Sandboxing

Contractor & Freelancer Access

Demo & Trial Environments

Team Budget Allocation

Which use case fits you?