Newsphere: How to Make Your Company AI-Native

“Your model isn’t the bottleneck—accessing your tribal knowledge is.”

— McKinsey, The Seven Operating Truths of AI-Native Companies (2026)

Every AI tool we used was good at its slice of the work. Not one of them could see the whole company. One knew the project in front of it, one the document on the screen, one the inbox it was plugged into. All useful, and all working from a fraction of the picture: not where we were headed, not what we promised a client months ago, not how today’s deal connected to a project we delivered two years back. We were getting real help on every piece, and none on how the pieces fit together.

Right now everyone is talking about becoming an “AI-native company.” It is the phrase of the moment. And most of the time that is all it is, a phrase. It gets said in meetings and printed on websites far more often than it gets put into practice, and when companies do reach for it, they usually start in the wrong place, with a new tool or a new model, and then wonder why nothing really changes.

We wanted to actually do it, not just say it. What we built, we call Newsphere. The name is a play on an older idea, the noosphere. A century ago a few thinkers described the biosphere as the layer of living things wrapped around the planet, and then named a layer above it, the noosphere, from the Greek word for mind. It was their term for the sphere of human thought and knowledge, the shared layer of everything we collectively know. That is almost exactly what we set out to build for a single company. One sphere that holds everything the business knows, that our people and our AI both think inside of. Newsphere.

If you want a more familiar reference point, you already know two things close to it. A Center of Excellence is a single home for a company’s best practices and standards, so everyone works from the same playbook instead of reinventing it. An ISO standard is the discipline of writing down how you do things, so quality stays consistent and anyone can follow it. Newsphere is both of those ideas, your whole way of operating captured in one place and kept current, with one change that makes all the difference. It is built to be read by an AI as easily as by a person.

Here is how we got there, and how you could build your own.

So we made our whole company AI-readable

So we built the foundation first. We took everything the company knows and put it in one place, in a form an AI can read across the whole business at once, instead of one slice at a time.

That foundation is what we call the substrate, and it is less about format than about order. It does not mean a special database or a new app, and it does not mean retyping anything. It means your real work, your documents, PDFs, contracts, notes, and spreadsheets, kept in one organized, labeled home instead of scattered across inboxes, drives, and people’s heads.

Your projects, your clients, your pricing, your past work, the way you say things, all in one place where an AI can read across the whole of it at once. The AI sits on top of it. Almost everything good that follows comes from getting this part right.

Frontmatter: the label on every file

If the substrate is the foundation, frontmatter is what makes it navigable. It is the single most important detail in the whole system, and it is the part almost everyone gets wrong.

Frontmatter is a short label at the top of every file: a handful of structured fields that say what this file is, what is inside it, and how it connects to everything else. A contact’s file does not just hold notes about them. At the top it carries who they are, the company they work for, the deals they touch, the last time we spoke, and the next step we owe them. Every file points to other files by name, so an AI can follow the thread the way you would in your own head: from a person, to their company, to the open deal, to the project you delivered for them two years ago.

That is the difference between an AI that walks straight to the right room and one that wanders the halls opening every door. Without good labels, even a clean pile of files leaves the AI guessing, and it will confidently hand you the wrong thing. With them, it goes straight to what matters and brings back answers grounded in your real records.

Here is the honest part. Good frontmatter is harder than it looks. Too many fields and they rot the moment someone forgets to update one. Too few and the AI is lost. Inconsistent from one file to the next and nothing lines up. The formula for frontmatter that is minimal, consistent, and actually earns its place, plus the tooling that keeps it from drifting, is the part we have spent the most time refining. It is a large part of what separates a substrate that works from a folder of files that just sits there.

The habits that made it work

Keeping the substrate clean comes down to a few simple habits.

The first is one source of truth for every fact. A thing is written down in exactly one place, and everywhere else points to it instead of copying it. A client’s address, a price, a decision we made: each lives once. When you keep two copies, they drift, and then the AI has to guess which one is right. One copy means there is nothing to guess.

The second is a map, not an encyclopedia. Each corner of the company carries a short guide to what is there and where to look next, not a dump of everything it might ever need to know. The AI follows the map to the room it needs and reads deeply only there. This is what keeps it fast and accurate as the company grows, instead of slower and more confused with every file you add.

