Data Gravity: Why Your Company Knowledge Should Live Where Your AI Works

23 September 202616 min read

Co-founder at Superkind

A central mass pulling scattered fragments inward, representing data gravity consolidating company knowledge

Ask any operations leader where the answer to a specific customer question lives, and watch what happens. They will not name one place. They will name six. It is probably in the CRM, or maybe in someone’s inbox, or in a Teams thread from March, or in a SharePoint folder nobody has opened since the last reorg, or in the ERP, or in the head of the one person who is on holiday this week.

This is the real state of company knowledge in most mid-sized and enterprise businesses. Not missing - scattered. And scattered knowledge can never become leverage, because leverage requires that the knowledge be reachable from one place, by both your people and, increasingly, by AI. There is a name for the force that keeps pulling knowledge apart and makes it so hard to bring back together. Engineers call it data gravity.

This article is for the CTO, operations lead, or Geschaeftsfuehrer who has heard the AI pitch, maybe even run a pilot, and quietly noticed that the tools are impressive but the results are thin. The reason is almost never the model. It is that the knowledge the AI needs is spread across a dozen systems, and no assistant can act on context it cannot reach.

TL;DR

Data gravity is the tendency of data to attract applications, services, and more data - coined by engineer Dave McCrory in 2010. The bigger a data store gets, the more work moves toward it, because moving the data is harder than moving the work.

Scattered knowledge is the tax you already pay. Employees lose 1.8 to 2.5 hours a day just searching for information that already exists somewhere in the company.

AI makes this urgent. Gartner predicts organizations will abandon 60 percent of AI projects through 2026 for lack of AI-ready data. An AI that cannot reach your context cannot do your work.

The answer is a centre of gravity. A Company Brain consolidates the knowledge that matters and connects to your real systems, so AI employees act from complete context instead of fragments.

The payoff compounds. Knowledge that survives turnover, and more output without more headcount, because the leverage lives in the company and not in the individuals.

What Data Gravity Actually Means

The term comes from cloud engineering, but the idea is older than any cloud. In 2010, engineer Dave McCrory coined “data gravity” as an analogy to physical mass: large accumulations of data attract applications, services, and more data, the same way a planet attracts objects around it¹. The more data you gather in one place, the more everything else wants to be near it.

Mass attracts - As a data store grows, applications and services are drawn toward it because being close means faster access and less friction¹.
Data resists moving - The larger a dataset gets, the harder and slower it is to relocate, so the work moves to the data rather than the other way around¹⁵.
Gravity compounds - Once workflows cluster around a data store, they generate still more data there, which strengthens the pull further¹⁸.
It can be natural or artificial - McCrory distinguished naturally occurring gravity from forces created by pricing, throttling, and lock-in that trap data in a location on purpose¹.
It decides architecture - Where your data has gravity determines where your applications, and now your AI, can realistically run¹⁹.

“When you have a sufficient enough amount of data, it ends up with an attractive force like when you have a large enough amount of mass, say a planet or something, it has an attractive force.”

- Dave McCrory, creator of the term “data gravity”²

For years this was a data-centre conversation about latency, egress fees, and where to run workloads. But the same physics governs something much closer to the daily reality of a business: your company knowledge. When knowledge accumulates in one reachable place, it pulls work, tools, and decisions toward it. When it is scattered, nothing can build on it.

From data centres to company knowledge

The mistake most companies make is treating knowledge as if it has no gravity at all - as if a document in SharePoint, a decision in a Teams thread, and a customer note in the CRM are all equally reachable. They are not. Each lives in its own well, with its own owner and its own version of the truth.

