FAQs

Known Entity.ai builds AI Authority infrastructure for suppliers to the food and drink industry.

Your buyers have stopped searching and started asking. When a technical, NPD or procurement manager at a food or drink manufacturer asks AI who can supply something, what comes back is a shortlist of about five names. We make sure you are on it, and that the facts about you are right.

The way we do that is the EntityCore™: a machine-readable single source of public truth, hosted on your own subdomain, that every major AI platform can find, trust and cite. Known Entity.ai is based in Folkestone, Kent, and is a trading brand of WorkSpot Ltd.

In AI and knowledge systems, a known entity is an organisation the system holds enough verified information about to name with confidence. That is the status we build. The name is the outcome.

Properly Known is our brand essence. It means being cited for the right things, in the right context, to the right buyers.

Appearing is not the goal. A business can appear in an AI answer with an expired certificate, a discontinued product or a capability it no longer has, and lose the deal faster than if it had not appeared at all. Properly known means the facts a buyer agent finds about you are current, complete and checkable.

Known Entity.ai was co-founded by Helenor Rogers, Chief Executive Officer, and Mike Rogers, Chief Technology Officer.

Helenor brings more than twenty years of strategic marketing experience across global brands and B2B sectors, and leads client delivery and the Authority and Evidence Layer.

Mike previously co-founded Troo Foods. He has sat on the buying side of this industry, specifying ingredients and packaging and qualifying suppliers, which is why the audit is built around the questions a technical manager actually asks rather than around marketing keywords. He leads the technical build and the agent pipeline.

Specialist content, technical and PR people are brought in per project.

Suppliers to the food and drink industry. Ingredients and raw materials, packaging, materials, and the equipment and services that sit alongside them. If your customer is a food or drink manufacturer, a brand owner or a private label producer, you are who we are built for. We also work with food and drink producers and brands themselves, where how AI describes them affects commercial outcomes.

We chose this sector because it is the one where an AI shortlist is decided almost entirely on facts a machine can check. When an NPD manager asks which suppliers can provide a given material with BRCGS certification, held in UK stock, in a specific pack size and quantity, the AI is not weighing reputations. It is running a qualification filter: is the product listed, is the certification held and in what scope, what is the specification, where is the stock, what is the minimum order. Suppliers whose answers are findable and checkable get shortlisted. Everyone else does not appear.

That is also why the work compounds here. Once a recipe or specification has been validated on a competitor's material, switching means re-trials, shelf-life work and relabelling. The supplier who was on the list when the question was asked usually keeps the business for the life of the product.

We do not work on consumer search, we do not take on repositioning or brand strategy work, and we do not work with early-stage technology companies.

Three things.

We build infrastructure, not recommendations. Everyone else works on the website you already have. We build a separate, purpose-built layer designed for AI systems, on a subdomain you control.

We address the whole problem. Around 90% of what AI systems say about your business comes from outside your website. Optimising the website alone leaves the other 90% untouched.

We are already on the next layer. Discoverable and trusted are today. Transactable is where the buyer journey is going.

We do not run paid advertising, manage social media or work on search rankings.

The buying journey has already shifted. A technical or procurement manager asks AI once and works from the answer.

Among buyers at companies with more than 3,000 employees, 42% already depend on AI to find vendors, against 18% at mid-sized firms (Responsive, survey of 350 B2B buyers, October 2025). In food and drink, the companies with more than 3,000 employees are the major manufacturers and the retail brand owners. The accounts you most want are the ones most likely to be asking.

And it is not only new business. Your existing customers run the same questions when they dual-source or re-tender, and increasingly they do it through a tool rather than a phone call.

The deals that never reach you. Buyers filter before they make contact, so an absent or wrongly described supplier loses the brief without ever seeing it. Your sales team never hears the question, so nothing appears in your reporting.

The defensive case is often the larger one. Existing accounts re-tender. If a buyer's AI says your ISO certificate expired in January, a six-figure account is at risk and nobody in your business will ever hear about it. We have found exactly that on a company we audited.

Two things are happening, at different speeds, and they are worth separating.

Delegated discovery is happening now. A buyer asks AI once, takes the answer and does not search. This is where shortlists are being decided today, and it runs on the open web: the AI reads directories, registries, trade body lists, certification registers, trade press and your own published record.