The third is to let the system rebuild what it can. Our pipeline forecast, our directory of who is who, the index of where everything lives: none of it is typed up by hand. It is generated from the underlying files, so it is always current and never quietly wrong.

And we built the tooling to keep those habits true without leaning on anyone’s memory. It handles the routine upkeep on its own, and when something genuinely needs a person, a judgment call, or a fact it cannot verify, it surfaces the question to a founder instead of guessing or letting the data quietly go stale.

We manage it like a team

We do not think of the AI as a tool we operate. We think of it as something we manage, the way you manage a capable team.

A good manager does not do every task by hand, and does not rubber-stamp everything either. They let the team do the work, review the decisions that actually matter, and set the direction. That is our job here. The AI drafts the proposal, prepares the follow-up, pulls the answer together. We review the calls that count and decide where things go. Nothing important leaves the building or gets committed without one of us signing off, and the corrections we give it become how it gets better next time, the same as coaching anyone else.

Your company already has the pieces

Here is the part worth sitting with. You almost certainly already have everything you need to start this. You are not missing a tool.

You already own the storage. Your file cabinet might be SharePoint, Box, Dropbox, or a shared drive, but you have one, and you are already paying for it. And you already have the knowledge. It is just scattered, spread across those drives, buried in inboxes, locked in the heads of the people who have been there longest. The raw material of an AI-native company is sitting in your business right now. What is missing is not the cabinet. It is the order inside it.

That order is the work. The labeling, the single source of truth for each fact, the structure that turns a pile of files into something an AI can read across. It is not glamorous, and it is not a weekend project. It is the part that takes real intent.

And here is the honest catch, and how we handle it. Order is not a one-time cleanup. The substrate only keeps working if it stays current as the business changes, and stale data is the one thing that breaks the whole system. An AI reading last quarter’s pricing or a deal that closed months ago will hand you a confident, wrong answer, and confident and wrong is worse than nothing. McKinsey found the same in its study of AI-native companies:

“When data gets old and stale, agents will confidently serve up outdated answers, which rapidly erodes user trust.”

— McKinsey (2026)

This is exactly why keeping the substrate current is part of what Newsphere automates. The system uses agents to keep itself up to date, capturing new work as it happens, refreshing what it can on its own, and flagging anything that looks out of date for a quick human check instead of letting it quietly rot. Staying fresh becomes mostly the system’s job, not a standing chore for your team. Get that part right, and everything changes.

Our process

This is what we do at BPN. We guide companies through the full process of building their own Newsphere, and we run it in a deliberate order.

We begin with capture. Before anything can be organized, the data has to exist in one place, and in most companies a great deal of it is never written down at all. So we make capture automatic. One example is an AI note-taker connected directly to the substrate, so every meeting and call is recorded and filed without anyone having to remember to do it. The knowledge that used to live only in someone’s memory stops slipping away.

From there we organize. We give the substrate a real structure, applying the folder layout and frontmatter standards we have battle-tested running our own business. This is the work that makes a substrate navigable, the difference between a full archive an AI cannot move through and one it can read across in seconds. It is the part most companies underestimate, and the part where our experience matters most.

With capture and structure in place, the foundation is ready. It is what makes the deeper automations possible, the ones that change how a company runs day to day. Based on what we have seen so far, we estimate it has improved our productivity by about 175 percent, and we are still collecting data and refining the process every day. It keeps us moving toward our goals without letting the small things slip through the cracks.

Where we go the deepest

This is where the foundation pays off. Once your substrate is in place and stays current, the AI on top of it can take on real work, the kind that used to need a person for every step. These are the automations we run on our own business and build for the companies we work with.

Some of it runs quietly in the background, keeping the company current without anyone touching it.

• Everyone stays in sync automatically. When one of us logs a call, moves a deal, or makes a decision, that change is waiting for everyone else the next time they open the company. Nobody has to broadcast what they did, and nobody is working off last week’s picture.
• The forecast builds itself. Our view of the pipeline is not a spreadsheet anyone keeps up to date. It is generated straight from the underlying deal records every time we look at it, so the moment a deal moves, the forecast already reflects it. No monthly scramble, and no version that is quietly wrong.
• Meetings file themselves. An AI note-taker writes the summary, but the part that matters is what happens next. The notes flow into the right client and contact records and update them, capturing the next step and the new detail without anyone typing it up.

