Apply to Moonlabs
Glossary

The operator glossary.

The AI, commercial, and investment terms we use most at Moonlabs, defined in operator-direct English. Not the textbook version — the version we use in the room.

AI engineering

Agent
A loop that gives a model access to tools and lets it choose, iteratively, which tool to call. In 2026 the "agent" word is overloaded — pin it to the loop, not the marketing.
Claude Code
Anthropic's terminal-native coding agent. The thing we use to ship unsupervised work in parallel while we are doing something else.
Cursor
AI-first IDE we use as the default editor across the team. Multi-file reasoning is the feature that matters.
Eval
A deterministic test for a non-deterministic system: fixed input, desired property of the output, measurement of whether the property held. The single biggest predictor of whether an AI product will survive contact with users.
Embeddings
Numerical representations of meaning, used to compare similarity between texts. We use them sparingly — Postgres + pgvector covers 90% of the cases people reach for a dedicated vector DB for.
Hallucination
When a model generates plausible-sounding output that is factually wrong. The frequency has fallen; it has not gone to zero. Operator taste is the defence.
Inference cost
The £/token cost of running a model against a query. Falls roughly an order of magnitude every 18 months. Plan your unit economics for the version that lands in two years, not today.
LLM
Large language model. Increasingly a misleading label — multimodal models are now the default and "language" is a subset of what they do.
pgvector
Postgres extension for storing and querying vector embeddings. Almost always sufficient until you cross the millions-of-vectors threshold.
Prompt
The text you send a model. The skill is taste, not cleverness. A 300-word brief that produces a usable first draft is a different universe of capability from "help me write a deck."
RAG
Retrieval-augmented generation. The pattern of looking up relevant context, then feeding it to the model alongside the prompt. The default architecture for any knowledge-grounded AI feature.
Tool use
A model's ability to call structured functions (search, lookup, write to a database). The shift from "chatbot" to "agent" runs through this primitive.
Token
The unit of input/output a model is priced and rate-limited by. Roughly 0.75 of an English word, often less for code.

Commercials

ACV
Annual contract value. The yearly revenue from a single customer contract. The metric that turns "users" into a business.
ARPU
Average revenue per user. Useful for products with many small accounts; misleading once you have a few large ones.
ARR
Annual recurring revenue. The headline metric for SaaS companies. Booked, not invoiced — pilots count when they are signed, not when they pay.
CAC
Customer acquisition cost. Total sales and marketing spend divided by net new customers acquired. The denominator is harder than it looks.
Churn
The rate at which customers leave. Logo churn = number of customers; revenue churn = pounds. Always look at both.
Discovery call
The first structured conversation with a potential buyer. The goal is to learn, not to sell. Pricing intuitions are built by running fifty of these.
LTV
Lifetime value of a customer. ACV × average lifetime − cost to serve. Wildly miscalculated by most early-stage founders.
Net dollar retention
How much existing-customer revenue grows year-on-year after accounting for churn and expansion. >100% means your existing customers are paying more this year than last. The gold standard SaaS metric in 2026.
Payback period
How long it takes for a customer's revenue to repay the cost of acquiring them. Shorter is better. Twelve months is healthy at seed; eighteen is concerning; twenty-four is broken.
Pilot
A time-bounded, scoped contract designed to prove the product can deliver before a longer commitment. Always paid. "Free pilots" are usually proofs-of-nothing.
Pricing intuition
The internalised model of what a buyer is willing to pay for a given outcome. Cannot be read off a spreadsheet; built through reps of asking the question.
Positioning
Where in the buyer's mental map you sit, relative to alternatives. Pricing, language, and product all flow from positioning.
Wedge
The narrowest version of the product that is genuinely valuable to a specific buyer. The thing most early-stage companies fail to find before running out of money.

Investment

Angel
An individual writing personal money into early-stage companies. UK angels typically £25k-£250k cheques; the better ones write four-line emails not 14-page diligence packs.
Cap
On a SAFE, the maximum valuation at which the SAFE converts to equity. The lower the cap, the better for the SAFE holder, the worse for the founder.
Cap table
The ledger of who owns what percentage of the company. Worth keeping clean from week one. The fixings on a messy one are expensive.
Convertible note
A loan that converts into equity at a future round, typically at a discount. Mostly replaced by SAFEs in the UK now.
Data room
The structured folder of company documents an investor reads during diligence. The single largest factor in whether a deal momentum survives the diligence phase.
Dilution
The reduction in your equity percentage that happens when new shares are issued. Compounds across rounds. Painful to look at; unavoidable to live with.
EIS / SEIS
UK tax-relief schemes for individual investors backing early-stage companies. The reason most UK angels can write at the cheque sizes they do.
HNWI
High-net-worth individual. People who are not professional investors but write substantial cheques. Network-gated; often the right capital for operating businesses.
Lead investor
The investor who sets the terms of a round and the pace of its close. Other investors follow the lead's diligence. Without a lead, most rounds drag.
Liquidation preference
The clause that decides who gets paid first if the company is sold. 1x non-participating is the friendly default; anything more aggressive is a yellow flag.
Post-money
The valuation of the company after a round closes. Pre-money + round size = post-money. Always specify which one you are quoting.
Pro rata
An existing investor's right to participate in future rounds to maintain their percentage. Important to negotiate clearly; expensive to retrofit.
Round
A defined fundraising event with a target amount and an opening and closing date. "Always-on fundraising" is a euphemism for "not actually fundraising."
SAFE
Simple Agreement for Future Equity. Originally a YC instrument, now standard in UK seed. Lightweight, founder-friendly. Stack a few of them and the cap table gets complicated fast.
Term sheet
The non-binding summary of the terms a lead investor is offering. Signing it is the moment the round shifts from sales to execution.
Vesting
The schedule by which founders and employees earn their equity over time. Standard is 4 years with a 1-year cliff. Skip it at your peril.

Moonlabs specifics

Tech-for-equity
Our 1% equity slice in exchange for the technical foundation and operator time. Calibrated to be small enough that we have to genuinely help to make it worthwhile.
Success fee
Our 10% fee on the capital raised in a round we help you close. Paid on success, not on slides.
Commitment fee
The £999 a founder pays on signing the Incubator agreement. Credited back against the success fee at round close, so on net it costs nothing if you raise. Exists to filter for founders willing to put a small amount of skin in the game.
The wedge
Our internal shorthand for the smallest defensible version of a company. The thing every Academy student writes a document about in week one.
Three pillars
Coding, Commercials, Investment — the three skills the Academy curriculum is built around. Every week of the cohort touches all three.
Demo day
The cohort's public moment at the end of week twelve. Invited audience of investors, operators, and hiring observers.

Missing a term?

We will add it. Email founders@homemove.com with the word you want defined the operator way and we will write it up.

Your next chapter starts here.

Tell us about the company you want to build. If we’re a fit, we’ll get back within a week.