Level 5: Building AI Products — Cheat Sheet

Key terms

AI Product Thinking

Starting from a frustrating customer problem and asking whether AI improves the outcome. It's Team B's habit, systematized. Users don't buy AI; they buy time saved, effort reduced, and better decisions.

Problem-First Reasoning

Choosing the problem, then discovering whether AI is the right tool. Team B started from "43% abandon during data import" and shipped a column-mapping classifier, not a chatbot, while the feature-first team shipped a chatbot 4% of users tried once.

The AI Product Equation

User Value + Business Value + Technical Feasibility + Responsible AI = successful product; weakness in any dimension drags down the whole.

Substitution Test

Replace "AI" with "a very fast intern" in your pitch. If the sentence still creates value, the idea rests on a real problem, not on the technology's glamour.

Workflow Mapping

Charting the user's full flow (receive request, gather information, analyze, create, review, deliver) and adding AI only at the highest-value steps, instead of asking "what can AI do?"

Opportunity Canvas

The structured checklist for any AI idea: who has the problem, what frustrates them, how they solve it today, how AI improves it, why invest, what could go wrong, how success is measured. (see AI-042)

MVP

The simplest version delivering enough value for real customers to use and give feedback, the stage every earlier step in the lifecycle exists to reach with less uncertainty.

Trust

The compounding adoption driver built from transparency, reliability, explainability, privacy, and human oversight. It's one of the strongest competitive advantages an AI product can hold. (see AI-038)

Product Discovery

The structured process of reducing uncertainty before building: user research, problem validation, prototyping, business analysis. The worked example's one-week discovery killed "AI writes the email" and saved "AI briefs the rep": same team, a tenth of the risk, aimed at the real bottleneck.

Jobs to Be Done

Focusing on the task users are actually hiring the product for: "prepare tomorrow's meeting in 15 minutes," not "I need AI." The job reveals what the product truly must do.

User Observation

Watching users work rather than asking. Shadowing 5 reps revealed 16 of 20 minutes went to research, not writing. Users normalize their problems; observation surfaces what interviews miss.

Problem Statement

User + need + reason: "Project managers need a faster way to summarize weekly updates because reports take hours." Clear problems lead to better solutions.

Feasibility Spike

A cheap, fast test of whether the AI can do the job: hand-crafting one great prompt against real inputs and judging honestly. The wizard-of-oz test is discovery's cheapest experiment. (see AI-010)

Data Audit

Checking whether the inputs the model needs actually exist. The CRM notes were sparse and stale, which killed the personalization idea on day 2. (see AI-006)

Four-Dimension Validation

Desirability (users want it), feasibility (we can build it), viability (business value), responsibility (safe to deploy). A successful AI product satisfies all four.

Blast Radius Check

Weighing where output lands: customer-facing outreach makes hallucination a brand risk, while internal briefs make it a nuisance. Risk assessment reframed the product. (see AI-060)

AI UX

Designing interactions so users achieve goals intuitively and confidently. The worked example proves the stakes: identical model, but auto-insert killed adoption in a month while suggest-and-diff grew it. AI UX is the discipline of that difference.

Designing for Uncertainty

Showing honest signals instead of false confidence: "I found two possible interpretations," confidence shading, "human review recommended." Honesty builds trust more reliably than pretended certainty.

Confidence Indicator

A high/medium/low signal beside responses helping users decide when to verify before acting; it's the shading in the suggest-and-diff design.

Progressive Disclosure

Starting simple and revealing advanced features as users gain confidence, reducing cognitive load so new users succeed quickly.

Error Recovery

Replacing "Error" with guidance, such as "I couldn't find that document. Upload it again?", so mistakes move users forward instead of dead-ending them.

