The platform behind Intelligence

AI Hub, Blueprints, Signals, Rules Engine, and full audit trail: the foundation that makes every agent work, every workflow configurable, and every action traceable.

Your Rules. Always.
AI executes within boundaries you set, in plain language, not code
Intelligence is not a black box. You write the instructions. The AI follows them. When it can't execute safely, it stops and asks for approval. Every action is logged, traceable, and auditable.

AI Hub

April 2026

Your command center. Everything Intelligence does flows through here.

The AI Hub replaces the current Review Queue with a structured, prioritized workspace. One screen tells you what happened, what matters, and what to do next.

Today's priorities

Three counters at a glance:

  • Items to Map: Smart Inbox items needing customer/contact mapping
  • Items to Review: agent checkpoints waiting for your team's approval
  • AI Suggestions: insights and recommendations from Account, Part, and Inventory signals

Next best action

The top 3 most critical items requiring your attention, ranked by urgency and confidence score. Intelligence decides what matters most. You decide what to do about it.

Review queue

Card-based view replacing the old table layout. Each card shows a priority badge, confidence score, short reasoning, and one-click actions: Review, Approve, Reject, Map. Filter by My Items or All Items.

Rules engine: your rules run first

Live

Intelligence is not a black box. You write the rules. AI executes within them.

Every agent follows instructions you configure in plain language, no code, no technical skills required. The system is hybrid by design: deterministic rules run first, AI enhances and fills gaps, and your team approves at checkpoints you define.

How rules work

  • Structured instructions: describe the behavior you want in plain language. No code required.
  • Sequenced execution: instructions run in order, with state tracking between steps
  • Condition checks: rules run only when your criteria are met, or always execute when flagged unconditional
  • Approval gates: review checkpoints where the agent pauses for your team before proceeding

Your team stays in control

  • Review checkpoints: agent pauses and waits for your team's approval before proceeding
  • Manual price protection: the agent never overwrites a price your team already entered
  • Enable/disable per feature: every agent and capability can be turned on or off independently
  • Request Review: any step can require employee sign-off before the workflow continues

The key principle: the AI never overrides your team's decisions. If your team has already priced a line, the agent leaves it alone. Rules define exactly how the agent behaves in every scenario.

Models: the knowledge layer

Agents execute workflows. Models provide the knowledge those agents draw from. Each model is configurable to your business and improves as your transaction history grows.

Live

Pricing Model

Defines how Intelligence prices parts. You configure the factors: inventory cost, historical sale price, customer contract pricing, market data, and margin rules. Each factor can be weighted and combined. The model produces a suggested price with full rationale so your team can see exactly why it priced what it priced.

How to configure:

  • Select which pricing factors to include
  • Write advanced pricing instructions for complex scenarios
  • Add customer-specific contract pricing overrides
  • Write advanced pricing instructions for complex scenarios

Used by: Quote Agent, Pricing Agent

Live

Source of Supply Model

Defines how Intelligence searches and ranks inventory. Configurable search criteria determine which locations to check, how to match condition codes, what trace documentation is required, and how to prioritize by lead time and availability. The model also incorporates customer-specific requirements from the quote request and from the account record, ensuring the match reflects what the customer actually needs.

How to configure:

  • Set condition code matching rules
  • Incorporate customer requirements from quote requests and account records
  • Define trace documentation requirements
  • Configure lead time ranking and tie-breakers
  • Set search scope by location, warehouse, or business line

Used by: Quote Agent, RFQ Agent, Work Order Agent

April 2026

Technician Suggestion Model

Defines how Intelligence recommends technicians for work order tasks. Matches technician certifications, qualifications, skills, and availability against task requirements to suggest the best assignment.

How it helps:

  • Matches certifications and qualifications to task requirements
  • Considers technician availability and current workload
  • Avoids duplicate assignments
  • Works across all work order statuses (except Shipped/Canceled)

Used by: Work Order Agent

April 2026

MRO Planning Model

Defines how Intelligence assists with work order planning and scheduling. Uses historical performance data on similar work scopes to validate dates, estimate timelines, and suggest task sequencing.

How it helps:

  • Date validation against realistic completion timelines
  • Task sequencing based on historical patterns
  • Timeline estimation informed by similar past work orders
  • Proactive identification of scheduling conflicts

Used by: Work Order Agent

April 2026

Vendor Suggestion Model

Defines how Intelligence identifies the best vendors to source from. Analyzes your vendor transaction history, past pricing, performance metrics, lead times, and availability patterns to recommend who to RFQ when you don't have a part in stock.