Data gravity in the data centre	Data gravity in your company knowledge
Workloads run near the data to cut latency	Work gets done near where the knowledge already lives
Moving large datasets between clouds is slow and costly	Moving knowledge out of an expert’s head or inbox rarely happens at all
Egress fees punish every transfer between regions	Every copy-paste between tools is a hidden tax on your team’s time
Consolidation improves performance and control	Consolidation into a Company Brain lets AI act from complete context
Silos raise cost and slow analytics	Silos mean no single source of truth and constant rework

The lesson from a decade of cloud architecture is simple and it transfers directly: you do not fight data gravity, you decide where it should be strongest. For a company that wants AI to actually work, the answer is one consolidated place the knowledge orbits - not a dozen.

The Cost of Scattered Company Knowledge

Fragmented knowledge does not announce itself as a line item. It hides inside every process as delay, rework, and lost time. But the aggregate cost is enormous, and the research is consistent.

Hours lost to searching - McKinsey found employees spend on average 1.8 hours a day, roughly 9.3 hours a week, searching and gathering information⁷. IDC puts it even higher at about 2.5 hours a day, near 30 percent of the workday⁸.
Tool sprawl keeps growing - The average company now runs well over 100 SaaS applications, with large enterprises far higher, and adds far more each year than it retires⁹.
Most of it never reaches IT - A majority of SaaS apps are adopted without IT’s knowledge, so the knowledge inside them is invisible to any central system¹⁰.
Silos undercut AI directly - Fragmented data is repeatedly cited as a top barrier to getting value from AI, because the context the model needs is trapped in systems it cannot reach¹¹.
Most knowledge is unstructured - Gartner estimates around 80 percent of enterprise data is unstructured - locked in emails, documents, tickets, and transcripts rather than tidy databases⁶.
The single-source-of-truth problem - When the same fact lives in six places, none of them is authoritative, and every decision carries the risk of being made on a stale version¹².

Key Data Point

If a knowledge worker loses two hours a day to searching and re-searching for information that already exists, that is roughly a quarter of every salary you pay - spent not on the work, but on finding the inputs to the work. Multiply that across a department and the cost of scattered knowledge stops being abstract.

Where the knowledge actually hides

The knowledge you need to run the business is real and it exists. The problem is purely location. A single customer relationship might be documented across all of the following at once.

System	What it holds	Why it stays stuck there
Email and Teams	Decisions, commitments, context behind the deal	Private to individuals, unsearchable across the company
SharePoint and drives	Contracts, specs, proposals, versions	Folder structures nobody agrees on, duplicates everywhere
CRM	Contacts, pipeline, activity history	Only as complete as the last person who updated it
ERP	Orders, invoices, delivery, inventory	Structured but walled off from the surrounding context
People’s heads	The why behind exceptions and workarounds	Never written down, leaves when they do

No AI assistant, however capable, can turn this into leverage as long as it has to hop between wells and most of the context never reaches it. The fragmentation is the ceiling.

Why Data Gravity Matters More in the AI Era

Scattered knowledge was always a drag on productivity. What changed is that AI has raised the stakes: the same fragmentation that slowed your people now caps what your AI can do. Here is why the problem became urgent in the space of two years.

AI is only as good as its context - A general model with no access to your systems can draft an email, but it cannot answer “what did we promise this customer” because the answer lives in your knowledge, not the internet⁵.
Data readiness is now the bottleneck - Gartner predicts organizations will abandon 60 percent of AI projects through 2026 that are not supported by AI-ready data⁴. The blocker moved from the model to the data.
Pilots stall for the same reason - Impressive demos fail to scale because the demo used clean sample data and the real company runs on scattered, messy context the assistant never sees¹².
Enterprises are re-centralising on purpose - After a decade of spreading workloads everywhere, organizations are pulling strategic AI work back toward where the data has gravity¹⁴.
The cost of moving data is rising - As AI constantly retrieves and synchronises across clouds and systems, the hidden expense is increasingly the movement of data, not the processing of it¹³.
Spend is scaling faster than value - AI-native application spend for large enterprises now averages millions a year, much of it scattered across subscriptions with no agreed definition of value²⁰.