Verification by Design

Making review the path of least resistance: per-paragraph insert with cited sources meant agents skimmed citations naturally, versus auto-insert where habituated agents stopped reading. (see AI-053)

Feedback Loop

Thumbs up/down, issue reports, and regeneration requests: among the most valuable signals for improving both the AI and the experience. (see AI-022)

Failure-State Design

Sketching what the user sees when the model is wrong before designing the happy path. If the wrong-state design is hard, the feature may need a lower autonomy level.

Prompt UX

Designing the prompting experience so users communicate naturally without learning prompt engineering: the meal-planner's rewrite took first-menu completion from 38% to 74% on the same model. Every system-prompt line is an interface decision.

Conversation Design

Structuring the multi-turn journey (greeting, understand goal, gather information, respond, clarify, confirm, offer next step), like an experienced service rep who guides rather than expects a perfect brief. UX writing with state.

Clarification

Asking one focused question when a request is incomplete ("what type of report?") instead of guessing generically, and asking one question per turn, not a form-shaped interrogation.

Empty State

The screen before the user types, filled with example prompts and buttons for common answers, because free text is a blank-page problem for users too. The first interaction determines long-term adoption. (see AI-043)

Repair

When parsing fails, naming what was understood and asking a narrower question, never the dead-end "I didn't understand." Recovery should always move users forward.

Follow-Up Suggestion

Contextual next actions after each response (improve this, create slides, generate an email), keeping conversations moving and revealing the product's range.

Conversation Memory

Retaining stated goals, uploaded files, and choices across turns so users never repeat themselves, using an explicit cue like "I'll remember the allergy; change it anytime." (see AI-035)

Prompt Template

A reusable Role → Task → Context → Constraints → Output Format structure giving users a predictable starting point for recurring work. (see AI-010)

AI Product Architecture

The blueprint organizing frontend, backend, AI services, knowledge, and integrations into a maintainable system. Mostly placement decisions: where each responsibility lives and what happens when it fails, as the reply-suggestions worked example shows.

AI Service Layer

The layer owning prompt construction, context assembly, memory, model routing, response validation, and guardrails, kept separate from business logic so both evolve independently. Putting all logic into prompts is the classic anti-pattern.

Latency Budget

The hard time limit user experience imposes: reply suggestions must land in under 1.5 seconds or typing outruns them, forcing a small fast model, streaming, and aggressive caching. (see AI-008)

Context Recipe

The deliberate choice of what enters the prompt (last 10 messages plus thread topic, not the whole channel history) for cost, privacy, and attention-quality reasons. (see AI-020)

Degradation Path

What happens when the AI dies: the feature quietly disappears, but chat itself never blocks on AI. If the whole product dies with the model endpoint, that's an architecture smell. (see AI-030)

Privacy Boundary

The line where data leaves your control: messages going to a provider require a data-processing addendum, no-training clauses, and regional routing for enterprise tiers. (see AI-056)

Build vs Buy

The per-component decision to build, buy, or use a managed service: auth is usually bought, core business logic built, vector databases depend on scale. It drives development speed and total cost of ownership.

Provider Gateway

Abstracting the model API behind an internal service so the provider can be swapped without touching forty call sites. (see AI-040)

Minimum Viable Product (MVP)

The simplest version delivering real value while enabling rapid learning. The contract summarizer took four weeks: fake it, thin slice, trust features, charge. Not unfinished software; the smallest product worth using.

Concierge Test

Week 1's move: no product at all, a human runs the AI workflow manually and emails results back. 20 users and "the summary missed payment terms" — learning at zero build cost.

Feature Prioritization

Sorting every idea into Must Have / Should Have / Could Have / Future and building only the must-haves. It's the discipline that returns every new suggestion to the framework instead of the backlog.

Build–Measure–Learn

The Lean Startup cycle applied to AI: build, release, measure, improve, repeat, with every iteration answering a specific question about user behavior or value.

Feature Creep

Scope expanding as every suggestion becomes a feature: delayed releases, confusing interfaces, rising costs. The contract MVP deferred RAG entirely because contracts fit in the context window. (see AI-061)

Kill Criteria

Failure conditions agreed before building, such as "if fewer than X of Y users do Z by week 4, we stop." The cheapest founder discipline there is.