Autonomous agent buying is arriving, but not on the open web first. Every supplier-discovery agent currently shipping inside the major procurement platforms and marketplaces searches a closed supplier network rather than the open internet. Gartner's construct is that 90% of B2B buying will be AI agent intermediated by 2028. Intermediated is the word that matters: an agent in the loop, not an agent buying unattended.

So the work that pays today is being found and qualified on delegated questions. The work that pays later is being present in the data those networks run on. The same record serves both, which is why there is no second build.

The record you build to be cited in 2026 is the record that qualifies you in 2028. The same facts, the same evidence, the same provenance, read by a different kind of consumer. Nothing gets rebuilt.

What changes is who is reading: a chat answer today, a procurement agent or a supplier data network later. Businesses that wait will be building the record while the questions are already being answered without them.

AI Authority is the degree to which AI platforms can find your business, describe it accurately, and follow the evidence behind what it claims. A business with AI Authority is named and recommended when a buyer asks. A business without it is absent, however well established it is in the real world.

It has two gates.

Retrieval. Are you present in the sources a platform actually pulls when it answers that question.

Qualification. Once you have been found, can the facts about you be checked.

Most established businesses fail the first gate and have no idea. In the one ingredients supplier we have audited so far, the business was recognised in 100% of branded questions and appeared in fewer than one in five unbranded buying questions.

Search optimises a website so it ranks in a list, and the buyer chooses where to click. AI platforms do not return a list. They assemble one answer from sources they have learned to trust, and the buyer usually sees nothing further.

The goal moves from appearing in results to being in the answer. And the test moves from persuading a reader to satisfying a check.

Yes. A well-structured, crawlable site with genuine domain authority feeds the signals AI systems use, and you should keep doing it.

But a business can rank on page one and be entirely absent from AI answers, because most of what AI says about you comes from sources you do not own. Keep your agency. This is not what they do.

Those are the industry's names for making content more likely to be mentioned in AI answers. It is real work and it helps with the first of the two gates.

It is not what we do. Mention is not shortlist. A buyer agent choosing a supplier is running a qualification check on facts: is the product listed, is the certification held and in what scope, what is the specification, where is the stock, what is the minimum order. Answering that needs a verified record behind the mention, not more content in front of it.

We measure five: ChatGPT, Perplexity, Claude, Gemini and Copilot.

Each draws on different sources and weights evidence differently, so a supplier can be shortlisted on two and absent from three. Every question in your panel is run against all five, on the consumer apps with web search enabled, which is what your buyers actually use.

An EntityCore™ built on confused positioning just makes the wrong business machine-readable. Before we make your business machine-readable, we make sure the machine is reading the right business.

The AI Authority Audit classifies what is actually wrong into four gaps: strategy, communication, authority and structure. Authority and structure gaps are ours to fix, at Stages 3 and 2 of the Ladder.

Where the audit finds a strategy gap we say so plainly, and we build on the factual layer that does not depend on it: products, certifications, customers, identifiers, locations. We do not take on repositioning work. When you resolve positioning with your own advisers, EntityWatch™ updates the record.

AI systems keep repeating what the web still says, long after you have moved on, because they are reading sources you no longer control. That makes the record and the correction work more valuable during a transition, not less.

What we do is publish the new facts in a form a machine can check, then find every third-party source still carrying the old ones and route corrections to them. What we do not do is the repositioning itself.

The machine-readable layer that makes a business discoverable, trusted and transactable.

Discoverable. AI systems can find your business when a buyer asks. The right entity exists, in the right registries, with the right facts, in a form a machine can read without guessing.

Trusted. AI systems represent your business accurately and can follow the evidence behind every claim. Independent verification, third-party authority and continuous correction hold the record true over time.

Transactable. Buyer agents can act on their buyer's behalf: interrogate your catalogue, check suitability and specifications, request samples and quotes, check availability, book meetings and eventually place orders.

Today's clients buy discoverable and trusted. Protect is what turns the build into a recurring relationship. Transactable is the destination.

Monitoring tools tell you what AI says about you. Content services try to make AI say more. Neither changes whether a machine can check the claim.

We build and maintain the thing that gets checked: an authored record of what is true about your business, with every claim tied to a source a machine can follow, kept current, with a log of what was found, what was wrong and what we did about it. Closer to a custodian than to an agency.

Because the record earns its keep on today's question before it earns anything on tomorrow's. It is what gets you shortlisted when a buyer asks AI this month.

Stage 5 is not the reason to start. It is the reason the work does not need doing twice.

The AI Authority Ladder™ is our five-stage methodology. Each stage answers one question.

