Questions business owners ask.
Nothing maps to that yet. Ask on the right and we'll email you an answer.
Potentially, yes. Many AI companies and specialized data buyers license historical operational data from businesses to help train AI systems.
We generally use the word "license" rather than "sell" because, depending on the agreement, you may continue to own the underlying data while giving a buyer specific rights to use a copy of it.
Exactly what rights you're granting, how long they last, whether they're exclusive, and how the data can be used should all be spelled out in the agreement with the data buyer.
And exploring the opportunity doesn't mean agreeing to anything. You can find out whether buyers are interested, understand what your data might be worth, ask questions, and then decide whether you want to move forward.
AI companies and specialized data buyers are looking for real-world examples of how people actually get work done.
They're not necessarily interested in your company because of what you sell. They're interested in the history your team created while running it: conversations, decisions, projects, documents, workflows, problems solved and outcomes.
That kind of operational history can help AI systems learn how work happens in the real world.
Goldcache isn't the data buyer. We're Data Value Advisors on the business side of the table. We help businesses understand what they have, navigate potential buyers and figure out which opportunities may put the most value on what they built.
The first step is simply figuring out what you have.
You give us some basic information about your organization: how long you've operated, how many people have worked there and what systems they've used. Estimates are fine.
From there, Goldcache can help determine whether potential data buyers may be interested and what the opportunity might be worth.
If there's interest, you can talk directly with potential buyers, understand what data they want, ask questions about privacy and de-identification, review the proposed terms and decide whether you want to proceed.
The important part: exploring is not agreeing.
Filling out the form, finding out there's interest, talking with a buyer or receiving an offer doesn't mean you have to license anything.
The process is:
→ Tell us what you have
→ See if there's buyer interest
→ Find out what it might be worth
→ Talk with potential buyers
→ Ask questions and review the terms
→ Decide
You can always say no.
No. Here's why.
How AI training works
Most people picture their data going into something like ChatGPT, where anyone can look it up later. Training doesn't work that way. It's more like an apprenticeship. Picture a new hire who shadows a thousand different offices. They won't remember anyone's client list or pricing. What they learn is how work gets done: how a request becomes a plan, how a task gets handed off, how a team fixes a mistake.
Why it can't copy your company
Your records are combined with records from thousands of other businesses, so the AI learns what they have in common, not what makes yours unique. Everything is anonymized first, with names, clients and your company name removed. A competitor can't ask it "how does Acme price its jobs?" and get your playbook. That isn't what it learns, and nothing in it points back to you.
What buyers actually want
Today's AI is good at polished, public work like articles, code and textbook answers. It struggles with the messy, human middle of real work: the email thread that sorts out a problem, the Slack chat where a team changes course, the follow-up that saves a deal. The more AI gets wrong at your team's kind of work, the more valuable your data is.
So buyers aren't after your secret sauce, your client deliverables or your IP. They want the everyday back-and-forth you'd never think twice about. To you, it's just Tuesday. To an AI learning how people work, it's the most valuable part.
Everything is anonymized before it changes hands. Names, emails, phone numbers, job titles, company names, passwords and payment details are replaced with placeholder tags, so nobody can trace a record back to a person, a client or your company. You also decide what's included. You can leave out any tool, any type of content, or anything covered by an NDA or a client contract. And the licensing terms are written down and agreed before anything is shared. If you have questions about your rights to certain data, we'll flag them early so you can check with your own attorney.
Much less than you'd think. It won't tie up your team for weeks. The longest part is usually telling us roughly how much data you have in each tool, and we walk you through exactly where to find that, or do it with you on a call. Buyers have connectors for the major tools, so nobody on your team has to sort, export or clean anything.
The goal isn't to package up your organization and hand it to a competitor.
The buyers are typically AI companies using the data to train models to understand how work happens. Depending on the buyer and the deal, identifying information may also be removed before the data reaches the ultimate AI company.
This is absolutely something to ask about before signing anything: Who can receive the data? What can they use it for? Will they know which organization it came from?
Get those answers directly from the potential data buyer and make sure you're comfortable with them before moving forward.
Not by exploring the opportunity.
Filling out the Data Footprint form, finding out whether there's buyer interest, receiving an offer, or talking with a potential data buyer doesn't give anyone rights to your data.
If you decide to move forward, the agreement with the data buyer spells out exactly what you're licensing and what rights you're granting.
Read that part carefully. Ask questions. And if you're not comfortable, don't sign.
Completely fair concern — and one you should ask every potential data buyer about.
AI companies generally aren't interested in the fact that Jane Smith works at Acme Corp. They're interested in the work happening around her: how something gets requested, discussed, worked on, reviewed and completed.
Different data buyers handle anonymization differently, so ask them to explain exactly what gets removed, when it gets removed, and what the ultimate AI buyer sees.
You can understand all of that before deciding whether you want to move forward.
It depends on the data buyer, but this is something you'll understand before you agree to anything.
Typically, identifying information — names, email addresses, phone numbers, customer names and other personal details — is removed or replaced while preserving what makes the data useful: how people communicated, made decisions, solved problems and got work done.
Different buyers have different processes, and many can walk you through real examples of what data looks like before and after it's de-identified so you can see exactly what that means.
