IntuidyIntuidy
Sign InGet Started

Platform / Ontology

Teach your AI how your business works.

Ontology is where Intuidy keeps what it knows about your company: what each table holds, how every number is calculated, and how your team likes things done. Every agent reads it before it answers.

See what it holds
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Your Ontology
How much your agents actually know, what to fix first, and exactly what they receive.
Your agents know 42 of 46 data objects, 31 business terms, and 9 of 10 company facts.

The other 4 objects are silent. Intuidy found them by scanning your data, but a scan only learns names. Until someone writes down what one means, agents get nothing for it at all.

Data objects
42 of 46
4 silent: agents get nothing for them
Business terms
31
Sent to every agent
Company facts
9 of 10
Missing: Current goals
Notes
24
2 waiting for your approval
Your biggest gaps

Undescribed objects, ranked by how much depends on them. Describing these few changes more answers than describing everything else combined.

Scrap log
used by 2 workflows and 1 dashboard · ~212,940 rows
Described just now · every part scrapped, with its reason code and cost
Describe Described
Maintenance work orders
used by 1 workflow and 1 agent · ~18,402 rows
Describe
Supplier certifications
used by 1 workflow · ~1,240 rows
Describe
Start here

Each step on its own makes agents measurably better. You don't need to finish to benefit.

Tell us who you are

A handful of company facts every agent reads before answering. It can be drafted from your website.

Open

Describe the data your agents use most

Start with the tables your workflows actually read. The gap list above is ordered for exactly this.

Open

Define the words your business uses

Metrics especially: write down how each number is calculated, so every agent computes it the same way.

Open

Write down how you work

Rules, preferences, and process notes that aren't stored in any table.

Open

[ What it holds ]

Everything your agents should know, in one place.

A prompt is only as good as what the AI knows behind it. Ontology gives every agent, workflow and dashboard the same understanding of your business.

01

Your data

What each table and record means, the words your business uses, and how data moves between your systems.

02

Your business

Who you are, how your team likes things done, and the facts agents pick up along the way once you approve them.

03

Access

Which outside systems can reach your workflows, and which data may never reach an AI or leave the company.

[ Self-educating ]

It teaches itself. You have the final say.

Intuidy fills in its own ontology as it works. People step in wherever they want to: to confirm what it learned and to fill in what it couldn't.

Intuidy learns on its own

  • ·Scans every system you connect and lists what it finds
  • ·Drafts meanings for tables, fields and metrics
  • ·Flags what to re-check when a source changes

You verify and fill the gaps

  • ·Accept or reject each AI draft before it's saved
  • ·Verify definitions, stamped with your name and the date
  • ·Fill in the gaps, ranked by how much depends on them
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Data Objects
What your tables and records mean. Agents get nothing for an object you have not described.

Your apps, databases, and connected integrations are scanned automatically; agents also fill in meanings as they work. Anything you write here is authoritative. They can refine an empty definition, but never overwrite yours.

Search objects, fields, and definitions…
Status
AllNeeds description4Needs review2Verified27
Data objects (42 of 46 described)
Deep scan Enrich with AI Add object
Scrap log Silent · used by 2 workflows, 1 dashboarddiscovered
managed-db:2:scrap_log

Plant data · ~212,940 rows · used by 2 workflows · 1 dashboard

Details Verify
What does this object represent, in business terms?
Grain
One row per… (what exactly is a single row here?)
Production jobs In prompt Verified Internal
integration:epicor:JobHead

Epicor · ~48,210 rows · 3 joins · used by 2 agents · 6 workflows · 3 dashboards

You provided
One job per production order released to the floor at either plant. Status, quantities and due dates come from Epicor; labor and material post against the job until it is closed.
Grain
One row per job (production order), not per operation
▾ Field meanings (14/22 described)
Plant varchar
TUL-02 is the Tulsa plant, ICT-01 is Wichita. TUL-WH is a warehouse, never a plant.
ReqDueDate date
The date promised to the customer. OTD is measured against this, not the ship date.
JobClosed bit
True once the job is costed and closed. Open jobs still collect labor and material.

PartNum→Part JobNum→scrap_log CustNum→Customer

Review AI suggestions

Nothing has been saved yet. These are drafts, based on your schemas and a small sample of rows (up to 5 per table). Anything you accept is stored as AI-inferred, so it can never overwrite text you or your agents wrote.

Include sample rows: reads up to 5 rows from each undescribed table and sends them to the AI. Your data-policy rules still apply.
managed-db:2:scrap_logEvery part scrapped on the floor, with the reason code, the job it came from and what it cost.grain: One row per scrap eventalso applies: 2 joins · 6 field descriptions · value meanings for 1 field
integration:fiix:work_orderPlanned and breakdown maintenance on presses, lasers and welders. Closed orders carry labor hours and parts used.grain: One row per work orderalso applies: 1 join · 8 field descriptions
First pass yieldShare of parts that pass inspection the first time, without rework or scrap.computed as: (parts inspected - parts reworked - parts scrapped) / parts inspectedalso applies: category: metric · 2 related objects
managed-db:2:supplier_certsMaterial and process certificates on file for each supplier, with the date each one expires.also applies: 4 field descriptions
CancelApply 3 suggestions

[ Data objects ]

Give every table a meaning.