“Gartner predicts that through 2026, organizations will abandon 60 percent of AI projects unsupported by AI-ready data.”

- Roxane Edjlali, Senior Director Analyst at Gartner⁴

The pattern is clear. The companies getting real value from AI are not the ones with the best model access - everyone has that now. They are the ones that gave their knowledge a centre of gravity so the AI can reach it.

The Company Brain: Giving Your Knowledge a Centre of Gravity

If data gravity is the force, a Company Brain is what you build to point it where you want. It is the consolidated layer that your knowledge orbits and that your AI employees act from - not another silo, but the place the silos feed into.

Learns your company, not the internet - The Company Brain holds your processes, rules, decisions, and institutional context, so AI acts from what is true in your business rather than generic training data.
Connects to your real systems - It links to email, Teams, SharePoint, CRM, and ERP, consolidating the knowledge that matters without forcing anything out of the systems that own it.
One source of truth - Instead of six versions of a fact, there is one place the answer is reasoned from, so decisions stop being made on stale copies.
Improves with use - It gets sharper as your team interacts with it, because every real task adds context, the same way gravity strengthens as mass accumulates.
Sits on top, not instead of - No rip-and-replace. Your existing tools stay exactly where they are and keep doing their job.
Model-agnostic foundation - The knowledge lives with the company, not inside one vendor’s assistant, so you are never locked into a single model to keep using your own context.

The Core Idea

A Company Brain does not try to move all your data into one database. It gives your knowledge a centre of gravity - one consolidated layer that connects to where the data already lives, so both your people and your AI employees can finally act from complete context. Consolidate the knowledge, not the systems.

Company Brain versus the usual alternatives

Approach	What it does	Why it falls short
Wiki or knowledge base	People write things down manually	Goes stale because keeping it current is nobody’s job
Point-to-point integrations	Pipes data between two tools	Moves data around but creates no centre of gravity
General AI assistant	Answers from public knowledge and a chat window	Cannot reach your systems, so it cannot do your work
Data warehouse project	Moves structured data into one store	Ignores the 80 percent that is unstructured, takes years
Company Brain	Consolidates knowledge and connects to live systems	The centre of gravity AI employees actually act from

Give your knowledge a centre of gravity

Book a 30-minute call. We will map where your knowledge is scattered and where the leverage is hiding.

Book a Demo →

Scattered fragments consolidating into one solid block, representing knowledge gaining a centre of gravity

AI Employees That Act From the Centre

A Company Brain on its own is a foundation. What turns the foundation into results is the layer that acts on it: AI employees connected to your real systems, doing routine work end to end rather than waiting to be asked.

They reason from the Brain - Before acting, an AI employee draws on the consolidated context - the customer history, the process rules, the prior decisions - so its output fits your business.
They act through your systems - Connected to CRM, ERP, and your document stores, they create the order, post the invoice, update the record, and send the reply, not just draft text for a human to paste.
They handle the full process - Not a single step, but the whole routine workflow from trigger to close, with human-in-the-loop checkpoints where judgement matters.
They stay in one place - Because they work from the Company Brain, they do not need a person to gather context from six systems first. The context is already there.
They compound the gravity - Every task an AI employee completes adds to the Brain, so the centre of gravity gets stronger and the next task gets easier.

Capability	General AI Assistant	AI Employee on a Company Brain
Knows your company	No, answers from public data	Yes, reasons from consolidated context
Reaches your systems	Limited or none	Connected to CRM, ERP, docs, comms
Takes real action	Drafts text for a human	Completes the process end to end
Waits to be asked	Yes, one prompt at a time	No, takes work from the queue
Gets better over time	Resets each session	Compounds as the Brain grows

This is the difference between a chat window and a colleague. The chat window helps a person work faster; the AI employee owns the work, because it has a centre of gravity to act from.

