If you ask an AI to create something for your company, what does it actually need to know?
The obvious answer is your brand guidelines.
But that is only one layer.
A system can know your logo, fonts, colors, and tone of voice and still make strategically bad creative decisions.
Because creative work is downstream of much more than brand.
A useful AI creative system needs a working model of the company itself.
01
What the company does
What you sell, how the business works, the products, services, capabilities, strategies, offerings, and distinctions that matter inside the category.
02
Who the audiences are
Customers, prospects, investors, employees, partners, media, recruits, and what each audience already knows, cares about, fears, and needs to believe.
03
Positioning
What the company should own in someone's mind, how it is meaningfully different, what alternatives it is compared against, and where it resists category conventions.
04
Messaging architecture
Primary and supporting messages, proof points, value propositions, objections, claims, language to avoid, and which messages matter to which audiences.
05
Evidence
Customer names, metrics, case studies, experience, research, outcomes, testimonials, credentials, and product capabilities that substantiate claims.
06
Brand identity
Logo, color, typography, image direction, illustration, motion, shape language, layout principles, tone of voice—and how those ingredients behave.
07
Creative precedents
Approved sections, slides, campaigns, and examples that should become the company's own reference library.
08
Rejections
What was tried and rejected, why it was rejected, and what that teaches the system about future choices.
09
Goals
What the company is trying to accomplish now: raise a fund, launch a product, recruit, move upmarket, enter a category, explain complexity, or increase credibility.
10
Quality standards
What “good” means for this company: information density, boldness, proof expectations, writing polish, visual restraint, and what should never feel generic.
11
History
What changed in positioning, leadership, offerings, and brand—and why those changes happened.
12
What is relevant right now
The system must know which slice of company knowledge is useful for the task in front of it.
Why each layer matters
If the system misunderstands the business, everything downstream inherits the error.
If it treats every audience the same, the company will sound generic.
If it lacks positioning, it will create competent category-average work.
If it lacks messaging architecture, it will try to say everything everywhere.
If it lacks evidence, it will produce confident assertions without support.
If it knows brand assets but not brand behavior, it can remain technically compliant while feeling wrong.
If it forgets precedents and rejections, it repeats mistakes and throws away learning.
If it ignores goals, it can make beautiful work that solves the wrong business problem.
From company database to company understanding
It is tempting to think the answer is simply “put everything into a knowledge base.”
That is a start.
But a pile of files is not understanding.
→
Understanding
Structure + relationships + priority + recency + provenance + retrievalThe system needs rules about what can change and what should remain stable.
And ideally, it should learn from the creative work itself.
This is what we mean by company intelligence
At Morphic, we think the valuable layer around generative models is increasingly the persistent understanding of the company.
The model will keep getting better.
The generation will keep getting better.
But the model still needs to know who it is working for.
A great creative partner is not great because they can make attractive things.
They are great because they understand the business well enough to know what the right thing is.
AI creative systems should be held to the same standard.