Intuidy scans the databases and systems you connect and lists everything it finds. Write down what a table holds and what one row stands for, and every agent works from the same understanding.

  • ·Found automatically when you connect a system
  • ·Drafted by AI, confirmed by a person
  • ·Your definitions are never overwritten by an agent
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Data Objects
What your tables and records mean. Agents get nothing for an object you have not described.

Your apps, databases, and connected integrations are scanned automatically; agents also fill in meanings as they work. Anything you write here is authoritative. They can refine an empty definition, but never overwrite yours.

Search objects, fields, and definitions…
Status
AllNeeds description4Needs review2Verified27
Data objects (42 of 46 described)
Deep scan Enrich with AI Add object
Scrap log Silent · used by 2 workflows, 1 dashboarddiscovered
managed-db:2:scrap_log

Plant data · ~212,940 rows · used by 2 workflows · 1 dashboard

Details Verify
What does this object represent, in business terms?
Grain
One row per… (what exactly is a single row here?)
Production jobs In prompt Verified Internal
integration:epicor:JobHead

Epicor · ~48,210 rows · 3 joins · used by 2 agents · 6 workflows · 3 dashboards

You provided
One job per production order released to the floor at either plant. Status, quantities and due dates come from Epicor; labor and material post against the job until it is closed.
Grain
One row per job (production order), not per operation
▾ Field meanings (14/22 described)
Plant varchar
TUL-02 is the Tulsa plant, ICT-01 is Wichita. TUL-WH is a warehouse, never a plant.
ReqDueDate date
The date promised to the customer. OTD is measured against this, not the ship date.
JobClosed bit
True once the job is costed and closed. Open jobs still collect labor and material.

PartNum→Part JobNum→scrap_log CustNum→Customer

[ Business glossary ]

Define every number once.

Write down what OTD or scrap rate means at your company and how each one is calculated. Every agent, dashboard and workflow then computes it the same way.

  • ·Metrics, terms, acronyms and processes
  • ·A formula and an owner on every metric
  • ·Rename a term and everything it carries moves with it
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Business Glossary
Your terms and metrics: what your business calls things, and how each number is calculated.
Search terms, definitions, and formulas…
Kind
AllMetric14Entity9Process5Acronym3
Written by
AnyoneAI-inferred6Learned in chat4You provided21
Business glossary (31)
Import terms Add term
OTDMetricApproved
You provided
On-time delivery: the share of order lines shipped complete on or before the date we promised the customer. Measured against the promised date, never the ship date.

Updated Oct 30 by Priya Shah

How it's calculated (metrics) You provided
COUNT(lines shipped complete where ShipDate <= ReqDueDate) / COUNT(lines shipped), by week, per plant
Kind
Metric
Status
Approved
Owner
Priya Shah
Scrap rateMetricApproved
AI-inferred
Scrap cost as a share of material issued to jobs. Plant managers read scrap cost first, then the rate.

Updated Nov 12 by Dana Ruiz

How it's calculated (metrics) AI-inferred
SUM(scrap_log.cost) / SUM(JobMtl.IssuedCost), per plant, per week
Tulsa plantEntityApproved
Learned in chat
Site TUL-02 on Broken Arrow Rd. The Tulsa warehouse (TUL-WH) is not part of it and is never counted in plant output.

Updated Nov 18 by Dana Ruiz

[ Business profile ]

Every agent starts with who you are.

A short profile of your company that every agent reads before it answers. Intuidy drafts it from your website, and anything you write yourself stays exactly as you wrote it.

  • ·Drafted from your website in one click
  • ·Shows who wrote each fact: you, an agent or a sync
  • ·Read by every agent, with every message
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Business Profile
The facts about your organization that every agent reads before answering.
Re-draft from
quarrowfab.com
Re-draft
Company name
updated 4 months ago From signup
Quarrow Fabrication
Industry
updated 4 months ago AI-inferred
Contract metal fabrication: laser cutting, forming, welding and powder coat
What the company does
updated 2 months ago by Dana Ruiz You provided
We build welded steel assemblies and enclosures to print for equipment makers. Two plants: Tulsa (TUL-02) and Wichita (ICT-01). Most revenue is repeat production on blanket POs.
Customers & market
updated 3 weeks ago Learned in chat
Agricultural and HVAC equipment makers across the central US. A few large accounts drive most of the volume.
Key metrics & KPIs
updated 3 weeks ago by Priya Shah You provided
OTD against the promised date Scrap cost and scrap rate, by plant Gross margin by job
Current goals
What the business is focused on right now.

