AskSage — Kearney AI Sandbox

What is AskSage?

AskSage is a secure AI platform built for government and defense use. It gives you access to the same AI models you may already use personally, through a channel authorized for sensitive federal work.

When you use ChatGPT directly on the open internet, your conversations may be used to train future models and are not protected under any government data handling agreement. AskSage solves that problem. It is authorized at DoD Impact Level 5, meaning it meets the security and data handling requirements for sensitive but unclassified government information.

Why this matters for Kearney

Any AI work on a DoD-scoped engagement must go through AskSage. Using consumer AI tools like the public ChatGPT website for client work is a policy violation and a potential FAR compliance issue. AskSage is the compliant path.

What models are available through AskSage?

AskSage provides access to multiple AI models through a single credential. You do not need separate accounts or API keys for each one.

GPT-4o

OpenAI's flagship model. Strong across analysis, writing, summarization, and reasoning. Good default choice for most tasks.

Best for: general analysis

o1 / o3 Reasoning Models

OpenAI's reasoning-focused models. Slower but more accurate on multi-step problems, logic, and complex financial calculations.

Best for: complex reasoning

Gemini

Google's AI model family. Strong at document analysis and long-context tasks. Useful when working with lengthy reports or policy documents.

Best for: long documents

Others

Additional models are available depending on your AskSage tenant configuration. Check with your team lead for the current model list.

Confirm with team lead

How to access AskSage from the sandbox

AskSage is accessed programmatically from your SageMaker Studio notebook environment, not through a separate website. The API key is stored securely in AWS Secrets Manager and is available to your notebook session without you having to manage it manually.

Open your SageMaker Studio notebook

Log into the AWS Console with your sandbox credentials, navigate to SageMaker, and open your Studio workspace. From there, open a new Jupyter notebook or use an existing one.

Load the AskSage API key from Secrets Manager

The key is stored securely. Retrieve it in your notebook using the Python code your team lead provides.

Make your first call

Once you have the key, you can call AskSage using the asksageclient Python library. Your team lead can provide a working starter notebook. Start with a simple test prompt before moving to more complex analysis.

Key not yet populated

The AskSage API key shell has been provisioned in the sandbox but the key value itself has not been populated yet. Contact your team lead to obtain the AskSage credential and have it added before you attempt to use it.

When to use AskSage vs. the Prompt Assistant

Use the Prompt Engineering Assistant (on the home page) when:

You want to turn a rough idea into a well-structured prompt you can then use in ChatGPT or Gemini. This is the right starting point for anyone new to AI tools. It requires no technical knowledge and works directly in your browser.

Use AskSage when:

You are working on a DoD-scoped client engagement and need to run AI analysis in a compliant environment. You are building a repeatable workflow that will run on data regularly rather than a one-off question. You need access to the more powerful reasoning models like o1 or o3 for complex financial analysis.

Questions about AskSage access

Contact sandbox-support@kearneyco.com or your team lead. Confirm the AskSage tenant details (URL endpoint and tenant assignment) with your team lead before attempting to connect.