Technical Debt

Acceptable shortcuts that accelerate learning, but never on security, privacy, data integrity, or reliability, and always with a resolution plan.

Trust Features

The week-3 additions that doubled conversion: clause-level "show source" links and a visible "verify amounts" note. Trust is a shippable feature, not a vibe. (see AI-043)

Metrics Tree

The layered instrumentation of one feature: north star (hours saved per user per week), product metrics (acceptance rate, edit distance, retention), quality metrics (groundedness, thumbs-down), system metrics (p95 latency, cost per draft). Its power is diagnosis: the tree traced an "AI got worse" report to a mobile CSS bug in minutes.

North Star Metric

The single outcome metric everything else serves: hours saved per active user per week for the email-drafting feature. An accurate AI nobody uses is not a successful product.

Acceptance Rate

The share of AI suggestions users actually adopt (42% in the example), a product metric that captures real value delivery better than raw usage counts.

Edit Distance

How much users change AI output before using it (31% of characters), a proxy for draft quality that caught the mobile truncation bug when quality metrics stayed flat.

Retention

Whether users return (Day 1, Week 1, Month 1), one of the strongest indicators of sustained value, and harder to fake than adoption spikes.

Vanity Metric

Impressive-looking numbers that inform no decision: total prompts sent, parameter counts, feature counts. Define "active use" with the harsher option: accepting output, not just opening the panel.

Leading vs Lagging Metrics

Leading metrics (trial usage, feature adoption) predict; lagging metrics (revenue, retention, ROI) confirm. Both are needed for trajectory.

A/B Testing

Controlled comparison of two variants across adoption, satisfaction, task success, cost, and quality, letting evidence beat internal opinion. (see AI-058)

Phased Rollout

Expanding access gradually: internal → beta → pilot → 5% → 20% → 100%, so issues surface before they reach everyone. Boring launches are engineered, not lucky: the worked example's provider slowdown at hour 3 triggered the fallback and users never noticed.

Eval Freeze

Locking golden-set scores a week before launch: any prompt or model change after freeze restarts the clock. The quality half of go/no-go. (see AI-022)

Red-Team Pass

Running adversarial prompts (injection, PII extraction, brand-damage requests) before launch; the example found two failures, gated them, and retested. (see AI-027)

Capacity Math

Projected launch-day traffic × tokens per request versus provider rate limits: the example carried 4× headroom plus a configured fallback provider. (see AI-008)

Rollback Rehearsal

Actually executing the rollback before you need it; a rollback that has never been run is a hope, not a plan. (see AI-029)

Comms Honesty

Reviewing marketing copy against eval results: "never makes mistakes" was removed because the evals said otherwise. Overclaiming is a disclosure risk. (see AI-060)

Support Readiness

The top-10 expected complaints with prepared responses and an escalation path carrying trace IDs to engineering; support should know the product before customers do. (see AI-028)

Change Management

Communication, training, executive sponsorship, and feedback channels for the humans whose work the product changes; the organizational half launches forget.

Scaling

Growing users, capabilities, infrastructure, and business together while holding quality and cost: the study-helper's 2,000→90,000 users broke the bill, the provider limits, and the eval coverage, and none of it was the model. Scaling AI is scaling the system around the model.

Linear Costs

Costs that grow with every single user: tokens, storage, human review. The linear list is what breaks at 50×; each item needs a sub-linear plan: cache, route, batch. (see AI-039)

Rate Limits

Provider caps hit at peak hours as usage grows — mitigated with request queuing, graceful "high demand" messaging, an overflow provider, and pre-negotiated quota raises. (see AI-024)

Eval Coverage Drift

New users bringing new inputs the golden set never covered — chemistry questions doubling thumbs-down on that slice. Fix: mine production logs into the eval set weekly. (see AI-025)

Capacity Planning

Forecasting compute, API quotas, storage, and headcount from growth trends so expansion is proactive, not an emergency reaction.

