StackLift
Glossary

Software delivery & AI engineering, defined.

Clear, practical definitions of the terms that come up when you scope, price, and ship software - written for decision-makers, not just engineers.

Story points

A unit of relative effort used to size software work instead of raw hours.

Read definition

MVP (Minimum Viable Product)

The smallest version of a product that delivers real value and validates the core bet.

Read definition

AI software factory

A governed delivery system where AI agents accelerate every phase and engineers own every decision.

Read definition

AI agents (in software delivery)

Specialized AI workers (scope, code, QA, DevOps...) that draft work for human review.

Read definition

Human-in-the-loop

A workflow where AI proposes and a human reviews and approves before anything ships.

Read definition

RAG (Retrieval-Augmented Generation)

Grounding an LLM's answers in your own data retrieved at query time.

Read definition

Vector database

A database that stores embeddings to enable semantic / similarity search.

Read definition

AI copilot

An assistant embedded in a product or workflow that suggests actions with human oversight.

Read definition

iPaaS (Integration Platform as a Service)

A platform for connecting and automating data flow between business systems.

Read definition

Workflow automation

Replacing manual, multi-step business processes with software that runs them reliably.

Read definition

System integration

Making separate software systems work together by sharing data and triggering actions.

Read definition

API (Application Programming Interface)

A defined contract that lets software systems talk to each other.

Read definition

CI/CD

Automated building, testing, and deploying of code on every change.

Read definition

Observability

Being able to understand a system's internal state from its logs, metrics, and traces.

Read definition

Release management

The controlled process of shipping software safely with the ability to recover quickly.

Read definition

Delivery operations

The operational floor - CI/CD, QA, observability, security - every production build needs.

Read definition

Technical debt

The future cost of shortcuts taken in code or architecture today.

Read definition

Code review

A peer (or senior) checking code for correctness, security, and maintainability before merge.

Read definition

Change request

A formal, priced change to agreed scope after a project starts.

Read definition

Software estimation

A structured prediction of a project's scope, cost, and timeline - ideally a delivery plan in draft.

Read definition

Agentic AI

An approach where AI systems pursue a goal on their own: planning, using tools and adjusting as they go.

Read definition

MCP (Model Context Protocol)

An open protocol that lets AI apps connect to tools and data sources through one standard interface.

Read definition

Agent orchestration

The coordination layer that decides which agent or tool runs next, passes context and handles failures.

Read definition

Multi-agent system

A setup where several specialized AI agents work together on one task, each with its own role and tools.

Read definition

Tool calling

How a language model asks software to run an action, such as a search or database query, and uses the result.

Read definition

Agent memory

How an AI agent keeps information across steps and sessions: short-term context plus stored long-term facts.

Read definition

Agent evaluation

Testing whether an AI agent completes tasks correctly, safely and efficiently, using repeatable scenarios.

Read definition

AI guardrails

Controls around an AI system that limit what it can see, say and do, checked in code, not just in the prompt.

Read definition

Prompt injection

An attack where text fed to an AI model, by a user or hidden in content, overrides its instructions.

Read definition

AI red teaming

Deliberately attacking your own AI system to find failures, unsafe behavior and security holes before others do.

Read definition

Large language model (LLM)

An AI model trained on huge amounts of text that reads and writes language, code and structured data.

Read definition

Tokens

The small chunks of text a language model reads and writes, and the unit most AI APIs bill by.

Read definition

Context window

The maximum amount of text, in tokens, a model can take in and produce in a single request.

Read definition

Embeddings

Lists of numbers that capture the meaning of text, images or other data so similar items sit close together.

Read definition

Fine-tuning

Further training a model on your own examples so it behaves in a consistent way for a specific task.

Read definition

LoRA (Low-Rank Adaptation)

A fine-tuning method that trains a small add-on instead of changing every weight in the model.

Read definition

Prompt engineering

Writing and testing the instructions, examples and context you give a model to get reliable output.

Read definition

Prompt caching

Reusing the already-processed start of a prompt across requests to cut latency and cost.

Read definition

Hallucination

When an AI model states something false or invented with the same confidence as something true.

Read definition

Temperature

A setting that controls how predictable or varied a language model's word choices are.

Read definition

Semantic search

Search that matches the meaning of a query, not only its exact words.

