Introducing Trellis, our client workspace: see the work, approve what sends.

Glossary

The vocabulary, defined by operators.

Every term in the GTM engineering system, with the context the textbook definition skips. Written from pipelines we actually run.

229 terms · 4 topics

GTM Engineering

AI in GTM

AI SDRAn AI SDR is software marketed to autonomously handle the prospecting and outreach work a junior sales development rep used to do, though most tools carrying the label still run on rules a person wrote, with AI filling in personalization inside a fixed template.
AI agentAn AI agent is a system that decides its own next step based on what it just observed, without a person specifying that exact branch in advance, as distinct from a tool that does one lookup or a workflow that follows a fixed sequence someone already wrote.
AI enrichmentAI enrichment is an AI agent reading public pages row by row inside a workflow tool to answer a specific research question and fill a field, distinct from a static database lookup that only returns whatever a provider already has on file.
AI guardrailsAI guardrails are the rules built around an AI system that limit what it can act on unsupervised: a strict boundary between data and instructions, mandatory confidence labels on every claim, and a human review gate before anything reaches a prospect.
AI hallucinationAI hallucination, in a GTM research context, is a research agent stating an invented fact, like a funding amount or investor name, with the same flat confidence as a verified one, because the model is pattern-completing an answer rather than retrieving it from a real source.
AI-native GTMAI-native GTM describes a system where AI is embedded by default, not bolted on, distinct from AI-augmented GTM, where AI assists a workflow a team still runs manually, and from GTM AI, often just a marketing label rather than an architecture claim.
Agentic AIAgentic AI describes a system that decides its own next step, based on what it just observed, without a person approving that specific branch in advance, as distinct from a workflow that only follows branches a person already wrote.
Black-box scoringBlack-box scoring is any lead or intent score that nobody in the room can trace back to a specific, verifiable signal, whether it comes from an opaque AI model or an old spreadsheet formula nobody remembers, making the number impossible to defend, debug, or override.
Blast radiusBlast radius is how much damage a wrong AI decision can do before a person catches it, the measure that should decide which decisions get handed to an agent, not how sophisticated the agent looks in a demo.
Buyer intelligenceBuyer intelligence is a single, decision-ready view of an account: who's involved, what they've said publicly, and what changed recently, assembled in one place, not a synonym for having more raw data points scattered across separate tools.
Command-line interfaceaka CLIA command-line interface (CLI) is a way to operate software by typing text commands in a terminal instead of clicking through screens, which makes a tool easy for a coding agent to drive directly.
Confidence labelA confidence label is a required tag, such as verified, inferred, or unverifiable, attached to every claim an AI research agent returns, so a person or a downstream rule can tell a retrieved fact from a plausible-sounding guess before it reaches a prospect.
Context engineeringContext engineering means deliberately designing what information an AI system can see at the moment it acts, which records, which history, which fields, so its output is grounded in something real instead of a prompt with nothing behind it.
Context rotContext rot is the drop in an AI model's accuracy and recall as the amount of text stuffed into its context window grows, meaning a larger context window doesn't protect a task like personalization from getting less accurate as more source material gets crammed into one prompt.
Escalation pathAn escalation path is a written, specific trigger, such as a confidence score below a threshold or several consecutive skips, that hands a decision from an AI system to a named person, instead of relying on someone noticing a problem eventually.
HeadlessHeadless means a tool can run its logic and complete actions without anyone using its visual interface, driven instead through an API, CLI, MCP server, or webhook, so a script or an AI agent can operate it directly.
Human in the loopaka HITLHuman in the loop means no AI agent's output, a scored lead, a drafted email, a field written back to the CRM, reaches a prospect or a system of record without a person checking it first, so one gate catches both a hostile prompt injection and an ordinary model mistake.
Inspectable logicInspectable logic means a person who wasn't in the room can trace exactly why an AI system or scoring rule did what it did, for any single record, without guessing or asking the model to explain itself after the fact.
MCP serveraka Model Context ProtocolAn MCP server exposes a tool's functions and data to an AI model through the Model Context Protocol, so an agent like Claude can call that tool directly instead of a person clicking through its interface.
Prompt engineeringPrompt engineering is the practice of writing the instructions that shape what an AI system does with the material it reads, including which sources count as evidence, when it can say it doesn't know, and what format its answer has to take.
Prompt injectionaka indirect prompt injectionPrompt injection is an attack where text hidden or embedded in content an AI system reads, a webpage, a document, a resume, is written to be interpreted as an instruction instead of data, causing the system to act on the attacker's command rather than the user's.
Research agentA research agent is an AI system that reads public sources, web pages, filings, press coverage, to answer a specific research question about a company or person, as distinct from a static enrichment lookup that returns whatever a database already has on file.
Retrieval-augmented generationaka RAGRetrieval-augmented generation (RAG) is an AI architecture where a model looks up a specific source before answering and grounds its response in what it actually found, instead of generating an answer from patterns alone.
Source groundingSource grounding is requiring an AI system to tie every claim to a specific passage in a named, admissible source, retrieval instead of pattern completion, so a reader can tell a verified fact from a plausible-sounding guess.

Enterprise Sales

GTM Toolkit

Newsletter

One email a month on GTM engineering.

Ready to engineer your GTM motion?

Tell us how your motion runs today. We'll show you what we'd engineer.