Platform Thinking

The evolution from product to platform (plugins, developer APIs, extensions, marketplace, partner ecosystem), distributing growth across many contributors.

High Availability

Staying up through component failures via redundancy, multi-region deployment, automatic failover, and health checks; reliability becomes a competitive advantage at enterprise scale.

Product-Market Fit

The precondition for scaling: strong retention, organic growth, users measurably worse off without the product. Scaling before fit is the classic fatal mistake.

Capstone Project

Designing a complete AI product from problem discovery through launch and scaling. The worked example compresses eleven lessons into one page for a freelance-designer proposal tool. If a curriculum section leaves no trace in your page, revisit it.

Product Vision

The aspirational statement of what the product achieves, for whom, and what success looks like; it's the tiebreaker when priorities conflict.

Product Lifecycle

Problem Discovery → Research → Product Thinking → UX → Architecture → MVP → Metrics → Launch → Scaling → Continuous Improvement. This is the full cycle the capstone runs once, as the foundation for repeating it. (see AI-041)

One-Page Scoping

Time-boxing the whole design (problem, evidence, solution shape, architecture, MVP plan, metrics tree, risks) to 90 minutes on one page, then sharing it with one honest reader whose first confused question marks the section to rework.

Kill Criterion

The pre-agreed failure condition (fewer than 3 of 10 pilots active by week 6) that stops the project honestly instead of letting it drift. (see AI-046)

Executive Presentation

Outcome-focused communication for leadership on problem, solution, business value, launch, and growth, built around confidence in business outcomes rather than technical detail.

Evaluation Rubric

The capstone's balance test: problem understanding 20%, architecture 20%, UX 15%, AI design 15%, scalability, launch, and business value the rest. A technically brilliant product with unclear business value fails.

Business Case

The structured investment argument: problem, solution, expected benefits, risks, and timeline to value. It's what turns a design into a funded project.

Fine-tuning

Continuing to train a pre-trained model on a smaller curated dataset so its weights shift toward the behavior you want — the generalist consultant trained into a full-time employee who no longer needs the briefing document. Not training from scratch: you're adjusting defaults, not teaching language. (see AI-005)

LoRA (Low-Rank Adaptation)

The most widely used parameter-efficient technique: freeze the base weights, add two small trainable matrices A and B whose product approximates the update, merge at inference (W' = W + A×B). Only 0.1–1% extra parameters — a 7B model fine-tunes on a single consumer GPU.

QLoRA (Quantized LoRA)

LoRA with the frozen base model loaded in 4-bit precision to slash GPU memory, keeping adapter weights in full precision — the technique that put large-model fine-tuning within reach of consumer hardware. (see AI-065)

PEFT (Parameter-Efficient Fine-tuning)

The family of methods updating only a small subset of parameters while freezing the rest — near full fine-tuning quality at a fraction of the cost, since full fine-tuning of billions of weights is impractical for most teams.

Instruction Tuning

Training on (instruction, response) pairs so a base model follows instructions reliably — how base models become chat assistants. (see AI-067)

RLHF (Reinforcement Learning from Human Feedback)

Using human preference annotations to train a reward model that guides reinforcement learning, aligning outputs with human values. DPO achieves similar results by training directly on preference data, skipping the reward model.

Catastrophic Forgetting

Losing previously learned capabilities when fine-tuning aggressively on new data — managed with careful learning rates, regularization, and mixed-task training data.

The Decision Triangle

Prompting for quick iteration, RAG for dynamic knowledge, fine-tuning for consistent style and narrow task mastery — and the best systems combine all three. Avoid fine-tuning with under 100 quality examples or frequently changing knowledge. (see AI-016)

Multimodal AI

A system that processes and/or generates more than one data modality — text, images, audio, video — reasoning across them simultaneously, like the doctor who integrates the X-ray, the patient's account, and the lab report. Understanding models take multimodal input and answer in text; generation models produce images, audio, or video.