Knowledge That Survives Turnover

The most expensive form of scattered knowledge is the kind that lives only in a person. When they leave, it leaves with them, and the company pays to rediscover what it already knew. A centre of gravity changes the economics of turnover.

The why does not walk out - Exceptions, workarounds, and the reasoning behind decisions get captured in the Brain through daily work rather than living only in someone’s memory.
Onboarding starts warm - A new hire, or a new AI employee, inherits the consolidated context instead of starting from a cold start and re-asking questions that were answered years ago.
Vacation and sick cover stop breaking - Work does not stall because the one person who knows is unavailable, since the knowledge is reachable from the centre.
Reorgs do not reset the clock - When teams move and roles change, the institutional knowledge stays put because it has gravity of its own.
The company owns the leverage - The value accrues to the business, not to individuals who can take it with them or hold it hostage.

Knowledge trapped in individuals

✗ Leaves on the last day - resignation resets the process
✗ Single points of failure - one holiday stalls the work
✗ Slow onboarding - new hires start from zero context
✗ Invisible to AI - nothing in a head is reachable by a system

Knowledge in a Company Brain

✓ Stays with the company - turnover no longer means loss
✓ Always available - no single person is a bottleneck
✓ Warm onboarding - people and agents inherit context
✓ Usable by AI - the centre of gravity is machine-reachable

More Output Without More Headcount

The reason data gravity is a business argument and not an IT hobby is what it does to capacity. When knowledge has a centre and AI employees act from it, output stops being tied one-to-one to headcount.

Routine work leaves the team - The repetitive, context-heavy tasks that ate hours move to AI employees, so your people spend their time on the work only people can do.
Capacity grows without hiring - You add throughput by putting another AI employee on another process, not by opening another req in a market where the talent is scarce.
The search tax disappears - The 1.8 to 2.5 hours a day lost to hunting for information collapse when the context is reachable from one place⁷.
Onboarding cost drops - Every new process inherits the existing centre of gravity instead of rebuilding context, so the second and third use cases are faster than the first.
Spend maps to outcomes - Instead of scattering budget across per-seat subscriptions, you pay for completed work with a defined value, which is the opposite of the scattered AI spend most enterprises struggle to justify²⁰.

Why It Compounds

Hiring adds capacity linearly - one person, one seat, one salary at a time. A centre of gravity adds capacity non-linearly, because every process you move onto the Company Brain makes the next one cheaper to move. The leverage is in the consolidation, and it strengthens with every task, exactly like the physics the term is named after.

How to Build Data Gravity Into Your Company

You do not need a two-year data strategy to start. You need one process where scattered knowledge costs you the most, and a centre of gravity built around it. Here is the practical sequence.

Name the costliest fragmentation - Find the process where the same question gets re-asked, the handoff always drops, or the work waits on one person. That is where the gravity is weakest and the payoff is clearest.
Map where the knowledge lives - For that one process, list every place the context actually sits: which inboxes, which systems, whose head. Do not try to inventory the whole company.
Consolidate, do not migrate - Build a Company Brain that connects to those sources and consolidates the knowledge that matters, leaving the systems of record exactly where they are.
Put an AI employee on the process - Deploy an agent that reasons from the Brain and acts through your systems, with human checkpoints where judgement is required.
Measure against the baseline - Compare time, errors, and throughput before and after. Prove the value on one process before expanding.
Expand the same centre of gravity - Move the next process onto the same Brain. Because the foundation exists, each addition is faster and the pull gets stronger.

