Glossary
The words behind One3
Plain definitions of the ideas the platform is built on.
A
- AI credits
- The metered unit for AI usage in One3, mapped to real model cost with a margin and protected by a hard cap.
- AI disclosure
- One3's rule that people always know when they are dealing with AI.
- AI incident response
- How One3 detects, contains, reverses or compensates, explains and fixes agent mistakes.
- Action
- A typed, permissioned and audited verb that is the single way anything changes in One3.
- Action Bus
- The kernel service that validates, authorises, executes and audits every Action.
- Agency OS
- One3's toolkit for digital agencies: sub-accounts, white-label, snapshots, rebilling with margin, client portals and agent resale.
- Agent Investment Statement
- The monthly statement that shows each client what it invested in agents and what they returned, with evidence.
- Agent Network
- A planned network where verified company agents discover, negotiate and transact with each other under policy.
- Agent Performance Arena
- Where agent performance is measured against the tenant's previous process and role targets, with evidence packs for claims.
- Agent Rental Marketplace
- Where clients browse, trial, rent, return and replace Digital Employees and squads.
- Agent Replacement Engine
- Swaps an agent's version or template while keeping the tenant's context, history and ROI lineage.
- Agent Training Gym
- The pipeline through which Digital Employees improve — only by tested, versioned releases.
- Agent identity
- The separate, non-human principal every agent acts as, with scoped short-lived tokens and least privilege.
- Agent template
- A versioned, reusable definition of a Digital Employee role, built from packs by the Role Factory.
- Approvals Inbox
- The single place where people review and decide pending R3 and R4 Actions from agents, workflows and colleagues.
- Automations
- One3's workflow builder: triggers and steps made of governed Actions, with run history.
- Autonomous loops
- The Enterprise Brain's four loops — operational, tactical, strategic and transformational — each with its own autonomy ceiling.
- Autonomy level
- How much an AI agent may do on its own for one skill, from A0 (observe) to A4 (autonomous).
- Avatar & Voice Rights Registry
- The registry of consent, scope, expiry and revocation for any real-person likeness or voice.
- Avatar classes
- One3's three avatar classes: Synthetic (default), Consented real-person representative, and Prohibited.
B
- Baseline
- The frozen "before" measurement that an agent's results are compared against.
- Budget
- Money, compute or time allowances for actors, goals, squads and workflows, enforced at the Action Bus.
- Budget guardrails
- Caps and stop triggers that pause an agent's spend automatically when limits or return thresholds are breached.
- Business Twin
- A simulation of the business that can test interventions before they are made, with stated maturity levels.
C
- C2PA provenance
- An open standard for signing media with its origin, used by One3 to mark AI-generated content.
- Capability (C-1…C-200)
- A planned feature set with its own objects, Actions and AI skills — the unit of planning, not of sale.
- Capability Brief
- The approved one-page brief every app, Pack, agent family or connector needs before it enters a sprint.
- Certification levels
- Listed, Verified and Certified Outcome — the trust levels an agent template must earn.
- Client mandate
- The scoped authority a client issues before a rented agent may act: purpose, Actions, channels, spend limit and expiry.
- Clouds
- Curated compositions of apps with a shared cockpit, shared agents and cross-app flows, such as Revenue Cloud or CX Cloud.
- Company Constitution
- A machine-readable policy document per tenant that every agent obeys: brand voice, DoA, forbidden actions, escalation and contact rules.
- Complexity Budget
- A cap on how many core infrastructure systems One3 runs in production in its early waves.
- Connector Health Center
- Shows the health, freshness and auth status of every connection — and agents check it before acting.
- Context Engine
- Assembles the minimum sufficient context for each agent step from records, knowledge, graph, metrics, memory and the Constitution.
- Context Graph
- The network of entities and relationships across apps and connectors that agents and the Enterprise Brain reason over.
- Copilot
- One3's assistant for asking questions and taking actions across the apps a tenant subscribes to.
- Cross-tenant isolation
- The rules that keep each client's data, memory and playbooks inside its own tenant, even when renting the same agent template as others.
D
- DPDP
- India's Digital Personal Data Protection Act and Rules, which One3 designs for from the start.
- Data quality score
- A score per record, object and metric; below the threshold, One3 makes no recommendation and shows no ROI figure.
- Decision Engine
- Frames decisions, generates options, estimates expected value and risk, and recommends with confidence.
- Decision Journal
- The record of significant decisions with their expected and actual outcomes, used to calibrate people and agents.
- Definition of Done levels
- Five levels, L1 to L5, that an app must reach — L3 is the minimum for general availability.
- Delegation of authority (DoA)
- The matrix that says who may approve which amounts and decisions — for people and agents alike.
- Digital Employee
- A governed AI agent with a job, KPIs, a budget, an identity, a named human owner and an employment record.
- Digital Employee Store
- One3's first-party catalog of agent templates built by the Role Factory.
E
- Enterprise Brain
- The reasoning layer that runs Sense → Understand → Decide → Act → Learn → Govern over the kernel.