1 · Discover. Can AI find you? Deliverable: the AI Authority Audit.

2 · Structure. Can AI understand you? Deliverable: the EntityCore™ build.

3 · Verify. Can AI prove what you claim? Deliverable: the Authority and Evidence Layer.

4 · Protect. Can that understanding stay accurate? Deliverable: EntityWatch™.

5 · Agent readiness. Can AI do business with you? Deliverable: agent-ready infrastructure.

The Ladder is how you buy and sequence the work. The EntityCore™ is what you own. Every stage after the second is work performed on the same record, which is why the record compounds.

We agree 30 to 50 real buying questions with you, the ones a technical, NPD or procurement manager at your customers would actually type. Not “best packaging supplier”, but the specific ones: this material, this certification and scope, this pack format, this quantity, this lead time, this country of stock. Then we run them across ChatGPT, Perplexity, Claude, Gemini and Copilot. Every answer is scored as shortlisted, omitted or misrepresented, with the competitors named and the sources cited.

We score nine dimensions: branded recognition, unbranded supplier discovery, accuracy, recommendations, competitor share, citation health, source quality, misinformation and gaps, and technical machine-readability.

It also includes a workshop with Helenor, the four-gap classification, and a verification inventory: every material claim you make, and whether a machine could check it today.

You get a documented baseline, a reason for every exclusion, and a scoped plan.

Two weeks. Included in the £5,500 setup fee covering Stages 1 to 3.

We ingest your website, product pages, PDFs, specifications and certificates, together with the authoritative external sources about you, and turn them into a canonical model: company, products, specifications, applications, markets, capabilities, certifications, locations, people, evidence. Our team resolves what a machine cannot determine confidently, and you sign it off.

It publishes at a subdomain of your own domain, entity.yourbrand.com, covering who you are, who works there, what you sell, how you do business and on what terms, who you are connected to, what buyers ask, who you are for, what you have done, where else you are recorded, and who has written about you. Every claim carries who asserted it, when, and from what source. Buyer agents query it directly through a dedicated endpoint.

The EntityCore™ is not a copy of your website and it is not a content site. It is a record. It sits on your domain, it is yours, and it is portable if you ever leave us.

No website rebuild, no CMS access, one DNS record. Four to six weeks. Included in the £5,500 setup fee.

The EntityCore™ says a claim is true. Stage 3 asks whether an AI system can confirm it independently, without taking your word for it.

We connect every material claim to a source a machine can follow: accreditation registers, certification bodies, Companies House, regulators, trade bodies, manufacturer and supplier sources, reputable trade publications, customer evidence and third-party references. A certificate is checked at the issuer, with its scope and validity dates, rather than reproduced as a logo on a page.

Then we map the gaps. For every buying question you are losing, we identify which fact needs to appear in which named source for you to be retrievable at all, because a record nobody points at does not get read.

Trade coverage and third-party citation sit inside this stage as authority building, not as content marketing. Included in the £5,500 setup fee.

Once we have established what is true, we continuously compare it against what the outside world is saying.

EntityWatch™ discovers changes, extracts claims and compares them with your EntityCore™. It identifies inconsistencies, monitors certification expiry, catches stale information, tracks new products, people and capabilities, flags potentially harmful misinformation, and watches for AI systems beginning to repeat an error.

The loop is correct, verify, update.

Your monthly report opens with your shortlist position, question by question, against named competitors. Then the corrective action log: what was found, the root cause, what we did, verified or pending. Then the verification ledger, the buyer agent log, two to four recommended earned-media targets to pursue, and next month's brief.

Once a buyer agent can find you, understand you and verify you, it should be able to act: interrogate your catalogue, check suitability and specifications, request samples and quotes, check availability, book a meeting and eventually place an order.

Your EntityCore™ already publishes an endpoint a buyer agent can query directly. Today most production procurement agents search closed supplier networks rather than the open web, so we treat this as a capability that is ready rather than a channel that is busy. It is included at no extra charge until enquiry volume justifies a line of its own.

The audit gives you a documented baseline in two weeks. The record goes live four to six weeks after that, and the first thing that usually moves is accuracy: wrong facts stop being repeated, because there is now a checkable source that disagrees with them.

Retrieval moves more slowly, because it depends on third-party sources being updated and then re-read. Expect three to six months for shortlist position to move on the questions that matter, and to keep moving after that as the evidence layer thickens.