Read definition

Hybrid search

Search that runs keyword and vector retrieval together and merges the results.

Read definition

Reranking

A second pass that reorders retrieved results by how well they answer the question.

Read definition

Chunking

Splitting documents into passages that are indexed and retrieved one at a time.

Read definition

Knowledge graph

A model of things and the relationships between them, stored as a connected graph.

Read definition

GraphRAG

RAG that retrieves through a knowledge graph to answer multi-step and whole-corpus questions.

Read definition

Data pipeline

Automated steps that move data from its sources to where it is used, and prepare it on the way.

Read definition

ETL (extract, transform, load)

A pattern that extracts data, cleans and reshapes it, then loads it into a target system.

Read definition

Data warehouse

A central database built for reporting and analytics across the whole business.

Read definition

Data lakehouse

One platform that keeps data in open files with warehouse-style tables and SQL on top.

Read definition

LLMOps

The practice of running LLM applications in production: prompts, evals, deployment, monitoring and cost control.

Read definition

AI observability

Tracing, logging and scoring LLM and agent runs so you can see what happened and whether the answer was good.

Read definition

Model drift

When an AI system's results get worse over time because data, user behavior or the model itself changes.

Read definition

AI governance

The policies, roles, reviews and records an organization uses to control how AI is built and used.

Read definition

Responsible AI

Building and using AI that is fair, safe, transparent, private and accountable, with people kept in charge.

Read definition

AI cost optimization

Lowering what an AI feature costs to run by choosing the right model, trimming tokens, caching and routing.

Read definition

Model routing

Sending each AI request to the model best suited to it, by difficulty, cost, speed or data rules.

Read definition

Inference

Running a trained AI model to produce an output: the step that happens every time someone uses the AI.

Read definition

Small language model

A compact language model that runs fast and cheap and can handle focused tasks, sometimes on your own hardware.

Read definition

Open-weight model

An AI model whose trained weights you can download and run yourself, under the publisher's license.

Read definition

Proof of Concept (PoC)

A small, time-boxed build that tests whether one risky technical idea can work at all.

Read definition

Prototype

An early, interactive model of a product used to test design and flow before the real build.

Read definition

Product Requirements Document (PRD)

A written description of what a product must do, for whom and why, shared by the whole team.

Read definition

User Story

A short note of a feature from the user's view: as a user, I want a goal, so that a benefit.

Read definition

Acceptance Criteria

Specific, testable conditions a feature or story must meet to be accepted as complete.

Read definition

Definition of Done (DoD)

A shared checklist of quality standards every piece of work must meet before it counts as finished.

Read definition

Sprint

A fixed, short work cycle, one month or less, that ends with a usable, reviewed increment.

Read definition

Agile

A way of building software in small increments, with frequent feedback and room for change.

Read definition

Scope Creep

Uncontrolled growth of a project's requirements without matching changes to budget or schedule.

Read definition

Discovery Phase

The first stage of a project, where the team learns the problem and plans before building.

Read definition

Microservices

An architecture that splits one application into small services, each deployed and scaled on its own.

Read definition

Monolith (monolithic architecture)

An application built and deployed as one unit, with all its features in a single codebase.

Read definition

API gateway

A single entry point that routes, secures and limits requests to the services behind it.

Read definition

Serverless

A cloud model where the provider runs your code on demand and you pay for use, not for idle servers.

Read definition

Kubernetes

An open-source system that runs, scales and heals containerized applications across a group of machines.

Read definition

Infrastructure as code (IaC)

Managing servers, networks and cloud resources through version-controlled code instead of manual setup.

Read definition

Feature flags

Switches in code that turn a feature on or off for some or all users without deploying again.

Read definition

Blue-green deployment

A release method that runs two identical environments and switches traffic from the old to the new.

Read definition

Canary release

A release method that sends a new version to a small share of users first and widens it as metrics stay healthy.

Read definition

Service level objective (SLO)

A reliability target for a service, such as availability or response time, measured over a set period.

Read definition

Forward deployed engineer (FDE)

A software engineer who works inside a customer's environment to build and deploy a solution for that customer's real problem.

Read definition

Start in 60 seconds

Have a project in mind?

Skip the jargon - get an AI-assisted, architect-reviewed scope, cost range, and timeline for your build.