Solutions

Apply governed intelligence to the workflows that matter most.

EL-Ai-NE is designed for recurring enterprise work that spans people, systems, endpoints, networks, data, approvals, actions, and reporting.

Use cases

Twelve use cases, each in a few plain steps.

Every use case says who it is for, the problem, how it works, what you get and where the data stays. The status column tells you what you can see in a proof of concept today and what is switched on later or planned.

Available now

Switched on by default in the current baseline build. You can see it working in a proof of concept.

Included, enabled per qualification

Built into the platform and shipped switched off. An accountable owner turns it on for a deployment once it is qualified for that deployment's profile.

Planned

Specified in Baseline v1.0.0 and on the delivery path. Not yet part of a deployable build.

01A private AI assistant that keeps company data on siteStaff want an AI assistant, but prompts and documents must not leave the business.

Who it is for

Any member of staff; the security and data owner.

How it works

  1. A user signs in and asks a question in the chat or from an application that calls the standard chat interface.
  2. The gateway checks who is asking and what policy allows, then answers from a local model, reusing a cached answer when the caller is allowed to see it.
  3. Usage is counted per request and route; an evidence record is written.

What you get

Useful answers with no external call, and a record of who asked what, by which route.

Where the data stays

Prompts are processed on your infrastructure. Nothing leaves unless an external route is explicitly admitted.

Status at Baseline v1.0.0

  • Available nowSign-in, chat, local model serving, gateway stages, answer cache, usage metering.
  • Included, enabled per qualificationAI routing console.
  • PlannedSingle sign-on; loading your own document corpus; external providers.
02Network troubleshooting with evidence, not guessworkFinding out whether a branch is reachable, and why not, means stitching together several tools by hand.

Who it is for

Network operations engineers.

How it works

  1. Read-only sources feed topology, path, configuration and health observations into the operational record.
  2. The engineer asks a path, intent, exposure or health question. The answer names its sources and their age.
  3. Where evidence is missing the answer says "unknown"; it never implies healthy.
  4. A recommended next step is shown for a person to act on.

What you get

A reachable, blocked or unknown conclusion supported by named sources, with contradictions kept visible.

Where the data stays

Network facts stay in the operational record in your control cell; nothing is changed on the network.

Status at Baseline v1.0.0

  • Available nowOperational record and views; recommendations (shown, never executed).
  • Included, enabled per qualificationNetwork Truth (read-only); SNMP and lab network sources.
  • PlannedActive probes, write connectors and change verification.
03One approved action, with proof that it workedAutomation that "ran" is not the same as a change that worked, and nobody can show who approved it.

Who it is for

Operators and approvers.

How it works

  1. A workflow proposes one bounded action, for example a readiness refresh that changes no policy, routing or security posture.
  2. The action is checked against policy and given a consequence class. An approver who is not the requester approves that exact request.
  3. A connector performs it once with a short-lived credential.
  4. An independent check reads the result. Only then is the action recorded as successful, with a rollback path on record.

What you get

A complete receipt: request, approval, execution, verification and rollback evidence.

Where the data stays

Receipts live in the action journal for one to seven years, by consequence.

Status at Baseline v1.0.0

  • Available nowRollback rule, no self-approval, audit recording, read-only reconciliation.
  • Included, enabled per qualificationAction Center actions. Every action is refused until a signed action allowlist is installed.
  • PlannedWrite connectors for network devices.
04Advisory impact review for a branch estateA new vendor advisory arrives and someone has to work out which sites are affected.

Who it is for

Operations and security teams.

How it works

  1. The approved advisory is compared with the authorized inventory in the operational record.
  2. An impact summary lists affected sites with links to the evidence.
  3. A ticket is proposed and stays in draft until it is admitted and its creation is independently verified.

What you get

An evidence-linked impact assessment and a draft ticket, not an unreviewed mass change.

Where the data stays

Advisory and inventory evidence stay in your environment with their source rights.

Status at Baseline v1.0.0

  • Available nowOperational record; recommendations.
  • Included, enabled per qualificationTicket creation through the Action Center.
  • PlannedGenerated impact pages (Governed Web Experiences).
05Evidence for auditors and customers on demandAssurance evidence is assembled by hand from screenshots and spreadsheets.

Who it is for

Governance, risk and compliance teams; auditors.

How it works

  1. Every governed run already writes signed, append-only governance records.
  2. The Trust Center evaluates control packs against that evidence and shows a scorecard with stale items marked.
  3. A signed evidence bundle is exported for one tenant.

What you get

Evidence that can be verified independently, tied to the release that produced it.

Where the data stays

Evidence stays in the governance ledger; exports carry references and proofs, not raw sensitive content.

Status at Baseline v1.0.0

  • Available nowGovernance ledger and proofs; audit recording.
  • Included, enabled per qualificationTrust Center, evidence bundle, privacy registers, assurance views.
  • PlannedCustomer-facing trust portal and questionnaire service.
06Deletion that actually reaches every copyA document is deleted at the source but lives on in caches, indexes and AI answers.