Data Gravity Readiness Checklist

You can name the one process where scattered knowledge hurts most
That process spans at least two or three different systems
Key context for it lives in email, chat, or someone’s head today
Your systems have API access or export capabilities
A process owner is willing to champion the first use case
Leadership will judge success on outcomes, not activity
You are willing to start with one process, not the whole company
You want the knowledge to belong to the company, not a vendor

Company-wide data project first

✗ Years to value - the payoff is always over the horizon
✗ High failure rate - big-bang data programmes routinely stall
✗ No early proof - nothing to show leadership for months
✗ Scope creep - every team wants their edge case in

One process, one Company Brain

✓ Value in weeks - first AI employee live in about two weeks
✓ Proof before scale - measured against a real baseline
✓ Compounding foundation - the next process reuses the Brain
✓ Contained risk - start small, expand what works

How Superkind Fits

Superkind builds custom AI employees for SMEs and enterprises, on a Company Brain that gives your knowledge a centre of gravity. The approach is process-first, not technology-first - it starts with your workflows and the systems you already run, not a generic product you have to bend to.

Company Brain as the foundation - We build a consolidated layer that learns your company, not the internet, and becomes the place your AI employees reason from.
Connected to your real systems - AI employees plug into email, Teams, SharePoint, CRM like HubSpot or Salesforce, and ERP like SAP or Lexware, so they act, not just chat.
No island solution - The Brain sits on top of your stack. No rip-and-replace, no tool chaos, nothing new for your team to learn.
Live in about two weeks - The first AI employee goes to work on a single high-value process fast, because there is no long data-migration phase.
Knowledge that stays - Context is captured in the Brain through daily work, so it survives turnover instead of walking out the door.
Output without headcount - Your team grows in performance without new hires, because routine work moves to AI employees connected to the centre.
Model-agnostic - Your knowledge lives with you, not inside one vendor’s assistant, so you are never locked to a single model to keep using your own context.
Enterprise-grade control - The Brain works within your infrastructure and permissions, with access controls and audit logs, so consolidation improves governance rather than weakening it.

Pros

✓ Company Brain foundation - a real centre of gravity, not another silo
✓ Acts through your systems - completes work end to end
✓ Fast time-to-value - first employee live in about two weeks
✓ Knowledge survives turnover - context stays with the company
✓ No vendor lock-in - model-agnostic by design

Cons

✗ Not a self-serve tool - it is a build, done with our team
✗ Needs process access - we map how your work really runs
✗ Capacity-limited - we take a focused number of clients
✗ Overkill for one-off tasks - a simple macro does not need a Brain

Decision Framework: Is Your Knowledge Ready to Have Gravity?

Not every company needs to act on this today. Here is a simple way to read the signals.

Signal	What it means	Action
The same question gets re-asked across teams	Knowledge exists but has no reachable centre	Consolidate that context into a Company Brain first
An AI pilot impressed but did not scale	The model was fine; the context was not reachable	Re-scope around data gravity, not a bigger model
Key processes stall when one person is out	Knowledge lives in heads, not in the company	Capture it in a Brain through the daily work
You are adding tools faster than you retire them	Fragmentation is getting worse every year	Build a centre of gravity before sprawl compounds
You need more output but can not hire	Capacity is capped by headcount	Move routine work onto AI employees on a Brain
You have under 20 people and simple, shared context	Your knowledge already has natural gravity	Start with lighter tools; revisit as you grow

If most of these signals sound like your company, the constraint is not your ambition or your model access. It is that your knowledge has no centre of gravity yet - and that is a solvable problem.

Frequently Asked Questions

Data gravity is a concept coined by engineer Dave McCrory in 2010. It describes how large accumulations of data attract applications, services, and even more data, the same way a large mass attracts smaller objects. The bigger and more useful a data store becomes, the more work naturally moves toward it because moving the data itself gets harder and slower. Applied to a company, it means your knowledge should have a centre that everything else orbits, not be smeared across a dozen disconnected systems.

AI is only as good as the context it can reach. When company knowledge is scattered across email, Teams, SharePoint, CRM, and ERP, an AI system has to hop between silos, and most of the useful context never reaches it. Data gravity in the AI era means putting the knowledge in one consolidated place - a Company Brain - so AI can act from complete context instead of fragments. Gartner predicts that through 2026, organizations will abandon 60 percent of AI projects that are not supported by AI-ready data.