[ Notes and learned facts ]

Corrections that stick.

When someone corrects an agent, the fact it learned waits for your approval. Once approved it reaches every agent, so the same mistake doesn't come back next week.

  • ·Picked up from chats, agents and onboarding
  • ·Nothing is used until a person approves it
  • ·Auto-approve the categories you trust
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Notes & Learned Facts
How you work: notes you write, plus facts agents picked up and are waiting for you to approve.
Learned facts
148
+23 in 30 days
Sent to agents
131
Approved and in use
Corrections
4
This week, was 19
Waiting for you
2
Need your review
AI-Learned Knowledge
allpending 2approved 131rejected 15
Skips the queue: Onboarding Interview answers. Everything else waits for review.
Search learned facts…
All categories All sources Any confidence

“The Tulsa plant” means site TUL-02 on Broken Arrow Rd. TUL-WH is the Tulsa warehouse and is never counted in plant output.

Terminologyvia AI ChatHigh confidenceOrgadded by Dana RuizNov 18

Approved by you just now · sent to agents

Weekly scrap report: lead with scrap cost in dollars, then scrap rate. Plant managers read the first line only.

Preferencesvia AgentMedium confidencePersonalNov 17

Approved by you just now · sent to agents

Freight invoices under $250 auto-approve when the PO is open. Anything above routes to Marcus Lee in AP.

Rulesvia AI ChatHigh confidenceOrgadded by Priya ShahNov 11

Approved by Priya Shah on Nov 12

OTD is on-time delivery against the customer’s promised date, not the ship date.

Terminologyvia Onboarding InterviewHigh confidenceOrgOct 30

Approved on Oct 30

Month-end close: accruals are due by business day 3. Remind owners on day 2 at 9:00.

Workflowsvia AgentHigh confidenceOrgOct 21

Approved by Priya Shah on Oct 22

Line 4 downtime is always planned maintenance.

Rulesvia AgentLow confidenceOrgOct 14

Rejected by Dana Ruiz on Oct 14

[ Data graph ]

See what depends on every table.

Before anyone changes a table, they can see which workflows read it, which dashboards show it and which agents rely on it.

Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Search the graph…
reads writes / feeds
FlowObjectsWorkflowsDashboardsAIPI KeysNotebooksAgents
Production jobsCustomer shipmentsScrap logSupplier billsPurchase ordersMaintenance work ordersNightly ERP syncLate order alertsWeekly scrap reportInvoice to bill matchingPlant scorecardAP agingOps copilotAP assistantCustomer portal keyScrap root cause

[ No black box ]

See exactly what your agents are told.

Intuidy shows the literal text every agent receives. Type a question to see which notes it would pull in, and why.

Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
What your agents actually receive

This is the literal text sent with every message: 5,184 characters of standing context, before anything an agent looks up on demand.

Data objects & glossary3,412 / 4,000
## Data Objects & Glossary
Shared meanings for this organization's data. Ground your answers in these definitions; use describe_object / search_glossary for details.
Objects:
- Production jobs (integration:epicor:JobHead): One job per production order released to the floor at either plant. [grain: One row per job (production order), not per operation] [joins: PartNum→integration:epicor:Part.PartNum, JobNum→managed-db:2:scrap_log.job_num, CustNum→integration:epicor:Customer.CustNum] [Internal]
- Customer shipments (integration:epicor:ShipHead): Packing slips shipped to customers, one per truck stop. [grain: One row per packing slip]
- Supplier bills (integration:quickbooks:bill): Bills posted once a freight or materials invoice is matched to its PO. [Financial]
Glossary:
- OTD: On-time delivery: the share of order lines shipped complete on or before the date we promised the customer. [computed as: COUNT(lines shipped complete where ShipDate <= ReqDueDate) / COUNT(lines shipped), by week, per plant]
- Scrap rate: Scrap cost as a share of material issued to jobs. [computed as: SUM(scrap_log.cost) / SUM(JobMtl.IssuedCost), per plant, per week]
- Tulsa plant: Site TUL-02 on Broken Arrow Rd. The Tulsa warehouse (TUL-WH) is not part of it.
- Blanket PO: A customer purchase order covering a year of releases at a fixed price. Each release is its own job.
- Hot job: A job the customer has escalated. Hot jobs jump the schedule at both plants.
Business profile1,121 characters
Pinned notes651 / 1,200
Try a question

The block above is always sent. This shows what an agent would additionally pull in for a specific question, and why.