Vision-Language Model (VLM)

An image encoder (typically a Vision Transformer) bolted onto an LLM backbone via a projection layer, so image tokens are injected into the sequence alongside text tokens and the language model attends to both. (see AI-012)

Vision Transformer (ViT)

The transformer adapted for images — divide the image into fixed-size patches, treat each patch as a token, attend across all patches. A 512×512 image might become 256 patch tokens.

Image Token

An image's cost in the context window — typically 100–2,000 tokens depending on resolution and architecture, directly affecting API cost and context budget. (see AI-059)

Diffusion Model

A generator trained to reverse a noise process — learn to denoise progressively noisier images, then at inference walk from pure random noise to a coherent image guided by a text prompt. (see AI-011)

Latent Diffusion

Diffusion run in a compressed latent space (via a Variational Autoencoder) instead of pixel space — the trick behind Stable Diffusion and FLUX that makes high-quality generation feasible on consumer hardware.

CLIP (Contrastive Language-Image Pre-training)

The model aligning image and text representations in one shared embedding space, enabling zero-shot classification, semantic image search, and the text conditioning behind many diffusion models. (see AI-014)

Whisper

OpenAI's open speech-recognition model, trained on 680,000 hours of multilingual audio and transcribing 99 languages with word-level timestamps — the standard audio-input building block for multimodal pipelines.

Human-in-the-Loop (HITL)

Building AI systems where humans remain meaningfully involved in consequential decisions — not a sign of AI weakness but an architectural choice. The most reliable AI products are not the ones that automate everything; they are the ones that know exactly when to hand off to a human.

Human-as-the-Loop

The AI is purely advisory; the human makes every decision — the lawyer reads the AI's case suggestions but files everything themselves. For safety-critical domains or immature, untrusted models.

Human-on-the-Loop

The AI acts autonomously on most decisions while a human monitors the stream and can override — high-confidence content removals with a moderator sampling decisions. For high-volume, lower-stakes work after trust is established.

Confidence Threshold

The routing boundary: above it auto-approve, below it reject or review, the middle band goes to the queue. Insufficient alone — models can be confidently wrong (miscalibration), so 95% stated confidence is not 95% accuracy. (see AI-025)

Automation Bias

Reviewers deferring to AI recommendations even when the AI is likely wrong — the failure that makes "a human checked it" decorative rather than protective. (see AI-043)

Active Learning

Having the model route its most uncertain examples to human reviewers first, maximizing learning per label and turning the review queue into a training-data engine. (see AI-006)

Trust Calibration

Systematically reducing human oversight as the model demonstrates validated reliability — the controlled path from human-as-the-loop toward automation, never the reverse leap.

Inter-Annotator Agreement

How consistently different reviewers label the same item (measured with statistics like Cohen's Kappa) — the health metric of annotation quality and guideline clarity.

Cost-Per-Interaction (CPI)

The variable cost of serving one user request: (input tokens × input price) + (output tokens × output price). Profile a representative sample of real interactions rather than guessing — token counts vary widely by use case.

Token Economics

The core financial reality of AI products: you pay every time it is used, not once to build it. Flagship models cost 10–50× more per token than small ones, and output tokens cost several times input tokens. (see AI-059)

Cost of Goods Sold (COGS)

All variable and semi-variable serving costs — LLM APIs, embeddings, infrastructure, storage, third-party APIs, human review. Unlike traditional software, AI COGS scales with usage, so unmodelled growth can destroy margins.

Gross Margin

(Revenue − COGS) / Revenue — mature AI SaaS targets 70–80%. The metric that decides whether growth compounds profit or losses.

Unit Economics

Profitability analyzed at the level of one user or one interaction — establishing sustainable margin before scaling. If each user loses money, volume makes it worse. (see AI-046)

Usage-Based Pricing

Charging in proportion to consumption (per message, document, or API call), aligning revenue with variable cost — the natural hedge against AI's pay-per-use cost structure.

