GUIDE 1.1  •  FOUNDATIONS

How AI models work

Not a technical deep dive. A working mental model for B2B marketers: what an LLM actually does, why it gets things wrong, why it sounds so confident when it does, and what you need to understand to use it effectively for marketing work.

8 min read|Guide 1.1 of 4 in Foundations|No technical background required
AFTER READING THIS YOU WILL KNOW
Why AI predicts text rather than retrieving facts
What hallucination actually is and how to reduce it
Why the same prompt gives different results each time
What a context window is and when it matters
The difference between standard and reasoning models
Which tools to use for which marketing tasks at INR pricing

It is not searching. It is predicting.

Knowing how AI models work gives B2B marketers a practical edge — it changes how you write prompts, where you trust outputs, and which tools are worth the spend. This guide is written for marketers, not engineers: plain language, no maths, just the mental models you need to use AI effectively. According to McKinsey’s State of AI research, the organisations getting the most value from AI are those where non-technical teams understand the capabilities and constraints. Read Prompting 101 next to put this into practice.

The single most useful thing to understand about an LLM is that it does not look things up. It predicts. When you send a prompt, the model is doing one thing: working out what text is most likely to come next, based on patterns it learned from an enormous amount of written material.

That is it. There is no database it is querying. There is no search happening. It has seen enough text during training that it has developed a sophisticated ability to predict what a useful, coherent, relevant response looks like for a given input. Most of the time, that prediction is genuinely useful. Sometimes it is wrong in ways that are hard to spot because the prediction still sounds completely plausible.

This is why the same prompt can produce different outputs on different days. There is an element of statistical sampling in how the model generates text. It is not returning a fixed stored answer. It is generating a response fresh each time, which is also why it can be creative and why it can be inconsistent.

HOW TO THINK ABOUT IT
Think of a very experienced copywriter who has read millions of pieces of marketing content. They have never worked at your company and do not know your product. But if you give them a brief, they can produce something that sounds credible and well-structured, because they have seen enough similar work to know what good looks like. They can also confidently get specific details wrong if you do not give those details to them. That is roughly what an LLM is doing.

Why it sounds confident when it is wrong

Hallucination is the term used when an AI model produces something that is factually incorrect but stated with complete confidence. It happens because the model is optimised to produce fluent, plausible text, not to produce verified facts. It does not know the difference between something it knows to be true and something that pattern-matches to what a true answer would sound like.

For B2B marketing, this has specific consequences. Statistics, competitor data, analyst quotes, product specifications, and company details are all high-risk categories. An AI model will confidently cite a Gartner statistic that does not exist, attribute a quote to a real person who never said it, or describe a competitor product feature that was removed two years ago.

The practical rule is straightforward: never publish a specific fact, statistic, competitor claim, or named quote from AI output without verifying it against a primary source. Use AI for structure, tone, and framing. Verify anything specific.

HIGH-RISK VS LOW-RISK TASKS
Verify before you use
Statistics and market data
Competitor product features or pricing
Analyst or research quotes
Named person quotes or attributions
Regulatory or compliance claims
Company news or recent events
Generally safe to use directly
Email and copy structure and tone
Headline and subject line variations
Content outlines and briefs
Rephrasing existing content you provide
Brainstorming angles and positioning ideas
First drafts when you supply the key facts

The context window: why it matters for longer tasks

Every time you interact with an AI model, it can only see a certain amount of text at once. That limit is called the context window. Think of it as the model’s working memory for a single conversation. Anything outside that window, it cannot see and cannot use.

Modern models have large context windows, typically enough for tens of thousands of words. For most B2B marketing tasks, this is not a constraint. But it becomes relevant when you are working with long documents. If you paste a 40-page research report and ask for a summary, earlier sections of the document may be less accurately represented than later ones as the model works through the context. This is sometimes called the lost-in-the-middle problem.

The practical fix is straightforward: for long documents, work in sections rather than pasting everything at once. Summarise section by section, then synthesise the summaries. You will get more accurate output than a single pass over a very long input.

WHEN CONTEXT WINDOW MATTERS IN MARKETING WORK
Summarising a long report
Work section by section. Summarise each, then ask for a synthesis of the summaries. More accurate than one big paste.
Long email threads
Paste the full thread and ask AI to extract action items or draft a response. Context is usually fine for email chains.
Multi-chapter whitepapers
Do not paste the whole thing and ask for a summary. Break it into logical sections and summarise each before synthesising.
Transcripts and call notes
Sales call transcripts can be long. Paste in chunks and ask for the key buyer objections, pain points, or stated priorities from each chunk.
Competitor website analysis
Paste sections of competitor copy and ask for a positioning comparison. Works well as long as you do not paste too many pages at once.
Campaign briefs
Keep your brief template concise. A 200-word brief with clear instructions will produce better output than a 2,000-word document with vague guidance.