We do not guarantee an outcome, and you should be wary of anyone who does. We guarantee the work: every exclusion traced to a reason, every reason inside our control fixed, everything re-measured next month.

Barely, and that is deliberate. The EntityCore™ is hosted at a subdomain, so we need one DNS record. No CMS access, no web development, no waiting for an agency backlog.

For the technical dimension of the audit we may want to look at your robots.txt and server configuration. We do not change anything. We tell you what to change and your agency executes it.

Parts of it. You could run prompts, and you could brief journalists.

What is hard to hold in-house is the discipline: a question panel that stays fixed so the numbers mean something, every claim traced to a source a machine can follow, certificates checked at the issuer, corrections filed and chased with third parties who have no obligation to reply, and the whole thing re-measured on the same basis every month.

That is a standing process, not a project. Most businesses have neither the time nor a reason to build one.

That is the common case, and correcting it is the headline of Stage 4 rather than an extra.

We trace the wrong fact back to the source carrying it, then fix it according to where control sits.

What you or we control, the record itself, registry entries, directory accounts, your own site, we correct directly or hand to you ready to submit.

Third parties that accept corrections, trade press, directories, business data providers, review sites, we file an evidenced request, chase it, and tell you the outcome either way.

What AI platforms themselves say is the third case. We fix the sources upstream, strengthen the record, use the platforms' own feedback routes, and re-run the question set next month.

What we will not do is promise that a model stops saying something. Nobody can promise that. What we do promise is that no wrong answer goes untraced.

Because around 90% of what an AI system says about your business comes from outside your website, and because a machine weights what it can check above what you assert.

There are two jobs here and they are routinely confused. Retrieval: your facts need to be present in the sources a platform actually pulls for that question, which in food and drink supply usually means certification registers such as the BRCGS Directory and IAF CertSearch, trade body membership lists, sector directories, the trade press and LinkedIn. Qualification: once you have been found, the claim has to hold up.

A trade article gets you retrieved. An issuer's certificate register gets you qualified. You need both, and they are not the same placement.

From your own audit, not from a media list.

Every question in your panel is run across the five platforms and every citation is captured, then collapsed to domains and scored on how many of your questions that source appears for, how many platforms it appears on, and how consistently it appears across repeated runs. We overlay where you and your competitors already appear, then classify each source by what it can carry, a verifiable fact or a mention, and by how attainable it actually is. A registry entry or your own directory profile is highly attainable. Trade press is medium. Some sources cannot be placed at all, and we say so rather than selling you the attempt.

You get a ranked list of the specific sources that decide your questions, refreshed monthly. It comes out of the same run that produces your shortlist score, so the map and the measure never disagree.

No. We are the brain, not the hands.

We tell your PR agency exactly which facts need to appear in which named sources, ranked by the buying questions they lose you today. That is a placement brief, not a topic list, and it is a far easier brief to act on than “write something about AI in packaging”.

If you have a PR agency, they keep the work, and we make it measurable: each month we show which placements were actually cited, by which platform, on which buyer question. If you do not have one, we can introduce you to freelance PR people we work with, and you contract them directly.

We do not take a margin on placements and we do not sell content. What we sell is the record, the evidence behind it, and keeping both true.

The same question panel, run the same way, every month.

The panel is agreed and locked at the start: the questions, their commercial weighting, the five platforms, the geography, the scoring rules and the number of runs. Nothing moves after that, because a measure you can change is not a measure.

Each month you get your appearance rate across the panel with its range, your position on the questions you told us are worth the most, an accuracy count of right, wrong and unverifiable, the competitors named against you, and the reason for every miss. A recovery has to hold for two consecutive months before we count it. We track your competitors on the same panel too, so a platform change that moved everybody is never reported to you as your gain.

A no-cost read of where you stand today. We run a sample of unbranded buying questions for your part of the food and drink supply base across the five platforms and send you a scorecard: how often you are named, how often you are recommended, how often the facts about you are right, and which competitors come up instead of you.

It is produced by the same pipeline that generates our sector index, so your score is comparable with every other supplier scored on the same panel. No obligation, and you do not need to sit through a meeting to get it.

A workshop with Helenor. Thirty to fifty agreed buying questions run across five platforms. Nine scored dimensions. The four-gap classification. A verification inventory of every material claim you make. A reason for every exclusion. And a scoped plan for what comes next.

Two weeks. Included in the £5,500 setup fee covering Stages 1 to 3.

Your question panel re-run across five platforms and scored, with the shortlist position on the front page.

