Last verified August 10, 2026

Element451 uses artificial intelligence throughout its platform to help higher education institutions communicate, analyze information, automate work, and support students.

Element451 does not train foundation models from scratch. We build the product layer around third-party AI and speech services, including Bolt Agents, retrieval and grounding, prompts, tools, permissions, workflows, evaluations, and safeguards.

Model assignments can vary by feature, environment, and staged rollout. The provider overview below is representative of current production use and is not a permanent or exhaustive model inventory.

We evaluate model and provider changes based on the needs of each product workflow. Changes can be introduced through staged rollouts, with fallback behavior where appropriate.

Element451 maintains a SOC 2 Type II audited security program. Platform data is encrypted in transit and at rest, and AI provider credentials remain server-side rather than being exposed to end users.

Learn more in SOC2 Type II Certification and Security and Compliance.

Element451 does not use institution or student data to train generalized foundation models, and we do not fine-tune shared models using customer data.

Third-party AI services are accessed through business APIs or managed cloud services under service-specific terms and agreements. Provider processing and retention terms can differ by service, so they should not be described as one universal policy across every provider.

The information processed depends on the feature being used. It can include:

Element451 designs each integration to provide the context needed for the requested task. The exact context varies because a voice conversation, a deduplication comparison, and an application analysis do not use the same data.

Element451 may retain conversations, transcripts, recordings, generated outputs, and other AI results as part of normal product functionality and in accordance with product settings, customer agreements, and platform retention practices.

Provider-side processing and retention vary by service and are governed by the applicable provider agreement. They should not be described as immediate processing with no retention in every case.

Your institution owns its data and controls how it is configured and used in Element451. Element451 processes that data to provide the contracted service. Specific privacy roles and responsibilities are governed by the applicable customer agreement and may differ for limited business operations.

Bolt Agents, default prompts, tools, and workflows are designed around higher education use cases. Institutions configure the knowledge, instructions, Custom Skills, permissions, teams, channels, and handoff behavior that shape each deployment.

Bolt Agents are designed to ground factual answers in the sources available to the feature, including the institution's Knowledge Hub and permitted Element451 or CRM context. Accurate, complete, and current institutional content remains essential.

Depending on the feature, Element451 uses controls such as:

These controls reduce risk, but they do not guarantee that an AI system will always be correct, remain in scope, resist every manipulation attempt, or identify every sensitive message.

When Bolt Agent Content Moderation is enabled for supported conversations, it can flag predefined categories such as hate, harassment, violence, self-harm, prompt engineering, and conflicting behavior instructions. Institutions can configure messages and actions for supported categories and can use Conversation Rules to assign flagged conversations for review.

Self-harm-related flags do not have default actions. Institutions must configure the escalation, assignment, notification, and review workflow they want. Automated flags and email notifications do not guarantee that every concern will be detected, seen, or acted upon, so institutions should maintain an internal monitoring process.

AI output can be inaccurate, incomplete, or unsupported even when grounding and guardrails are used. No large language model is 100 percent accurate.

Institutions should test AI configurations before launch, monitor production behavior, and use human review for sensitive or high-impact decisions. AI-assisted analysis should support authorized staff rather than replace institutional judgment.

Learn more in Element451 Bolt AI and the Problem of Hallucinations.

Element451 provides an application layer around third-party AI services, including retrieval, permissions, tools, workflows, logs, and configurable moderation. This reduces the work required to assemble and operate those components independently and supports more consistent deployment across higher education workflows.

A managed platform reduces operational complexity, but it does not eliminate AI risk or replace institutional governance, testing, monitoring, and human oversight.

Element451 surfaces Bolt Agent identity in supported product experiences. Institutions can customize agent names, greetings, instructions, and channel copy, so they should review each deployed experience to ensure AI use is clearly disclosed where required.

For inbound calls answered by a Bolt Agent, the default greeting identifies the agent as AI and states that the call is on a recorded line. When an institution configures a custom greeting, that greeting replaces the default. Institutions should preserve the appropriate AI and recording disclosure in custom greetings.

Messenger voice offers two optional transparency settings:

Voice recordings and transcripts are stored in the conversation thread. Phone calling uses separate recording settings, including whether recording is allowed or enabled by default.

Institutions are responsible for configuring disclosures, recording settings, consent experiences, and internal practices that meet their requirements.

Learn more in Advanced Voice Mode for Bolt Agents and Phone Calling.

Element451 is committed to improving the security, transparency, reliability, and responsible use of its AI systems. As providers, models, and product behavior change, we will continue updating our controls and documentation to reflect the current platform.