Glossary

An AI system that takes actions toward a goal rather than only answering a question. An agent decides which steps to take, calls tools or APIs, observes the result, and continues until the task is done or a limit is reached.

See also Agentic workflow, Orchestration, MCP

A process where one or more AI agents carry out multi-step work with some autonomy, chaining decisions and tool calls together. It differs from a fixed automation because the agent chooses the path instead of following a hard-coded script.

See also Agent, Orchestration

A go-to-market approach built around AI systems from the ground up, rather than adding AI features to an existing motion. It treats data, context, and agents as core infrastructure and favors coherent systems over collections of point tools.

See also Context engineering, RevOps, GTM

Shaping content so AI answer engines and assistants cite it accurately when they respond to a question. It extends search optimization from ranking on a results page to being the source an AI summary draws from.

See also LLM, RAG, Inbound

Assigning credit for a conversion to the marketing and sales touches that led to it. Every attribution model is an estimate built on assumptions about which touches mattered, so its output is a directional signal, not a fact.

See also Pipeline, RevOps

The rate at which customers stop paying for or using a product over a period. It is the inverse of retention and a core measure of whether a business keeps the value it wins.

See also Retention, Customer health score

The practice of assembling the right information for an AI model to work with: what to include, what to leave out, how to structure it, and when to retrieve it. In an AI-native system, context quality often matters more than prompt wording.

See also Context window, RAG, Prompt engineering

The maximum amount of text, measured in tokens, that a language model can consider at once. Everything the model reasons over, the instructions, the retrieved data, and the conversation, must fit inside it.

See also Context engineering, Token, LLM

Software that records, transcribes, and analyzes sales calls and meetings to surface coaching moments, deal risk, and customer signals. It turns unstructured conversations into searchable, analyzable data.

See also Customer health score

A composite metric that estimates how likely a customer is to renew, expand, or churn, based on signals like product usage, support activity, and engagement. It is a model, so its value depends on the signals feeding it.

See also Churn, Retention

Adding missing or updated information to a record, such as a company’s size and industry or a contact’s role. Enrichment is how a thin lead becomes a usable one.

See also Waterfall enrichment, Intent data

The structure that defines the objects, attributes, and relationships in a system such as a CRM. In an AI-native stack the data model is the foundation, because agents can only act safely on data that is modeled clearly.

See also System of record, RevOps

The likelihood that a sent email reaches the inbox rather than spam or a block. It depends on sender reputation, authentication, content, and volume, and it caps how much outbound email a domain can safely send.

See also Warmup, Outbound

A numeric representation of text or other data that captures its meaning as a vector, so similar meanings sit close together. Embeddings power semantic search and retrieval.

See also Vector database, RAG

Further training a pretrained model on a specific dataset so it adapts to a narrower task, tone, or domain. It changes the model’s weights, unlike prompting or retrieval, which change only its inputs.

See also LLM, Prompt engineering

Go-to-market: the combined strategy and motion a company uses to reach customers and generate revenue, spanning marketing, sales, and customer success.

See also AI-native GTM, RevOps

When a language model produces confident output that is false or unsupported. It is a core risk of generative systems and the reason AI-native workflows need verification and human review.

See also LLM, RAG

Ideal customer profile: a precise description of the accounts a product serves best, used to focus targeting and qualification. A sharp ICP is what keeps AI-scaled outbound from spraying everyone.

See also Outbound, Lead scoring

A motion where prospects come to you, drawn by content, search, referrals, or reputation. It trades a slower ramp for warmer, higher-intent conversations.

See also Outbound, Answer engine optimization

Signals that suggest an account is researching a problem or product, such as content consumption or search behavior. It is used to prioritize outreach toward buyers who may be in-market.

See also Data enrichment, Lead scoring

Assigning a value to a lead based on fit and behavior to rank who to contact first. Scores are estimates, and a model built on weak signals ranks confidently but wrongly.

See also ICP, Intent data

Large language model: an AI model trained on large amounts of text to predict and generate language. LLMs are the engines behind assistants, agents, and most generative GTM tooling.

See also Context window, Hallucination

Model Context Protocol: an open standard for connecting AI assistants to external tools and data through a consistent interface. It lets an agent call systems like a CRM without a custom integration for each one.

See also Agent, Orchestration

Coordinating multiple steps, tools, or agents so they work together toward an outcome. Orchestration is where control flow lives: what runs, in what order, and what happens when a step fails.

See also Agent, Agentic workflow

A motion where you initiate contact with prospects who have not raised their hand, typically by email, phone, or social. AI makes outbound cheaper to scale, which makes targeting discipline the real constraint.

See also Inbound, Deliverability, ICP

The set of open opportunities moving through defined sales stages toward a close, and the value they represent. Pipeline health is a leading indicator of future revenue.

See also Attribution, RevOps

A motion where the product itself drives acquisition, conversion, and expansion, often through a free tier or trial, rather than relying on sales to open every deal.

See also Sales-led growth, GTM

Crafting the instructions given to a language model to get a reliable result. In AI-native systems it is one lever among several, and context and retrieval often matter more than wording.

See also Context engineering, LLM

Retrieval-augmented generation: giving a language model relevant documents at query time so it answers from that material instead of memory alone. It grounds output in specific sources and reduces hallucination.

See also Embedding, Vector database, Context engineering

The share of customers or revenue kept over a period. It is the counterweight to churn and, for most software businesses, the largest driver of long-term value.

See also Churn, Customer health score

Revenue operations: the function that aligns marketing, sales, and customer success around shared data, process, and systems. In an AI-native stack, RevOps owns the data model that automation depends on.

See also Data model, System of record, GTM

A motion where sales reps drive acquisition and expansion through direct selling, common for higher-priced or complex products. It contrasts with product-led growth.

See also Product-led growth, Outbound

A predefined series of outreach steps across email, calls, and social, spaced over time. Sequences make outbound repeatable, and AI makes each step easier to personalize at scale.

See also Outbound, Deliverability

A tool that acts on data owned elsewhere, such as an outreach platform driving emails from CRM records. Separating systems of action from the system of record keeps truth in one place.

See also System of record, Orchestration

The single authoritative source of truth for a given kind of data, such as a CRM for customer records. Other tools may act on the data, but the system of record owns it.

See also System of action, Data model

The unit a language model reads and generates, roughly a word-piece. Model limits and usage costs are counted in tokens, which is why context has to be budgeted.

See also Context window, LLM

A database that stores embeddings and retrieves them by similarity, returning the items closest in meaning to a query. It is the retrieval layer behind semantic search and RAG.

See also Embedding, RAG

Email warmup: gradually increasing sending volume from a new inbox or domain while generating positive engagement, so mailbox providers build trust in it. Warmup protects deliverability before a real campaign begins.

See also Deliverability, Outbound

An enrichment method that queries data providers in sequence, falling through to the next only when the previous one lacks the field. It raises coverage and controls cost by not paying every provider for every lookup.

See also Data enrichment