Who it is for

Privacy and data owners.

How it works

  1. The owner accepts the deletion or rights withdrawal.
  2. A signed tombstone stops the item being served anywhere, first.
  3. Caches, indexes and derived content are invalidated; cleanup follows; evidence of deletion is kept.
  4. A later restore reapplies current tombstones before any data can be read.

What you get

Withdrawn data cannot resurface through a cache, an answer or a backup.

Where the data stays

Only the tombstone and the deletion evidence remain.

Status at Baseline v1.0.0

  • Available nowErasure and tombstones; a tombstoned item is never served again.
  • Included, enabled per qualificationData subject request workflow; retention bundles; signed backup manifest.
  • PlannedEnd-to-end erasure evidence across every derivative is part of production qualification.
07See what AI costs, by routeAI spend is spread across providers, local hardware and teams, with no shared unit.

Who it is for

Finance and platform owners.

How it works

  1. Each request records exact requests, tokens and the route used.
  2. A rate card turns usage into money figures.
  3. Measured cost, estimates and modeled savings are reported separately.

What you get

Cost per request and route you can defend, without mixing estimates into accounting figures.

Where the data stays

Usage and cost facts are append-only in the cost store.

Status at Baseline v1.0.0

  • Available nowUsage metering and cost windows.
  • Included, enabled per qualificationAI routing console; governed reporting.
  • PlannedFinancial close, chargeback and reconciliation.
08Stop everything, safelyWhen something looks wrong, AI activity has to stop without corrupting work in flight.

Who it is for

Security and platform administrators.

How it works

  1. A verified human administrator asserts the emergency stop with a stated reason.
  2. The stop reaches the gateway, agents, background work and the cache.
  3. Work whose outcome is uncertain is held for reconciliation. The stop is cleared by a human when it is safe.

What you get

A controlled halt with a record of who stopped what and why.

Where the data stays

The stop and its reason are written to the governance ledger.

Status at Baseline v1.0.0

  • Available nowPolicy epoch and emergency stop.
09A branch that keeps working when the WAN is downCloud-dependent tools stop when the link drops.

Who it is for

Site managers and operations.

How it works

  1. The site holds signed policy, artifacts and a local model path, all inside a finite lease.
  2. During the outage, local reads and low-consequence workflows continue; evidence queues locally.
  3. Nothing gains new permission because of the outage, and the lease cannot be extended by the site.
  4. On reconnection, revocations and tombstones are applied first, then the backlog.

What you get

Bounded local continuity and a clean, verifiable rejoin.

Where the data stays

Evidence waits in the site outbox and drains to the region or control cell.

Status at Baseline v1.0.0

  • PlannedQualified disconnected operation, cross-site state convergence and offline leases. No offline authority is issued until a lease duration is approved.
10Guide staff from unapproved AI to an approved assistantStaff paste company data into public AI tools because there is no approved alternative.

Who it is for

Security teams; every employee.

How it works

  1. Within the managed-work boundary, use of unapproved AI and SaaS is discovered.
  2. Policy returns permit, warn, redact, block or redirect.
  3. A redirect opens the approved EL-Ai-NE assistant with the permitted context preserved.

What you get

Safe adoption instead of blanket blocking, with the employee told what happened and why.

Where the data stays

Personal content is minimized. Personal applications, messages and devices outside the managed work context are not monitored.

Status at Baseline v1.0.0

  • PlannedShadow AI discovery and Shadow AI Guard.
11Supplier risk and customer questionnairesSupplier reviews and customer security questionnaires are slow and repetitive.

Who it is for

Security, procurement and sales engineering.

How it works

  1. A small inventory of AI suppliers and AI uses is recorded, with one trust portal and a readiness snapshot.
  2. An imported questionnaire is mapped to approved evidence; suggested answers cite it.
  3. An accountable reviewer approves each answer before it is published.

What you get

Faster, evidence-linked answers that are invalidated when the evidence changes.

Where the data stays

Confidential source documents are not redistributed because a questionnaire references them.

Status at Baseline v1.0.0

  • PlannedAssurance Exchange, supplier and exposure services (the Assurance Free and Risk plans).
12Endpoint inventory you can verifyInventory is self-reported and gaps are invisible.

Who it is for

Endpoint and IT operations.

How it works

  1. A managed endpoint enrolls with a device-held key.
  2. Inventory and posture arrive in sequenced batches; gaps are recorded as gaps.
  3. A single reversible action, collect inventory, runs under a signed lease.

What you get

Posture that separates reported, observed and verified state.

Where the data stays

Collection on personal devices is confined to the managed work profile.

Status at Baseline v1.0.0

  • Included, enabled per qualificationEndpoint estate and telemetry (partial: the console enrolls only its own host); collect-inventory action.
  • PlannedThe full endpoint client fleet.

