Elastic AI-assisted workflow live

Connect the Elastic AI Assistant to approved language models with governed data access and verified contextual prompt workflows. Fixed-scope platform delivery funded directly from your advisory credit pool, with top-ups available anytime.

Discuss this deliverable with a specialist

Outcome

One approved AI-assisted Elastic workflow is configured, secured, tested and usable for its intended users.


Overview

A defined Elastic task would benefit from repeatable AI assistance but needs defined access, security and an accepted operating workflow. This deliverable integrates the Elastic AI Assistant (for Observability or Security) with an approved Large Language Model provider, enforcing data redaction and access boundaries while validating interactive workflows for day-to-day operations.

Service request options

Option Credits Scope & objective Boundary
Standard 9,000 One defined workflow using approved data, model or inference access and supported Elastic AI capabilities. Additional workflows are separate Service Requests. Requires Delivery approval.

Scope

  • Configure the approved AI-assisted workflow within Kibana.
  • Establish connectivity to an approved Large Language Model (LLM) or inference provider.
  • Configure data sanitisation, field redaction and security boundaries.
  • Define contextual prompts and user interaction paths for the target operational workflow.
  • Test workflow scenarios to evaluate assistant responses and safety controls.
  • Document system configurations, user guidelines and completion evidence.

Customer responsibilities

  • Specify and approve the target operational use case and intended user audience.
  • Provide API access, credentials and commercial billing accounts for the approved LLM or inference provider.
  • Formally approve data access parameters, data masking and field redaction rules.
  • Maintain appropriate Elastic licence entitlements supporting AI Assistant.
  • Provide representative scenarios, queries or alert data for testing.
  • Participate in workflow review and provide formal acceptance sign-off.

Prerequisites and assumptions

Security policies, data access rules, intended users, model or inference access and the permitted action boundary must be approved before activation. Delivery approval is required.

Out of scope

  • Paying for or providing third-party LLM inference, token or cloud consumption charges.
  • Unrestricted or company-wide unmanaged AI adoption programmes.
  • Developing, training or fine-tuning custom foundation or open-source models.
  • Autonomous agent configurations that execute unapproved, destructive system changes.
  • Delivering multiple distinct operational workflows in a single request.
  • Continuous model operations, prompt drift engineering or managed chatbot operations.

Delivery approach

  • 1. Qualify use case and security boundary: Confirm target operational tasks, model provider, access rules and data masking needs.
  • 2. Configure workflow: Connect the AI connector, set up assistant preferences and map target indices.
  • 3. Apply access and safety controls: Configure role-based permissions, data filtering and redaction profiles.
  • 4. Test scenario: Execute typical operational queries and alert investigations using the assistant.
  • 5. Verify user result: Confirm that assistant responses are accurate, safe, helpful and conform to boundaries.
  • 6. Complete: Compile configuration manifests, export prompt documentation and close handover.

Completion and evidence

The defined workflow produces the agreed assisted result for the approved scenario, access and safety controls work, and the intended user accepts the workflow. Evidence delivered at sign-off includes:

  • Configured AI Assistant connector and workflow documentation.
  • Approved data access boundary and redaction filter records.
  • Access, security and role validation records.
  • Scenario test logs demonstrating valid assistant responses.
  • Documented exceptions (if any) and formal user acceptance sign-off.

Delivery credits

This deliverable is available only to customers with Elastic Specialist Advisory. Once the Service Advisor and Delivery have qualified the Service Request, the applicable credits are allocated from the customer's available credit pool. Outcome Credits represent the defined result, not consulting hours.

If the scope changes

If security review rejects external model transmission, target data contains unsanitisable sensitive fields requiring broader data architecture changes, or requirements expand into unconstrained automation, Observata pauses the affected work. The Service Advisor and Delivery then requalify the request to another option, another deliverable or a larger governed engagement before additional scope is performed or additional credits are consumed.

How to use this deliverable

  • 1. Propose the operational triage or summarisation task with your advisor in your working session.
  • 2. Your Service Advisor reviews model access, security boundaries and verifies 9,000-credit scope.
  • 3. The Service Request is activated, locking the permitted action boundary and credits upfront.
  • 4. Delivery connects the assistant, enforces redaction guardrails, verifies usability and hands over the workflow.

Delivery reference

  • Service family: HYPR Serve
  • Outcome ID: EO-AIWF
  • Specification Version: 1.3
  • Effective Date: 23 September 2026

Governing scope: Observata legal centre