Other parts work right next to you, doing the first ninety percent of a task and handing you the last ten.

• Documents draft themselves from your real history. A proposal, a statement of work, or a report, built from your own past projects, your real pricing, and the language you actually use, instead of a blank page. A document that used to take hours becomes a review that takes minutes.
• Follow-ups arrive written in your voice. After a meeting, a ready-to-send email is waiting in the sender’s own tone, built from what was actually discussed and the real next steps. A person reads it, adjusts what needs adjusting, and sends.
• You can ask the company anything. Because the whole business is readable in one place, you can question it the way you would a long-tenured employee who remembers everything. “What did we promise this client in March, and who owns it?” comes back in seconds, pulled from your real records rather than someone’s best guess.
• A recap lands on your phone at the end of the day. What moved, what shipped, and what needs you tomorrow. You did not open a dashboard or chase anyone for it. The company tells you where it stands on its own.
• Nothing goes cold quietly. Deals rarely die on purpose, they fade out of view. The system surfaces the relationships that have gone quiet and the deals waiting on a next step, and tells you exactly what each one needs.
• A new hire is useful on day one. Someone new does not spend three weeks shadowing people. They open the company and it explains itself: how we do things, where everything lives, the context behind the current work. People get productive in hours instead of weeks.

And the deepest layer connects the system to the specific way your business runs.

• Your repeatable jobs become one-command tools. Every business has work it does the same way every time: client intake, building a quote, running a quality check. We turn those into skills the AI performs on request to the same standard every time, so routine work no longer depends on the right person remembering the right steps.
• Custom integrations go deeper than the stock connectors. Most AI tools ship with a fixed set of built-in connections, and even when one exists for the app you use, it is often too shallow to rely on. Email was a clear example. The standard Gmail connector cannot reply inside an existing thread, so every message the AI sends starts a brand new conversation, and it handles attachments poorly, often returning a link instead of the actual file. So we built our own Gmail integration on an emerging standard called the Model Context Protocol, one that replies in the right thread and pulls attachments down as real, usable files. The same approach lets the AI reach the industry and in-house tools no off-the-shelf connector will ever support.
• Guardrails keep the sensitive things in. Opening your company up to AI should never mean opening it up to a leak. Automatic checks make sure private information stays where it belongs, and access follows the same boundaries your business already has, so a person or a system only ever sees what it is supposed to.

None of this is off-the-shelf. Every automation is shaped to how a particular business works, and each one is only as strong as the substrate beneath it. That is why we start with the foundation and not the features. So the honest question is whether all of this is worth the effort it takes. We think it is, and here is why.

Why it’s worth it

The work is real, and so is the payoff.

The first thing you notice is that you stop losing things. The detail from a call six months ago, the reason you priced a job the way you did, the promise someone made to a client, all of it is there, not trapped in one person’s memory or buried in a thread no one can find. The company remembers, so you do not have to.

From there, a small team starts to punch well above its size. The work that used to eat a week happens in an afternoon, because the AI does the gathering and the drafting and you do the deciding. Your decisions get better too, because they come from what actually happened across the business, not from guesswork or whoever spoke up last.

And it compounds. Every project, every call, every decision folds back into the foundation, so the company gets sharper the longer it runs instead of slower and more tangled. You are also not betting the business on a single tool or model. The models keep improving on their own, and your foundation stays yours, ready to sit under whatever comes next.

The quiet payoff is the one that is hardest to describe until you have it. The company feels calm. You spend your time on the work that matters instead of digging for information, chasing updates, and rebuilding what someone already figured out once.

There is a cost to waiting, too. As McKinsey put it in its study of AI-native companies:

“The companies that delay are not only standing still; they are ceding ground to competitors who are compounding their advantages with every cycle.”

— McKinsey (2026)

None of this started with a smarter model. It started with a foundation, the same one your business is already sitting on, waiting to be put in order. That foundation, and the intelligence and tooling we build on top of it, is what we call Newsphere.

Foundation first. Intelligence on top. A human keeping it sharp.

That is what being AI-native actually means, and Newsphere is how we get a company there, within reach for any business willing to build it.

Quotations from McKinsey & Company, “The Seven Operating Truths of AI-Native Companies,” June 2026.