Risk and assurance solutions Planned

Connect external cyber risk, Shadow AI, trust evidence, and remediation to the systems and workflows they can affect.

Deploy only the subset your program needs, then expand to broader governance, knowledge, endpoint, network, or Managed Fabric operations.

AI supplier and ecosystem risk

Continuous provider inventory, fourth-party dependencies, contract controls, concentration, continuity, and route impact.

External AI exposure

Customer-approved discovery of public AI services, tool servers, data stores, credentials, certificates, and forgotten environments.

Shadow AI Guard

Managed-work discovery, sensitive-data control, contextual guidance, and redirection to approved EL-Ai-NE AI routes.

Assurance Exchange

Trust portals, access protection, questionnaires, approved answers, and current release-bound evidence.

Risk cases and automation

Assignments, exceptions, approvals, integrations, typed remediation, verification, and evidence-backed closure.

Explainable posture

Multidimensional risk with freshness, confidence, coverage, exceptions, methods, trends, and recommended action.

AI governance operations Planned

Coordinate inventory, impact, oversight, transparency, monitoring, and assurance.

EL-Ai-NE provides a technical and evidence layer for organizations building a repeatable AI governance program across internal, third-party, local, regional, and approved external AI systems.

G1

AI system inventory

Discover, reconcile, assign owners, track material changes, and connect use cases to models, data, vendors, deployments, and affected populations.

Outcome: current accountability and fewer unmanaged AI systems.
G2

Impact and oversight

Assess benefits and harms, define interventions, prove competence, and record residual-risk acceptance for the exact use case.

Outcome: meaningful human control with evidence.
G3

Release and post-market assurance

Bind transparency, vendor evidence, monitoring, complaints, audit, corrective action, and technical qualification to each production release.

Outcome: defensible governance from launch through retirement.

Solution patterns

Where the same building blocks apply.

Nine patterns built from the capabilities above. The measures listed are what to track, not promised results. Check the capability status for what each pattern can use today.

01 / SERVICE OPERATIONS

Service desk and employee support

Bring together request, identity, entitlement, endpoint, configuration, knowledge, and telemetry context. EL-Flow manages the resolution process; Action Fabric applies approved changes; verification proves closure.

Potential measures: time to triage, time to resolution, repeat incidents, first-contact resolution, evidence completeness.
02 / SECURITY

Security and endpoint response

Correlate endpoint, identity, network, vulnerability, case, and business context. Recommend a safe response, issue a bounded endpoint or system action, and verify containment or remediation.

Potential measures: investigation time, containment time, verified remediation, analyst handoffs, reporting latency.
03 / NETWORK

Branch, network, and SaaS operations

Join application, user, device, topology, path, SD-WAN, DNS, DHCP, ITSM, and change context. Use out-of-band, endpoint-assisted, or service-insertion deployment patterns as appropriate.

Potential measures: incident isolation time, change success, unknown-path reduction, site continuity, service impact.
04 / KNOWLEDGE

Private enterprise knowledge

Provide permission-filtered answers, evidence-linked summaries, relationship-aware retrieval, approved memory, and reusable context without replacing source systems or training on every interaction.

Potential measures: time to answer, source coverage, citation quality, repeated context avoided, operator adoption.
05 / WORKFLOW

Agent and process operations

Convert requests and events into durable tasks, approvals, typed actions, verification, evidence, cost, and outcome records. Agents collaborate through bounded tasks rather than shared unrestricted authority.

Potential measures: cycle time, human touchpoints, exception rate, verified success, cost per completed case.
06 / REPORTING

Governed operational reporting

Define semantic fields and metrics, carry source watermarks and unknowns, apply masking and classification, and support correction, withdrawal, regeneration, and restatement.

Potential measures: report preparation time, disputed metrics, stale-source exposure, correction response, audit readiness.
07 / DISTRIBUTED SITES

Local-first branch and field operations

Keep eligible inference, retrieval, workflow continuity, cache, and evidence close to the site. Continue bounded work through WAN impairment with signed policy and locally present qualified artifacts.

Potential measures: local completion, WAN-dependent failures, latency, upstream calls avoided, reconnect reconciliation.
08 / MULTILINGUAL

Multilingual operator workflows

Support text, speech, translation, glossaries, and evidence-linked interaction while preserving originals and keeping structured action parameters separate from natural-language rendering.

Potential measures: completion by language, quality exceptions, escalation, operator satisfaction, protected-value preservation.
09 / ECONOMICS

AI cost and capacity operations

Measure provider tokens, local compute, model hosting, network, storage, and workflow economics. Route work according to privacy, quality, latency, reliability, capacity, and cost.

Potential measures: repeated-token reduction, external spend, local utilization, cost per verified action, accepted savings.
Ask EL-Ai-NEPublic website guide
Ask about capabilities, governance, privacy, Brain Trust, risk, deployment, pricing, comparisons, or public resources.