Ranking 18 Businesses by Their AI Disruption Risk

Get smarter on investing, business, and personal finance in 5 minutes.

This newsletter is adapted from a recent video Drew Cohen did on AI Risk Ranking.

AI is top of mind, and what everyone wants to know is whether or not AI is going to kill their businesses.

In this week’s Five Minute Money, we are going to rank 18 different businesses by their AI risk, along with covering some of the key headwinds from AI that these businesses are facing.

The ranking runs from S, the best, down to F, the worst, and is based solely on the AI risk inherent in the business model.

Something could rank very well because it does not have a big AI risk and still not necessarily be an attractive investment, because the valuation can be extraordinarily high.

These are not stock recommendations, just a ranking of AI risk, and it is important to know that some businesses may have real AI risk, but that risk may already be more than priced in.

The way to think about the tiers themselves is as a sliding scale of how real and how immediate the risk looks.

When there is some AI risk that still seems a little far-fetched, though still possible, that tends to land in A.

By the time it reaches B, there is more real AI risk showing up, though still not too bad, and by the time a business drops to C, that AI risk has become more formidable.

Let’s start the ranking at the low end at F Tier.

F Tier.

Chegg.

Chegg is an easy place to start, and a useful place to set the standard for what the worst possible outcome looks like.

It is an education company a lot of college students use to help learn, which in practice often means helping them get through their homework, and the stock had not been doing great even before ChatGPT arrived.

Afterward, it got absolutely decimated, down to an $84 million market cap.

This is not just about valuation and sentiment: Chegg's financials actually got hit hard, with negative 47% revenue over the last 12 months and negative 20% operating margins, so the business itself is shrinking, not just the multiple the market is willing to pay for it.

It has already been disrupted by AI, for the simple reason that everything a student used to pay Chegg for in terms of homework help, a simple chatbot can now do for free.

Even if the chatbot's answer is not 100% correct, it is good enough for most people, and most large language models are already quite good at landing on the right answer.

That is what the bottom of this ranking looks like, a business that AI has already taken apart rather than one merely facing a future risk.

Let’s move up the list to C Tier.

C Tier.

Atlassian.

Atlassian until recently ran a negative operating margin despite generating a lot of free cash flow.

Share-based comp came to $400 million against revenue of only $1.75 billion, meaning roughly 25% of revenue was being paid out in stock-based comp, quite high for a business this size.

In our Atlassian business breakdown, I flagged their heavy spending across the board, not just marketing, which was already high, but an R&D line that looked really high for a company that did not appear to be shipping much meaningful product improvement; expert-network transcripts quoted in that research showed real dissatisfaction, people saying the product had not meaningfully gotten better in years.

There was also a distracting stretch of co-founder tension playing out in public: one co-founder bought a roughly $100 million mansion, the other bought the house next door and started making construction changes without telling him, upsetting him in the process, and on top of that, one of them bought a stake in an NBA team.

None of it read like a company being tightly run, more like constant noise.

To be fair, and to some surprise, Atlassian pulled its numbers together faster than expected, moving from a deeply negative operating margin to a 12% operating margin in a single quarter, a real improvement.

Still, that is a modest margin for a software business that is two decades old and not a new company by any measure, and it does not erase a real, lingering question mark around how well this business is being managed, even though sometimes pressure and risk are what pull a team together.

Looking at the actual products: Jira is widely used by software developers for planning, tracking, and coordinating work, and it is only becoming more important as more code gets written and needs to be coordinated.

Confluence, used for creating and collaborating on documents, carries more of the AI risk, since more collaborative work may increasingly start inside an AI tool instead, whether that is a dedicated team app like Slack or, inside a smaller business, something like ChatGPT or Claude directly, and Confluence is trying to be the tool that work continues to start in, with real competition and a product roadmap that has not looked especially strong.

This one sits on the border with D.

The next company in this tier is Intuit.

Intuit.

Intuit runs three main segments.

QuickBooks, which small businesses use for accounting, splits into an older, slowing product and QuickBooks Online, a $10 billion business and Intuit's biggest, followed by TurboTax.

Credit Karma is a weirder business: an app where people hand over financial information for free credit scores, while Intuit sells leads generated from that data to financial institutions pre-qualifying users for loans, maybe 5 different institutions at a time.

