All AI features available in the Legistify platform are powered by Large Language Models (LLMs) accessed through AWS Bedrock. These models operate entirely within our secure AWS cloud infrastructure, ensuring that AI capabilities are delivered while maintaining strong data protection and regulatory compliance.

This architecture allows Legistify to provide advanced AI-powered capabilities such as contract analysis, drafting assistance, and document insights without exposing client data to external AI platforms.

Legistify uses state-of-the-art Large Language Models (LLMs) available through AWS Bedrock.

These models are foundation models trained on large-scale and diverse datasets to support advanced natural language processing tasks such as:

Model outputs are guided through controlled prompt design and application-level logic to ensure the responses are relevant, consistent, and aligned with legal workflows.

Legistify does not build, host, or train custom AI models outside AWS Bedrock for these use cases.

All AI workloads run securely within Legistify's AWS cloud environment.

Key architectural characteristics include:

This architecture ensures that AI capabilities remain fully integrated within Legistify's secure platform environment.

Legistify follows strict security and privacy practices when processing data through AI features.

Client data is processed only for the specific AI request initiated by the user.

Key principles include:

This ensures that client data remains private and is not reused beyond the immediate request.

Legistify protects data both in transit and at rest.

Security controls include:

These measures ensure that sensitive contract and legal data remains securely protected.

Strict access management policies are enforced across the platform.

Controls include:

These mechanisms ensure that only authorised systems and personnel can interact with AI services.

All AI interactions occur within a private and secure network environment.

Security measures include:

This prevents unauthorised external access to AI services.

To further strengthen responsible AI usage on the platform, Legistify has implemented AWS Bedrock Guardrails as an additional safeguard layer on top of existing security, privacy, and access controls.

Guardrails help ensure that both user inputs and AI-generated responses remain aligned with Legistify's responsible AI policies and are appropriate for legal and enterprise workflows.

AWS Bedrock Guardrails are applied during model inference within Legistify's secure AWS environment — on both input and output.

These controls help:

Guardrails complement existing controls such as prompt design, access management, encryption, and private network isolation.

If a user prompt or document context includes a sensitive identifier such as a PAN (Permanent Account Number), Bedrock Guardrails can mask that value on the input before the request is sent to the LLM. The same type of safeguard can also be applied on the output, so sensitive values are not returned to the user in clear text.

Illustrative example:

User input / document context:
Analyse the agreement linked to PAN ABCDE1234F.

What is sent to the LLM (with guardrails):
Analyse the agreement linked to PAN {PAN}.

What the user may see in the AI response (with guardrails):
The agreement is linked to PAN {PAN}.

In this flow, the actual PAN value is not passed to the LLM in clear text, and it is also not exposed in the final response. The rest of the request and answer can still support the legal workflow without revealing the sensitive identifier.

Legistify manages guardrails through AWS Bedrock's native governance workflow:

This ensures that safety policies can evolve over time while remaining under Legistify's operational control.

Legistify's AI architecture is designed to deliver powerful AI capabilities while maintaining strict standards for security, privacy, and compliance.

By leveraging AWS Bedrock within a secure AWS infrastructure, Legistify ensures that AI-powered features can analyse documents, generate insights, and assist legal workflows without compromising the confidentiality of client data. In addition, AWS Bedrock Guardrails provide an extra layer of responsible AI controls on both inputs and outputs, helping ensure sensitive information is protected and model responses remain aligned with Legistify's safety and usage policies.

This architecture allows organisations to confidently use AI within the platform while maintaining full control over their data.

AWS Bedrock Guardrails are policy-based safeguards applied during AI inference to help filter or mask unsafe, inappropriate, or sensitive content in both model inputs and outputs.

No. Where sensitive-information filters are configured, values such as a PAN can be masked on the input before the request reaches the LLM. Guardrails can also mask sensitive values in the output, so they are not returned to the user in clear text.

No. Guardrails run within Legistify's AWS Bedrock environment. Client data continues to remain inside Legistify's AWS infrastructure throughout the AI processing lifecycle.

No. Guardrails are an additional responsible AI layer. They work alongside existing security controls such as encryption, IAM-based access management, private networking, and data retention policies.

Read Frequently Asked Questions