How it helps:

  • Recommends vendors based on real transaction data, not guesswork
  • Considers vendor performance and reliability history
  • Learns from your purchasing patterns over time
  • Learns from your purchasing patterns over time

Used by: RFQ Agent

Signals: the nervous system

April 2026

Agents don't work in isolation. Signals connect them.

When an agent completes a task or encounters a condition, it emits a signal: a structured, machine-readable event. Other agents listen for signals and act on them. This is how Intelligence coordinates work without you managing the handoff.

How signals work

Every signal carries: a category (Map, Review, Suggest), a priority level, a confidence score, the source agent, and a human-readable message. Signals flow into the AI Hub and can trigger downstream agents.

The mesh: agents cooperating

Signals enable the mesh architecture, with agents triggering each other across ERP processes:

1

Quote Agent runs

Source of Supply searches on-hand inventory, alternates, on order, on repair, and secondary locations. No suitable stock found

2

NO_STOCK signal emitted

A deterministic gate confirms inventory is genuinely unavailable before the signal fires

3

RFQ Agent picks up the signal

Identifies qualified vendors and prepares outbound RFQs (sends automatically or pauses for approval)

4

Vendor responds

Smart Inbox parses the reply, matches to the original RFQ, updates the record

5

Quote Agent resumes

With newly sourced supply, the quote can be completed and sent

This is not a workflow you configure step by step. It's agents coordinating autonomously within the rules you've set.

Blueprints

April 2026

Start from a working configuration, not a blank screen.

Blueprints are exportable and importable packages of Intelligence configuration: pricing models, agent instruction sets, and workflow setups. Instead of configuring every agent from scratch, import a Blueprint designed for your business type and customize from there.

Import & customize

Import a Blueprint package into your AvSight org. Pricing rules, agent instructions, and sequence configurations come pre-built. Customize them for your specific operation: adjust thresholds, add conditions, change email templates.

Export & share

Built a configuration that works? Export it as a Blueprint. Share across orgs, teams, or use it as a backup before making changes. Your configuration becomes reusable infrastructure.

Business-type templates

Start with configurations designed for parts distributors, MROs, or government contractors. Each template reflects the common workflows and rules for that business type. You refine, not rebuild.

Intelligence Console

Live

Unified administration for all agents, models, and insights.

The Intelligence Console is where administrators manage the entire Intelligence platform: configure agents, set up models, adjust instructions, enable or disable features, and monitor system health. One screen to control everything Intelligence does across your organization.

Intelligence Activity Log

April 2026

Full traceability. Every action, every decision, every agent, every record.

Why this matters

In aerospace, traceability isn't optional. When an auditor, a Quality VP, or a customer asks how a decision was made, you need a complete paper trail. The Intelligence Activity Log provides that for every action every agent takes. A unique trace ID propagates through the entire execution chain, so you can follow a single transaction from the moment an email arrived through pricing, approval, and shipment.

Universal Intelligence panel

Embedded directly on Customer Quote, Sales Order, RFQ, and Work Order records. Shows all agent activity for that record in chronological order. You don't search a separate log. The full history is right on the record, next to the data it acted on.

Trace ID propagation

Every agent action carries a unique trace ID that connects the entire execution chain. When the Quote Agent calls the Pricing Model, which triggers a follow-up email, which leads to a sales order, one trace ID links every step. Click any point in the chain and see the full story.

Admin activity screen

Administrators get a centralized view of all Intelligence activity across the org. Filter by agent type, date range, account, or trace ID. Built for compliance review, troubleshooting, and understanding exactly what Intelligence is doing across your operation.

Security & privacy

Live

🔒

Purpose-built for aviation

Not a generic AI adapted for your industry. Built specifically for aviation workflows, terminology, and data structures.

🛡

Your data stays yours

Never stored externally. Never used to train AI models. Your data stays in your environment.

🔍

Full audit trail

Every action logged with trace IDs. Every feature independently enable/disable controlled.

For IT directors

Intelligence runs inside your existing Salesforce org. No new system, no new vendor, no new security review. Same profiles, permission sets, and field-level security your team already manages. Purpose-built for aviation, not a generic AI platform adapted for your industry. Your team gets AI capabilities without managing a new platform.

See the platform in action

AI Hub, Blueprints, Rules Engine, and full audit trail, built for aviation operations.

Schedule a demo Take the self-guided tour
PROTOTYPE