The corrective action log: what was found, the root cause, what we did, verified or pending.

The verification ledger: every material claim, its source, when it was last checked, and anything approaching expiry.

The buyer agent log: who queried your record.

EntityCore™ maintenance, including two to four new knowledge graph articles, two to four recommended earned-media targets to pursue, a strategy review call and next month's brief.

From £1,500 a month on a twelve-month minimum. The monthly fee is not discounted.

Because without it you cannot tell an improvement from a platform update, and neither can we.

A fixed panel, run the same way, with your competitors measured alongside you, is the only honest way to know whether the work is doing anything. It is also what makes a correction accountable: something was wrong, here is what we did about it, here is the re-measurement.

AI Buyer Visibility Score. Free.

Setup, Stages 1 to 3. £5,500. A single fee covering the AI Authority Audit, the EntityCore™ build and the Authority and Evidence Layer. This is the total for all three, not a price per stage. Audit two weeks, build four to six weeks.

EntityWatch™, Stage 4. From £1,500 to £3,000 a month, depending on how many claims are under governance and how many segments are tracked. Twelve-month minimum.

Agent readiness, Stage 5. Included at no separate charge until enquiry volume justifies a line of its own.

Every stage is fixed scope and fixed price. There are no day rates. Year one typically £23,500 to £41,500 depending on scope.

Worth setting against it: one mid-sized account at £100,000 a year at 20% gross margin, retained five years, is £100,000 of gross profit. It is also less than most suppliers spend on a single trade show stand, and this is working every day of the year rather than three.

Because the record only has value if it is kept true, and keeping it true is a standing process rather than a task.

Corrections take a cycle to file and another to verify. Third-party sources take weeks to update and longer to be re-read by the platforms. We will not count a recovery until it has held for two consecutive months. Under twelve months you would be paying for the setup and leaving before any of the compounding starts.

The setup fee is 50% on agreement and the balance on delivery. EntityWatch™ is billed monthly. We can spread the setup fee across three months for larger engagements.

The monthly fee is not discounted.

As a referral relationship. If you have clients supplying the food and drink industry, you introduce us, you earn a fee, your client's price does not change, and we deliver under our own name.

That keeps it clean for everyone, and it keeps you out of delivery risk on work you do not control. We do not compete with you: no paid, no social, no brand, no search.

We give you a brief you can act on, and a way to show the value of what you already secure.

The brief. Which facts about your client need to appear in which named sources, ranked by the buying questions those sources decide. Placements, not topics. In food and drink supply that is usually a mix of certification registers, trade body listings, sector directories and the trade press, and we tell you which is which so you are not pitching a story to a database.

The proof. Each month we show which of your placements were actually cited, by which AI platform, on which buyer question. Very few PR agencies can show that today, and it is increasingly what clients are asking for.

We do not pitch, write or place. That is your work, and we do not take a margin on it.

Some of it, and they should carry on doing it.

What they are generally not set up for is the verification half: checking a certificate at the issuer with its scope and validity dates, filing evidenced corrections with third-party data providers, holding provenance against every claim, and standing behind a record as its named custodian. That is a data governance job rather than a content job.

Their work makes AI more likely to mention your client. Ours makes AI able to qualify them.

The same referral structure. If you work on positioning, messaging, customer insight or measurement, we sit underneath it: whatever you decide the business is, we make it checkable by machine and keep it that way.

One model. You introduce, we deliver in our own name, your client pays what they would have paid us directly, and you earn a referral fee.

We do not put the work under someone else's brand. A record whose custodian is anonymous is not verifiable, and verifiability is the whole product. We will always tell you before we contact your client about anything, and we will not sell them services you already provide.

Ask for the free AI Buyer Visibility Score. Or send us three buying questions you would want to win, and we will run them across the five platforms and show you the answers, with the competitors named and the reason you were left out.

If you are on every list, we will tell you so.

The audit produces a scoped plan. Most clients go to the build, because that is where the missed questions actually get fixed. Some have a strategy gap to resolve first, and we will say so rather than sell around it.

There is no obligation, and you keep the report either way.

You keep the EntityCore™. It is on your domain, the content is yours, and it is portable. What stops is the watching, the correcting and the re-measuring, so the record ages from the day you stop.

Book a call with Helenor, email hello@knownentity.ai, or message +44 7411 934700 on WhatsApp.

Known Entity.ai, Folkestone, Kent. A trading brand of WorkSpot Ltd.