Terms Guide
Key terms and concepts used throughout the playbook. Jump to a letter or scroll to find what you need.
A
AI Deal Team
A set of specialized AI agents assigned to a single deal or account. The deal team maintains intelligence, prepares briefs, drafts communications, and keeps the CRM current between human touchpoints.
Agent
An AI system that can perform tasks independently, maintain its own memory, and take action without a human starting every step. In Era 3, agents monitor deals, prepare briefs, draft follow-ups, and flag risks on their own schedule.
Autonomous Workflow
A workflow that AI initiates and runs without a human triggering it. The system detects a signal, takes the appropriate action, and only pulls a human in when their judgment or presence would change the outcome.
C
Claude Projects
A feature in Claude that lets you upload reference materials, style guides, frameworks, and past examples that persist across conversations. Instead of pasting your sales methodology into every new chat, you set it up once and it’s always there.
Commercial Model
A data model built on top of your unified customer data that defines what “normal” looks like for your customers at each lifecycle stage. It identifies patterns that predict retention, expansion, and churn so AI can generate actionable signals.
Connected Workflow
A workflow where AI is plugged into your business systems and pulls context directly from your CRM, email, docs, and other tools. The human stops being the integration layer between systems. This is the defining characteristic of Era 2.
Context Assembly
The manual process of gathering information from multiple systems and pasting it into an AI chat window. This is how Era 1 works. The human collects the data, transfers it to AI, and copies the output back. Era 2 eliminates this step.
Context Layer
The infrastructure that gives AI access to your business data across systems. When your AI tool can read your CRM, email, documents, and meeting transcripts through a single connected layer, that is the context layer.
Copy-Paste Era
Another name for Era 1. People pull context out of their systems, paste it into an AI chat, get output, and copy that output back into whatever tool it belongs in. Fast, but manual and siloed.
D
Data Lake
A centralized repository that brings together data from multiple systems in its raw form. In Era 3, this is where your customer attributes, events, and unstructured data converge so AI can access the full picture from one place.
E
Era 1. AI as a Tool
The first stage of AI adoption. Individual contributors use AI on their own through copy and paste. No shared infrastructure, no connected systems. The gains are real but fragmented across individuals.
Era 2. AI as a Colleague
The second stage. AI is connected to your CRM, email, docs, and other systems. Work starts in the AI layer. Playbooks run as connected workflows. The productivity gains become organizational and consistent instead of individual and variable.
Era 3. AI as a Direct Report
The third stage. AI initiates work autonomously. Every remaining person becomes a manager of AI workers. The org structure changes. Fewer people generate the same or more revenue. The roles that survive are fundamentally different from what they were before.
Evals
Short for evaluations. The systematic process of measuring AI output quality across workflows. In Era 3, your ops team runs evals the same way a manager would run performance reviews on a human team. They identify where output is degrading, figure out why, and fix it.
Event-Driven
A system architecture where changes in one system immediately push data to other systems in real time. In Era 3, when an account’s product usage drops, that event flows instantly into the unified data layer and can trigger an autonomous workflow. The opposite of batch processing where data syncs overnight.
H
Hallucination
When AI generates information that sounds confident but is factually wrong. It invents a detail, misattributes a quote, or states something that didn’t happen. This is why human review of AI output matters at every era.
Human-Interaction Specialist
One of the two roles that remain in an Era 3 revenue org. These are the AEs and CSMs whose presence in front of customers changes the commercial outcome. Their value is the human connection, the trust, the room-reading, the relationship that AI cannot replicate.
Human-in-the-Loop
A safety model where AI can reason, analyze, and recommend autonomously but requires explicit human approval before taking actions that affect customers. Internal actions like updating a CRM field might run automatically. External actions like sending a customer email require a human to approve.
I
Infrastructure Rabbit Hole
When a talented rep or CSM quietly becomes a part-time systems engineer, spending significant time building AI workflows and integrations while their pipeline or book of business suffers. The line between productive experimentation and the rabbit hole is whether the person is still hitting their number.
M
MCP
Model Context Protocol. A standard that allows AI tools to connect directly to your business systems. When Claude connects to your CRM, email, or docs through MCP, the AI can read from and write to those systems without manual copy and paste.
O
Orchestration Layer
The infrastructure that coordinates multiple AI agents working together. In Era 3, you have specialized agents handling different aspects of a deal or account. The orchestration layer manages how they share information, hand off tasks, and collaborate.
Override Rate
The percentage of AI-generated outputs that a human rejects or rewrites instead of approving. A key metric in Era 3. High override rates on a specific workflow signal that the AI playbook needs improvement. Low override rates signal that the system’s judgment is earning trust.
P
Persistence Layer
The infrastructure that lets AI agents maintain memory between conversations. Without persistence, every AI interaction starts from scratch. With it, an agent remembers what happened in the last conversation, what it recommended, and what the human decided.
Plugins
Extensions that give an AI assistant the ability to connect to and take action in external systems. Plugins are how AI reads your CRM, sends email, searches your docs, or pulls data from enrichment tools. In Era 2, plugins turn a standalone AI chat into a connected platform. MCP is the emerging standard that is replacing proprietary plugin architectures.
Projects
Persistent AI workspaces that retain context, reference materials, and instructions across conversations. Instead of re-explaining your sales methodology or uploading your playbook every session, a project keeps it loaded. Claude Projects is the most common implementation. Projects are the bridge between throwaway chat sessions and repeatable AI workflows.
Prompting
The skill of giving AI clear, specific instructions with the right context to get useful output. In Era 1, prompting skill is the primary differentiator between people who get great results from AI and people who get mediocre results.
S
Shadow AI
AI tools that your team is using without official approval or IT awareness. Usually individual subscriptions to ChatGPT or Claude on personal emails. Common in Era 1 and early Era 2. Not malicious. People found something that works and didn’t want to wait for procurement.
Signal
A data-driven indicator that something meaningful is happening with a customer. In Era 3, signals are generated automatically by the commercial model. A drop in product usage, a change in communication patterns, a spike in support tickets. These signals trigger autonomous workflows.
Skills
Reusable, shareable AI capabilities that encode a specific workflow or task. A skill might handle deal prep, follow-up drafting, or account health scanning. In Era 1, skills are individual prompts people develop on their own. In Era 2 and Era 3, skills become organizational assets that run as part of connected and autonomous workflows.
System Builders
One of the two roles that remain in an Era 3 revenue org. These are the ops people who architect, maintain, evaluate, and optimize the AI workflows that run the entire revenue motion. They decide where human involvement has the highest impact. The highest-leverage role in the company.
System Prompt
The hidden instructions that define how an AI assistant behaves before the user says anything. System prompts set the role, tone, constraints, and context for every interaction. In a revenue org, system prompts are how you encode your sales methodology, brand voice, qualification framework, or account management playbook into AI so it operates consistently across the team.
T
The Bridge
The transition between Era 2 and Era 3. Not a technology change. An identity shift where people stop being executors augmented by AI and start being managers of AI that does the execution. The hardest part of the entire transformation.
U
Unified Data Layer
A single accessible layer that brings together customer attributes, events, and unstructured data from all your systems. The foundation for Era 3. Without it, AI agents are working with partial information from disconnected sources.