Budget Guardrail

Per-user rate limits, billing alerts, and circuit breakers preventing usage spikes, runaway automations, or viral growth from blowing the budget. (see AI-039)

Prompt Caching

Provider-side reuse of repeated prompt prefixes at a discounted rate — a direct CPI reducer for products with consistent system prompts. (see AI-061)

A/B Testing

Exposing a control group to the existing version and a treatment group to the change, comparing outcomes on a predefined metric. For AI features the control is already a distribution of outputs, not a fixed state — so log the actual response, not just the group assignment.

Non-Determinism

Identical inputs producing different outputs across calls due to temperature and sampling — the property that makes AI experiments a comparison of two distributions rather than two fixed experiences. (see AI-009)

Shadow Testing

Running the candidate on real traffic in parallel, logging outputs for offline analysis but never serving them — quality comparison at zero user risk. (see AI-026)

Canary Release

Exposing the new variant to ~1% of users and expanding as confidence grows — early problem detection before full deployment. (see AI-029)

Guardrail Metric

A metric that must not degrade even if the primary metric improves — latency, safety violation rate, cost per query. The insurance against optimizing one number at another's expense. (see AI-025)

Novelty Bias

Users engaging more with a new feature simply because it is unfamiliar, inflating early treatment metrics — one reason AI experiments need longer windows than button-color tests.

Delayed Feedback

The lag between an AI response and its measurable effect — a slightly wrong summary eroding trust over weeks. Effects that take time to manifest require patient measurement horizons.

Minimum Detectable Effect (MDE)

The smallest improvement the experiment can detect with significance — computed before starting to set sample size and duration, preventing underpowered experiments that "prove" nothing.

EU AI Act

The world's first comprehensive horizontal AI regulation (in force since August 2024), classifying systems into four risk tiers: unacceptable (prohibited), high (conformity assessment, human oversight, documentation), limited (disclose AI nature), minimal (voluntary codes). Most consumer chatbots land in limited/minimal; CV screening and credit scoring land in high.

GDPR

The EU's data protection law governing how organizations collect, process, retain, and delete personal data, including everything an AI system touches, from training data to conversation logs. Product teams make GDPR-affecting decisions daily. (see AI-037)

Automated Decision-Making

Decisions about individuals made without meaningful human involvement. GDPR Article 22 grants the right to human review and to contest significant decisions, one reason high-stakes AI ships as decision support. (see AI-053)

Lawful Basis

One of six GDPR Article 6 grounds (consent, contract, legal obligation, vital interests, public task, legitimate interests) required before an AI system may process personal data at all.

AI Impact Assessment

The structured pre-launch evaluation of an AI feature's risks to individuals and groups (data flows, vulnerable populations, mitigations, residual risk), analogous to a GDPR Data Protection Impact Assessment.

Data Processing Agreement (DPA)

The contract between controller and processor specifying purposes, security measures, and data-subject obligations; required whenever your AI product sends user data to a model provider. (see AI-045)

IP Indemnification

A model provider's contractual commitment to defend customers against IP claims arising from model outputs, a key vendor-contract term while courts actively rule on AI copyright cases.

Data Residency

The requirement that personal data and AI processing stay within a geographic region, constraining which providers and inference endpoints an enterprise product may use, and motivating EU-only routing paths.

Prompt Version Control

Tracking every prompt change as a discrete, auditable commit (who changed what, when, why), with rollback to any previous state. The antidote to the shared-Notion-doc failure where a "clarity tweak" silently dropped deflection 12% with no git blame to find it.

Semantic Versioning (Prompts)

MAJOR.MINOR.PATCH applied to prompts: major for breaking changes to output format or behavior, minor for non-breaking improvements, patch for typo fixes. A single word ("helpful" → "friendly") can warrant a version bump because it shifts tone in production.