Context PreviewSee what the AI receives
Why did OTD drop at the Tulsa plant last week?
Preview
4 / 24 notes selected~1,140 tokens
What the score means:1000 pinned, always sent50+ keyword match1–49 weaker content match0 not sent for this question
1000
Plant rulesrulesAlwaysOrg

pinned: always sent

86
Shipping cutoffsprocessOrg

keyword match: otd, tulsa, late

64
Learned factsknowledge-baseOrg

keyword match: tulsa plant, otd

131 approved · 2 pending · 15 rejected

18
Carrier contactsreferenceOrg

weaker content match: shipped

0
Month-end closeprocessOrg

[ Governance ]

Rules the system enforces, every time.

Data rules turn obligations like HIPAA into checks that run on every agent, workflow and export. They don't rely on an AI choosing to follow instructions.

  • ·Built-in tags for HIPAA / PHI, payment cards, pay data and credentials
  • ·Redact the values, block the operation, or ask a person first
  • ·Test a rule on sample data before you switch it on
  • ·Every decision logged, with the rule that fired
Ontology
Your Ontology
Your data
Data Objects4
Business Glossary
Data Graph
Your business
Business Profile
Notes & Learned Facts2
Access
AIPI Keys
Data Rules
Data Rules
What may never reach an AI, leave your organization, or land in a file. Enforced on every agent, workflow and export.
5 data rules enforced

A data rule is enforced by the system on every agent, workflow and export. It does not depend on the AI choosing to obey it.

New rule
Rules6 Classifications Detectors Test Activity
OnRuleActionApplies whenMatches
No PHI / HIPAA data to AIRecommended

Blocks any data tagged as health information, or carrying a Medicare or provider identifier, from reaching a model.

Block the operation
Sent to AI
2 tag(s) · 2 detector(s)
Health data stays in clinical systems

Removes anything tagged HIPAA / PHI before a write to an outside system, including the systems Intuidy syncs between.

Redact the values
Leaves your organization
2 tag(s)
Redact Social Security numbersRecommended

Masks US Social Security numbers everywhere, keeping the last four digits so records stay recognizable.

Redact the values
Sent to AILeaves your organizationDownloaded as a file
1 detector(s)
Payment card numbers never leaveRecommended

Blocks Luhn-valid payment card numbers from reaching a model, an external system, or a file.

Block the operation
Sent to AILeaves your organizationDownloaded as a file
1 tag(s) · 1 detector(s)
Approve exports of pay data

A person signs off before individual salary or bonus figures land in a file.

Require approval
Downloaded as a file
1 tag(s)
Personal ID badge numbers stay internalRecommended

Masks employee badge numbers before they reach a model, leave the organization, or land in an export.

Redact the values
Sent to AILeaves your organization
1 tag(s) · 1 detector(s)

[ Between systems ]

Your rules travel with your data.

Wire Intuidy in as the connector between two outside platforms and your ontology governs that traffic too. When a referral syncs from Epic to Salesforce, the diagnosis is removed and the SSN is masked before Salesforce ever sees them.

From

Epic

Referral created

Referring clinic
Tamsel Family Medicine
Service line
Cardiology
Diagnosis
I25.10 Coronary artery disease
Patient SSN
512-44-4821

Intuidy data rules

  • Referring clinicPassed
  • Service linePassed
  • DiagnosisRemoved: HIPAA / PHI
  • Patient SSNMasked: last four kept

Logged with the two rules that fired

To

Salesforce

Referral lead

Referring clinic
Tamsel Family Medicine
Service line
Cardiology
Diagnosis
Removed
Patient SSN
•••-••-4821
Explore security Open it to your own tools with AIPI and MCP

[ Get started ]

Teach your agents how you work.

Talk to us

Platform

  • Platform Overview
  • Dashboards
  • Intuidy AIPI
  • Ontology
  • Notebooks
  • How It Works
  • AI Agents
  • Agent Teams
  • Workflow Automation
  • Integrations
  • Security
  • Templates

Use Cases

  • AI App Builder
  • Custom Software
  • No-Code Backend
  • AI Agent Builder
  • AI Workflow Automation
  • Build Custom CRM
  • AI Coding Platform

Solutions

  • CRM & Sales
  • Operations
  • Supply Chain
  • Finance
  • IT Operations
  • All Solutions →

Industries

  • Builders & Infrastructure
  • Energy & Resources
  • Healthcare
  • Financial Services
  • Defense & Intelligence
  • All Industries →

Works With

  • Intuidy + Salesforce
  • Intuidy + SAP
  • Intuidy + HubSpot
  • Intuidy + QuickBooks
  • Intuidy + Oracle
  • Intuidy + Microsoft 365
  • All Integrations →

Company

  • About
  • Blog
  • Events
  • Videos
  • Newsroom
  • FAQ
  • Careers
  • Contact
  • Documentation

Subscribe to our newsletter

Stay up to date on Intuidy's latest products and insights.

Intuidy© 2026 Intuidy Technologies
Privacy|Terms|Cookie Preferences|Sitemap