Frequently asked questions

Yes. Ejento includes a no-code AI agent builder and visual workflow editor. Business users configure agents, connect data sources, and set guardrails without writing code. Engineers can also access the full platform via API for deeper integrations. Both paths produce agents with the same governance controls applied.

Most teams go live with their first assistant in under a day. You describe what you want the assistant to do, set access controls, connect your data sources, and publish. No engineering is required. The platform scaffolds the project automatically and walks you through each step.

Ejento is model-agnostic. It works with Claude, GPT-4, Gemini, and other LLMs your team prefers. You can run different models across different AI assistants and switch at any time without rebuilding. A/B model testing lets you compare performance on your own assistant evaluation set before committing to a change.

Yes. Ejento provides a one-line embed snippet that drops an AI assistant into any website, help center, or internal portal with no coding required. SSO passthrough keeps users authenticated. You can also call assistants programmatically via REST API or integrate via SDK for deeper workflows.

Ejento connects to SharePoint, Confluence, Google Drive, Notion, Slack, Dropbox, Box, OneDrive, HubSpot, any internal API or document store, and more. Sources are chunked, embedded, and indexed automatically. Re-indexing runs on a schedule or on demand when your documents change.

ChatGPT Enterprise and Microsoft Copilot are chat tools with AI built in. Ejento is an AI agent governance platform. It gives every agent a defined identity, scoped access, validated context, enforced guardrails, and an audit trail. It runs inside your own cloud (not shared infrastructure), supports any LLM, and is built for the accountability requirements of regulated industries.

No. Ejento is deployed inside your own AWS, Azure, or Google Cloud. There is no shared backend, no multi-tenant infrastructure, and no data stored on Ejento servers. Your environment is entirely separated out, and you retain full control over networking, access, and data residency.

No. Ejento is model-agnostic. Different AI assistants can run different models, and you can switch providers at any time without rebuilding. You can also run A/B tests across models on the same assistant evaluation set and pick the best option for each use case based on your own quality metrics.

Ejento has cleared InfoSec acceptance reviews with enterprise customers across finance, healthcare, and government. It is SOC 2 Type II ready, GDPR-compliant, and HIPAA-ready. RBAC (Role-Based Access Control), audit logs, PII (Personally Identifiable Information) redaction, and LLM penetration testing are built into the platform, not optional add-ons. Ejento can provide documentation packages for your InfoSec team.

Your prompts, documents, and agent outputs never leave your cloud environment, not in transit, not at rest. There is no shared backend, no multi-tenant infrastructure, and no data used for model training. Ejento runs inside your AWS, Azure, or Google Cloud VPC with private endpoints and firewall controls you own entirely.

No. Ejento runs inside your own virtual private cloud so your data, prompts, and AI agent outputs are never sent to Ejento's infrastructure. There is no shared backend, and no data is used for model training by Ejento or by the underlying model providers, since inference stays within your cloud perimeter.

Ejento is SOC 2 Type II ready, GDPR-compliant, HIPAA-ready, and CASA certified. Compliance documentation packages are available for your InfoSec team on request. The platform has cleared InfoSec acceptance reviews with enterprise customers in finance, healthcare, and government.

Ejento integrates with Active Directory, Microsoft Entra ID, Okta, Azure AD, OneLogin, Keycloak, Ping Identity, and any SAML 2.0-compliant provider. SSO passthrough keeps users authenticated when accessing embedded AI assistants inside your internal portals.

Yes. Ejento went through adversarial testing with over 20,000 prompts covering prompt injection, jailbreaking, data extraction, and policy bypass. Automated red team runs execute on every deployment, covering 6 OWASP LLM risk categories out of the box. Block rates, attack logs, and severity breakdowns are visible in the Ejento platform dashboard.

All customer data is encrypted at rest and in transit. Because Ejento runs inside your virtual private cloud, you control the encryption keys and cloud storage configuration directly. The platform applies encryption at the infrastructure level by default, with your existing cloud provider security controls fully intact.

Every AI agent in Ejento has a defined role, scoped data access, and enforced guardrails, not just instructions. The platform logs every action the agent takes: what it read, what it wrote, what it decided, and why. Your compliance team gets a structured audit trail, not just a conversation log.

Ejento agents operate within control layers before they act: jailbreak blocking, content safety, topic guardrails, and PII redaction run on every interaction. When something unexpected occurs, the full decision trace is available for review. Agents can also be paused, rolled back, or decommissioned in a single action without disrupting other agents.

Ejento uses a hierarchical permission model: organization, team, and project. Within each level, roles (Owner, Admin, Editor, Viewer) are scoped to individual agents and knowledge repositories. You can limit exactly which documents an agent can surface and which users can interact with or modify it.

Yes. Every AI agent has an independent staging slot. You build and test updates there while the live version continues running uninterrupted. When you are ready, promotion takes effect immediately with zero downtime. If something goes wrong, rollback to any previous version is instant.

Yes. Ejento tracks token consumption and LLM spend per agent, team, and user. You get monthly cost breakdowns with trend comparison, budget alerts, and automatic cost optimization recommendations. Reports are exportable for stakeholder review.

Every query, response, and agent action is logged with the user identity, timestamp, and full decision trace. Chat logs are searchable and filterable, and upvote and downvote signals are captured per response. Exports are formatted for compliance team review and meet the requirements of regulated industries.

Ejento uses hybrid retrieval, combining dense vector search with BM25 keyword retrieval. AI agents find the right answer whether a user phrases a question precisely or conversationally. Every search result is governed by the agent's RBAC policy. If a user is not authorized to see a document, the agent cannot surface it regardless of how the question is phrased.

Ejento supports over 90 integrations across communication tools, document and data stores, CRMs, HR systems, project management platforms, engineering tools, and SSO providers. New integrations are added every month. If your tool is not listed, you can request it or connect via the custom API connector.

Ejento agents can do both. Agents read from and write to your existing systems via MCP servers and OAuth integrations. They can look up CRM records, create tickets, send notifications, and trigger downstream processes, all within your security perimeter. Tool permissions are configured per agent, and every tool call appears in the audit trail.

Ejento runs evaluations directly on your AI assistants' live chat logs. Responses are scored for accuracy, faithfulness, and hallucination rate. Regression alerts fire automatically when a model swap or prompt change hurts quality. A/B model testing lets you compare two LLMs on your own assistant evaluation set before committing, and user thumbs up and down signals feed back into the same loop.