Prompt Registry

A central service storing versioned prompts, metadata, and deployment history, from which applications fetch the right version at runtime, decoupling prompt deploys from code deploys.

Prompt Template

A parameterized prompt with variable slots like {{customer_name}}, versioned as a structure independently of the runtime values that fill it. (see AI-010)

Golden Set

The curated inputs-and-expected-outputs collection run against every new prompt version to catch regressions before promotion. (see AI-022)

Prompt Drift

Behavior changing while the prompt text hasn't, caused by silent provider model updates, provider-side system prompt changes, or shifting input data. Detected only by continuous evaluation, never by reading the prompt. (see AI-007)

Prompt Staging Pipeline

Dev → staging → production environments a prompt change must pass, each with progressively stricter evaluation and access control, the same discipline software adopted forty years ago.

Regression Threshold

The pre-defined line a new version must not cross (e.g., no more than a two-point drop in golden-set accuracy versus current production), turning "looks fine to me" into a gate. (see AI-026)

Capability Commoditisation

A differentiating AI capability becoming available to everyone via provider API updates or open-source releases: the Q1 summarization edge that GPT-4o gives away free by Q3. The analysis question is never "what can they do today?" but "what advantage survives when capabilities converge?"

Data Moat

Proprietary data (from user interactions, exclusive partnerships, closed systems) that improves the AI in ways competitors can't replicate without equivalent data. One of the four foundations that outlast the ever-taller towers of model capability.

Workflow Moat

An AI feature so embedded in daily process (integrations, accumulated personalization, habit) that switching means significant re-learning and lost context.

Trust Moat

Brand capital from consistent accuracy, responsible data practices, and honest limitation disclosure, most durable in risk-averse domains like healthcare, legal, and enterprise security. (see AI-043)

Network Moat

Advantage compounding with usage: more feedback data improving the model, more user content benefiting others, more social proof de-risking purchases for new buyers.

Moat Assessment

The structured durability check across data, workflow, trust, network, and speed: separating temporary capability leads (sand) from structural advantages (deep footings).

Failure Mode Analysis

Probing how a competitor's product behaves when wrong, uncertain, or out of scope: the fastest read on their error handling, feedback loops, and honesty about limitations. (see AI-060)

Competitive Intelligence Cadence

Weekly scans, monthly landscape-table updates, quarterly deep-dives: the rhythm that keeps analysis current in a market where printed maps of the skyline are obsolete on arrival.

Forward Diffusion Process

The fixed, unlearned process of gradually adding Gaussian noise to a training image until it becomes indistinguishable from random noise.

Reverse Diffusion Process

The learned process where a neural network predicts and removes noise step by step, starting from random noise, to produce a coherent image. (see AI-052)

U-Net

The neural network architecture originally used in diffusion models to predict the noise at each denoising step.

Diffusion Transformer (DiT)

A newer diffusion model architecture that replaces the U-Net with a transformer, used in models like Sora and FLUX. (see AI-012)

Latent Diffusion

Running the noise/denoise process in a compressed VAE latent space rather than on raw pixels, the efficiency technique behind Stable Diffusion.

Variational Autoencoder (VAE)

A model trained to compress an image into a small latent representation and reconstruct it back, used to enter and exit latent space in latent diffusion.

Cross-Attention

The attention mechanism connecting a denoising network's image latents to text embeddings, steering generation toward a text prompt at every step. (see AI-013)

Classifier-Free Guidance

A technique that runs the diffusion model with and without text conditioning at each step, then exaggerates the difference to strengthen prompt adherence.

Sampler

The algorithm (DDPM, DDIM, DPM++) controlling how the reverse diffusion process steps from noise to image, trading off step count against image quality.

Text Encoder

A model (CLIP or T5) that converts a text prompt into embeddings used to condition image generation via cross-attention. (see AI-052)

Diffusion Transformer Video Model

A diffusion transformer extended across the time dimension to generate video, as in Sora, denoising a sequence of latent frames jointly.