That lead-selling model looks pressured, since personal-finance tasks are likely to shift to AI agents over time, though Credit Karma is a smaller piece regardless, about $2.6 billion of Intuit's roughly $21 billion in total revenue.

TurboTax is tougher to call: a good chunk of filings are simple enough that an AI could likely calculate them, but more complicated returns, and the growing TurboTax Live product that connects users to a live tax expert, still benefit from people wanting a brand to rubber-stamp their return before sending it to the IRS rather than trusting it entirely to an AI model.

Over time, trust in AI tax preparation is likely to grow, and the lower-end part of the segment has never been that profitable anyway.

Intuit trades around 15x LTM earnings and 10x forward earnings, but the base case is probably fine, with real AI risk spread across its products.

Adobe.

Adobe is a very contentious stock, and in my Speedwell Research report on them, I identified 8 potential AI risks that face Adobe.

None is a single business-model killer on its own, aside from a kind of existential risk that applies to every software company if AI simply does everything well.

Still, a handful of smaller risks chip away at the thesis: if revenue growth slows and operating leverage erodes as Adobe keeps reinvesting to defend against AI, an optically cheap stock around 10x earnings stops looking so cheap.

So far that has not clearly happened; revenue growth sits around 13% with operating margins still expanding, though annualized recurring revenue has been declining nearly every quarter, partly a currency effect Adobe expects to fade.

What stood out more than the financials was the fact pattern: a CEO succession that dragged on for roughly half a year without a named replacement, an internal pick that still seemed uncertain even to the new CEO and his predecessor when directly asked on the 3Q26 earnings call why he was chosen, plus a delayed price increase, something companies avoid unless feeling less confident.

New AI tools for generating video and creative content are not directly competitive with Adobe, which is about modifying creative assets rather than creating them from scratch, but the overall picture suggests more internal uncertainty about AI disruption than the financials alone show.

CoStar.

CoStar has been brutally beaten down, about $80 from its peak and down 60%.

Its main segment is the CoStar platform, the top resource for real estate brokers, especially lease brokers, who rely on its proprietary data to price office leases.

Its other major platform is Apartments.com, one of the first results that comes up when searching for an apartment, monetized through sponsored listings sold as a package, including virtual tours through a company CoStar acquired.

Apartment searching happens infrequently and typically starts on Google, so CoStar already has to pay to reacquire that traffic; the risk is that an AI agent could eventually scrape listings directly.

CoStar would likely try to block AI agents to protect its sponsored ad business, but with many other listing sites and most properties dual-listed, that restriction is hard to make stick, and a lot of listings are also available through MLS without being exclusively proprietary; if Zillow instead leans into becoming more of an API business, charging agents a fee just to ping its listings, that would pressure CoStar further.

There is also no actual transaction on the platform: once a renter finds a place, they connect directly with the property, often through a listing the owner already maintains elsewhere, which an AI agent could find and use instead.

A secondary platform, Homes.com, has burned a lot of capital without clearly paying off.

These sponsored ad businesses depend on giving people a strong reason to keep coming directly to the site, and CoStar's hold on that looks shaky.

Let's move on to B Tier.

B Tier.

Salesforce.

Salesforce should be fine on the AI risk itself, but more companies are likely to find more friction continuing to expand everything through a single platform like Salesforce, and more of them will have other options available to them than they did before.

The broader theme playing out in software right now is that incumbents are competing harder against each other than against outside disruption: Salesforce has pushed into ITSM, IT service management, ServiceNow's core turf, while ServiceNow has pushed into CRM, Salesforce's core turf, because both want to control the AI layer, and owning more of a customer's overall workflow matters more now than it used to.

Enterprises are likely to resist letting any single vendor take over too much of that workflow and will keep wanting multiple suppliers rather than concentrating everything with one, and it is still genuinely hard to switch off Salesforce given the so-called golden record of customer data it holds, so an imminent exodus away from the platform looks unlikely.

There is also a chance some companies decide to build some of this functionality in-house instead, though that feels different from migrating off an established platform: a brand-new piece of internally built software that is low-risk from day one and gets trusted gradually as it is iterated on over time is not the same as ripping out a mature system and creating a brand-new replacement from scratch, which looks much less likely.