Ask questions. Look at examples. Understand who sees what and when. And remember: exploring what your data might be worth doesn't mean handing it over. You get to understand the process and review the terms before deciding whether to move forward.
That's exactly the kind of question you should ask before doing a deal.
There are real things to understand: privacy, de-identification, customer and employee information, intellectual property, client-owned work, security, who ultimately receives the data and exactly what rights you're granting.
The important thing is that you don't have to resolve all of those questions just to explore the opportunity.
First, find out whether there's buyer interest and what your data might be worth. If there's a real opportunity, potential buyers can walk you through their process, show you examples of how data is de-identified and answer your questions before you sign anything.
Goldcache approaches this from the business side, not the buyer side. We want you to understand both the potential value and the terms before deciding whether a deal makes sense.
Start with your contracts.
If they clearly say the client owns something and you don't have the right to license it, don't include it.
But don't assume that means everything is off limits.
Your client might own the final deliverable while your organization still owns internal conversations, drafts, research, project management, workflows and other work that led to it.
For AI training, sometimes the process is more interesting than the final product anyway.
If you're unsure what you own, flag it when talking with a potential data buyer and have your attorney review anything that isn't clear before signing.
Flag it immediately.
Highly sensitive or regulated information requires additional scrutiny and may need to be excluded entirely.
Don't assume anonymization automatically makes everything okay.
Tell potential data buyers what kinds of sensitive information exist in your systems and ask how they would handle it before deciding whether to proceed.
Maybe. That's exactly what you're trying to find out.
There isn't a simple "$X per gigabyte" formula. Buyers care about things like how long your organization has operated, how many people have worked there, what systems you've used, how much history exists across them, and what kinds of work your team has done.
The important part: don't decide for the buyer that your data isn't valuable.
Give your best estimates and let potential buyers tell you whether they're interested and what they might be willing to pay.
Goldcache can help you get a sense of whether potential buyers are interested before you decide whether to take the process any further.
Yes. Closing a company is hard, and it doesn't erase what you built. As long as the data still exists in your old systems, it has value. There are lots of buyers for it, and we know the ones that will value yours most.
Usually, the everyday records of how work actually got done.
That can include email, Slack or Teams conversations, documents, project-management systems, CRM activity, support tickets, code, designs, files and other operational history.
What makes this interesting isn't necessarily any one document or message. It's the connections between them: a lead comes in, people discuss it, work gets assigned, decisions get made, something gets created, reviewed and ultimately completed.
Generally, the more people who contributed, the more systems they used and the more years of history you have, the more interesting the dataset may be to potential buyers.
You don't need to decide whether your data is valuable. Give us your best estimates and let the market tell you.
Think of an RL gym as a simulator. Instead of teaching an AI by having it read text, an RL gym teaches it by letting it learn through trial and error, exactly like a human learns a video game. The AI tries an action, the simulator reacts, and the AI gets a digital "reward" for a good move or a penalty for a mistake. By repeating this loop millions of times, the AI becomes an expert at that specific task.
To build these simulators, AI developers need massive amounts of real-world operational data (this is where you come in). Without your data, their virtual training arenas wouldn't mirror reality, and the AI would fail in the real world.
No. Your best estimates are fine. We're interested in the history of the business, not just what it looks like today.
Maybe.
It depends on how your organization is structured and what your agreements and governing documents say.
But you don't necessarily need to solve that just to find out whether there's an opportunity.
See if there's buyer interest and what the data might be worth first. If there's a real offer worth considering, determine what approvals you need before signing an agreement with the data buyer.
You don't need to hire a lawyer just to fill out the Data Footprint form or find out whether potential buyers are interested.
If you receive an offer you actually want to consider, that's a great time to have your attorney review the agreement with the data buyer — particularly if you have complicated client contracts, investors, sensitive information or valuable IP.
You can find out whether there's a real opportunity before spending money on legal review.
Finding out whether there's interest? A few minutes.
The initial Data Footprint form is designed to give potential buyers enough information to determine whether your organization might be interesting.
Actually completing a licensing transaction takes longer because the data may need to be reviewed, exported, processed, anonymized and ultimately approved.
You don't need to do all of that upfront.
First find out if there's interest and what the opportunity might be worth. Then decide whether it's worth taking the next step.
You don't have to make the decision now.
Finding out what you have, whether there's buyer interest, or what your data might be worth doesn't mean you have to proceed with a transaction.
You can learn what the opportunity looks like, talk with potential buyers, and decide whether now, later, or never makes sense.
It's very hard to weigh the risk against the upside when you don't know the upside yet.
What we recommend is to find out what a potential buyer would offer you, while asking the questions you have about any perceived risks. Ask about privacy. Ask about anonymization. Ask who gets the data. Ask about exclusivity. Ask what rights you're granting. Ask anything that makes you uncomfortable. Then weigh the actual offer against the actual terms.
Yes. This isn't only for profitable companies. Any organization qualifies, including 501(c)(3) nonprofits and trade associations, if it had 20+ people at peak, has been around 2+ years and used two or three of the tools we look for.
Yes. We have a separate partner deal for VCs, accountants, CPAs and financial planners whose companies or clients could benefit. Visit Partner with us to learn more.
No. Filling out the intake is free and doesn't commit you to anything.