Standard models vs reasoning models

Most AI tools you use in marketing are what are called standard generative models. You give them a prompt and they generate a response. They are fast, generally good at language tasks, and work well for the majority of what a B2B marketer needs to do: drafting copy, restructuring content, brainstorming, rephrasing, summarising.

Reasoning models are a newer category. They are slower and typically more expensive, but they spend more time working through a problem before producing an answer. They are better at tasks that require multi-step logic: analysing a complex brief, comparing multiple options against a set of criteria, or working through a strategic problem where the answer is not just a matter of good language.

For most marketing tasks, a standard model is the right choice. Use a reasoning model when you are trying to solve something structural: a positioning decision with multiple tradeoffs, a GTM sequencing question, or a campaign brief that needs to hold together logically across channels. For writing tasks, standard models are faster and produce output that is just as good.

Standard generative models
ChatGPT 4o, Claude Sonnet, Gemini 1.5 Pro
Email drafts and copy
Content outlines and briefs
Summarising documents
Rephrasing and editing
Generating headline variations
Research synthesis from provided text
Reasoning models
ChatGPT o3, Claude 3.7 Sonnet (extended thinking), Gemini 2.0 Flash Thinking
Complex positioning decisions
GTM strategy with multiple constraints
Competitor analysis with tradeoff evaluation
Campaign briefs needing logical coherence
Evaluating ICP fit across multiple criteria
Multi-step workflow planning

The main tools and what they cost in India

Three models dominate B2B marketing use. Each has a free tier and a paid plan. Here is what each costs, what it is relatively better at, and the honest assessment of where each falls short.

ChatGPT
OpenAI
Free tier available
Plus plan: ~₹1,700/month
STRONGEST FOR
Widest range of built-in tools and plugins
Strong at long-form drafts and structured documents
Image generation included (DALL-E)
Custom GPTs for reusable prompt templates
Web search on paid plans for current information
WATCH FOR
Tends toward verbose output by default. Can sound generic without detailed prompting. Free tier is significantly slower and rate-limited. Data used for training unless you opt out in settings.
Claude
Anthropic
Free tier available
Pro plan: ~₹1,700/month
STRONGEST FOR
Best at following nuanced instructions accurately
Strongest for tone calibration and brand voice work
Handles long documents better than most
Less likely to add unnecessary padding or filler
Strong at rewriting and editing existing copy
WATCH FOR
No image generation. Web search available but not as deep as ChatGPT plugins. Occasionally more cautious about certain content types. Fewer third-party integrations than ChatGPT.
Gemini
Google
Free tier available
Advanced plan: ~₹2,100/month
STRONGEST FOR
Tightly integrated with Google Workspace (Docs, Sheets, Gmail)
Strong at multimodal tasks: analysing images and PDFs alongside text
Good for teams already in the Google ecosystem
Gemini in Gmail and Docs reduces context switching
Strong real-time web access and Google Search integration
WATCH FOR
Writing quality is slightly behind Claude and ChatGPT for nuanced copy. Less predictable for creative or brand voice tasks. The value is highest if your team lives in Google Workspace.
Pricing note: INR pricing varies based on exchange rate and may include GST. All three tools bill in USD and convert at the prevailing rate. At current rates (April 2026), the paid plans are approximately ₹1,600 to ₹2,200 per month per user. Check current pricing directly on each tool's website before purchasing. All three have free tiers that are genuinely useful for getting started.
WHERE TO START
Pick one tool and spend four weeks using it for real work before adding another. The biggest mistake is subscribing to all three and using none of them deeply. If you are starting from zero, begin with ChatGPT Plus. The breadth of tools, the custom GPT feature for building reusable prompt templates, and the size of the community around it make it the fastest to get value from. Switch to or add Claude if your primary work is copy-heavy: brand voice, long-form drafts, or careful editing. Add Gemini if your team is Google Workspace-first.

Five things to remember when using any AI model

1
It is predicting, not retrieving
There is no database. It generates text based on patterns. The output is plausible, not guaranteed to be accurate.
2
Verify anything specific
Statistics, quotes, competitor claims, and recent events are high-risk. Always check against a primary source before publishing.
3
The first output is a draft
Treat it as a starting point. Iterate. The second or third version is usually significantly better than the first.
4
Garbage in, garbage out
Vague prompts produce generic output. The more specific context, instructions, and constraints you give, the better the result.
5
Context window has limits
For long documents, work in sections. Do not paste a 30-page document and expect a perfect summary on the first pass.

Get the Digital Uncovered newsletter

B2B marketing intelligence for India. No padding. No filler. Straight to the point, twice a week.