- Eval Coverage Score
- A 0–100 score of how well an agent skill has been tested, required for every autonomy promotion.
F
- Fact Register
- One3's register of facts that can change — laws, licences, provider rules, prices — each with an owner and an expiry.
G
- Goal
- A kernel primitive that links an objective to KPIs, a plan, Actions and a measured outcome.
- Goal Contract
- The versioned internal object that defines what a rented agent works toward, what it may do and spend, and how results are evidenced.
- Growth OS
- One3's Year-1 product for SMB growth: CRM, Marketing, Forms, Desk, Content, SEO, Ads, Dashboards, Billing lite and agents.
H
- Holdout experiment
- A pre-registered test that compares a treated group with an untreated control to estimate an agent's real effect.
- Human Expert Cloud
- Vetted human experts — accountants, lawyers, designers, ad specialists — that agents escalate to, paid per task.
I
- Incremental contribution
- The cohort-level extra contribution an agent produced, estimated from a final experiment — the only figure that may be called financial return.
- Industry Edition
- A preconfigured bundle of apps, modules, industry entities, workflows, agents and compliance templates for one industry.
- Investment & ROI Ledger
- The system that records every cost of a rented agent against the returns it can honestly be credited with.
K
- Kernel
- The small core every One3 app depends on: objects, Actions, events, workflows, policies and tenancy.
- Kernel primitive
- One of the 14 building blocks every One3 feature is built from.
- Kill switch
- A control that stops an agent skill, an agent, a tenant's whole AI workforce or a platform-wide model in seconds.
- Knowledge
- One3's permission-aware document knowledge base that answers with citations and detects gaps.
M
- MCP (Model Context Protocol)
- An open protocol through which external AI assistants can call One3 Actions under One3's governance.
- Migration Factory
- The guided pipeline that moves data from tools like HubSpot, GoHighLevel, Zoho or spreadsheets into One3, with dry runs and rollback.
- Model Hub
- The gateway and router that picks a model for every AI call by measured evals, cost, latency and data policy.
- Module
- A functional area inside a One3 app, sold as Core, Add-on or Standalone, sharing the app's data model.
N
- net_roi_ratio and return_on_spend_ratio
- The two named return ratios One3 uses instead of a bare "ROI" — break-even 0.0 and 1.0.
O
- Observed and assisted outcomes
- The two labels every individual outcome row carries — never "caused".
- Outcome
- A verified business result — a deal paid, a meeting held, a ticket resolved — linked to the Actions and Decisions behind it.
- Outcome Contracting
- The module that defines billable outcomes, attribution, exclusions, disputes and credit notes for performance pricing.
P
- Pack
- A versioned bundle of objects, Actions, events, workflows, policies, views, agent skills and evals that makes up an app module.
- Pass-through costs
- Third-party costs such as WhatsApp conversations, SMS, calls, SERP data and enrichment, billed at cost plus margin or capped.
- Persona Policy Engine
- Checks every agent's name, face, voice and disclosure against law and brand before each external message, render or call.
- Policy
- A declarative rule that decides who or what may do which Action, on which record, under which conditions.
- Previous-process baseline
- The tenant's own way of doing a task before renting an agent — the only benchmark One3 compares agents against.
- Public AI claim rules
- One3's written list of approved and forbidden language for AI claims, with evidence required for every claim.
R
- Rental lifecycle
- The nine stages of renting a Digital Employee, from discovery to offboarding.
- Revenue Squad
- A packaged growth squad aimed at a measurable outcome such as leads, local growth or cash collection.
- Reversibility class
- What can honestly be done after an Action runs: reverse it, compensate for it, or nothing.
- Risk tier
- A rating from R0 (read-only) to R4 (irreversible or financial) that sets the default approval gate for an Action.
- Role Factory
- The system that generates and tests agent templates from versioned packs, growing the catalog to 1000+.
S
- Segregation of duties
- Policies that stop the same principal both creating and approving sensitive changes such as payments.
- Shadow mode
- The first stage of an agent's work, where it proposes actions, humans act, and agreement is scored.
- Squad
- A goal-packaged team of Digital Employees with shared tools, KPIs and budget.
- Supervision cost
- The value of the human time spent approving, reviewing and correcting an agent's work, always counted as investment.
- Synthetic Customer Simulator
- Generates synthetic leads, objections and support cases so agents can be tested safely before they meet real people.
T
- Tainted input
- Untrusted content such as inbound emails or web pages, which can inform an agent but never authorise an Action.
- Tenant
- One customer's isolated space in One3, with its own data, users, agents, policies and subscriptions.
- Tenant Customisation Boundary
- The rules for what tenants can customise freely, what needs a sandbox, and what is forbidden.
W
- Wave
- One of One3's gated build stages, W0 to W6, each with exit criteria.
Y
- Year-1 Product Boundary
- The written list of exactly what One3 builds in Year 1 — and what it deliberately defers.
See how One3 would run your business
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.