A data silo is a system that holds data no other system can easily reach. Data gravity is the force that explains why silos form and why they are so hard to break: once data lives somewhere, the tools and workflows around it cling to that location. Silos are the symptom, data gravity is the physics. You cannot wish silos away, but you can choose where the gravity should be strongest by consolidating knowledge into one place your AI and your people both work from.

Point-to-point integrations help, but they do not create a centre of gravity. Every new tool multiplies the number of connections, and the knowledge still lives in a dozen places with a dozen owners and a dozen versions of the truth. Integrations move data around; a Company Brain gives the data a home. The goal is not more pipes between silos, it is one consolidated layer that your AI employees act from and your systems feed into.

No. A Company Brain sits on top of your existing stack. Your CRM, ERP, document stores, and communication tools stay exactly where they are. The Company Brain connects to them, consolidates the knowledge and context that matters, and becomes the place AI employees reason from and act through. There is no rip-and-replace, no migration project, and nothing new for your team to learn.

The numbers are stark. McKinsey research found that employees spend on average 1.8 hours every day, roughly 9.3 hours a week, searching and gathering information. IDC data puts the figure even higher, at about 2.5 hours a day or 30 percent of the workday. That time is lost not because the knowledge does not exist, but because it is fragmented across systems nobody can search in one place.

In most companies, a large share of what an experienced employee knows lives in their inbox, their private chats, their personal notes, and their head. When they leave, that context walks out with them, and the replacement starts from a cold start. When knowledge has gravity - when it is captured in a consolidated Company Brain that AI employees use daily - it stays with the company. Turnover stops resetting the clock.

A wiki is a place people are supposed to write things down and then read later. Most go stale because keeping them current is nobody's job. A Company Brain is different because it is actively used by AI employees to do real work, connects live to your systems rather than relying on manual entry, and improves as your team interacts with it. It is not a library your people occasionally visit; it is the working memory your AI operates from every day.

Most enterprise data is unstructured - Gartner puts the figure around 80 percent, locked in emails, documents, tickets, and transcripts. That is exactly the kind of context modern AI is built to read. You do not need a perfect data warehouse before you start. A Company Brain is designed to consolidate messy, real-world knowledge, and you begin with the processes where the value is clearest rather than boiling the ocean first.

Superkind typically has the first AI employee live in about two weeks, working on a single high-value process rather than the whole company at once. Because the Company Brain connects to systems you already run, there is no long data-migration phase. You see value on the first process, then expand the same consolidated layer to the next one, so the gravity compounds instead of being rebuilt each time.

Consolidation done properly improves control, it does not weaken it. A Company Brain works within your existing infrastructure and permissions, with access controls and audit logs, so you can see exactly what was used and by whom. Compare that to the real status quo: sensitive knowledge scattered across personal inboxes, private chats, and unmanaged files. Fragmentation is the bigger compliance risk, not centralisation under proper governance.

Start by naming the one process where scattered knowledge costs you the most - the handoff that always drops, the question that always gets re-asked, the work that always waits on one person. Map where the knowledge for that process actually lives today. Then consolidate that context into a Company Brain and put an AI employee on the process. One centre of gravity, proven on one process, is worth more than a company-wide data strategy that never ships.

Sources

Henri Jung

Co-founder of Superkind, where he helps SMEs and enterprises deploy custom AI agents that actually fit how their teams work. Henri is passionate about closing the gap between what AI can do and the value it creates in real companies. Before Superkind, he spent years working with mid-sized businesses on digital transformation and saw first-hand how many AI projects fail because the knowledge the AI needs is scattered across a dozen systems. He believes the Mittelstand has everything it needs to lead in AI - it just needs to give its knowledge a centre of gravity.

Ready to give your knowledge a centre of gravity?

Book a 30-minute call with Henri. We will find the process where scattered knowledge costs you most and outline how a Company Brain fixes it - no commitment, no sales pitch.