The more real risk is that new work increasingly starts on an AI layer instead of directly inside Salesforce's own interfaces, pushing Salesforce further into the back end as more of a plumbing pipeline, which makes cross-selling harder and makes it harder to raise prices, even if the existing core software itself is unlikely to be ripped out anytime soon.

That points to longer-term pressure rather than an immediate threat.

Booking and Airbnb.

Booking and Airbnb get covered together here.

By and large, in the immediate term, neither business looks threatened or likely to change much, but the more useful question to sit with is longer-term: does AI make each of these businesses better or worse 5 years from now?

When I ran my own test on Muse, I asked Muse to search for a room in Florence and then price-check the results it returned.

Booking had the better price on 3 of the 5 hotels, one was a wash, and the remaining one was actually cheaper booking directly through the hotel's own site, which was what actually happened when this test was run.

If that pattern holds up and becomes a habit, even losing a modest share of bookings to people going direct starts knocking a few points off top-line growth and nudges things in the wrong direction, on top of pressure on Booking's sponsored ad business as search behavior itself starts to shift toward AI agents.

Booking still has real insulation working in its favor: a mobile app, a loyalty program, and meaningful direct traffic.

Roughly half of all bookings come through people enrolled in that loyalty program, though there is still a good amount that does not, and how durable that loyalty stays if habits change is genuinely hard to say.

At around 14x LTM earnings and 12x forward earnings, how much of this risk is already priced in is an important question on its own, but on the narrower question of whether AI makes this a worse business, the view here is yes, that it probably does, acting more as an intermediary between Booking and the traveler over time.

That places Booking in the B tier.

Airbnb lands in roughly the same bucket for a different reason entirely: not because people stop choosing Airbnb, which has a genuinely great brand, a great app, and a meaningful amount of proprietary inventory that keeps people saying, when they think of that kind of trip, "let me just go to Airbnb."

The concern instead is that a lot of Airbnb's current strength rests on friction that keeps hosts from listing their properties anywhere else.

Booking actually has a sizable alternative-accommodation business of its own, roughly 2/3s the size of Airbnb's, and Expedia's Vrbo is another option sitting right there too, yet a lot of hosts only bother listing on Airbnb because it is the easiest platform to set up on, and most of them do not want the hassle of managing a calendar across three different platforms at once, in part because they are not particularly technical and do not know the software exists that could help them do it anyway.

If AI can now set a host up across all of those platforms simultaneously, it stops mattering that Airbnb had the best onboarding process in the category, since a host could just go to something like Muse and say, in effect, help me book across all of these different platforms and manage my calendar across all of them.

That would make previously exclusive listings show up elsewhere too, compounding with more AI-driven price-checking happening on the demand side at the same time.

That moves Airbnb up a little, closer to the A tier, since it still holds something of a walled garden around its proprietary inventory, even if that garden is increasingly under threat.

There is also a valuation wrinkle worth noting: for a long time, investors gave Airbnb some credit in its valuation for eventually building out a sponsored-ad business the way Booking and Amazon have, and at this point that looks less and less like it is going to happen.

Brian Chesky has called Airbnb's sponsored advertising opportunity unperishable, but it looks like AI might just mean that opportunity perishes after all.

Let’s move up to A Tier.

A Tier.

Microsoft.

Microsoft has many businesses, and cloud is now its largest profit center, with a lot of AI demand supporting that.

One specific risk is that a large share of its cloud backlog comes from a single customer, OpenAI, but that alone is not a big risk as long as AI demand stays real, since suppliers could always be swapped out.

The real risk is that workflows change: software is still roughly a $70 billion profit pool for Microsoft, and the question is whether work still opens with a Word doc or Excel sheet with a Claude plugin, or whether the process shifts to something more AI-forward, starting in Claude or ChatGPT and editing directly in those apps.

That has not happened yet, but there is no clear reason it would not eventually.

That lingering software headwind is why this is not an S, though it should manage it fine; a further risk is the so-called platform layer collapsing with AI doing more work at the database layer, which still seems far-fetched but possible.

S&P Global.

S&P Global, for some reason, has been caught up in a lot of the broader AI sell-off.

Its market intelligence segment is a roughly billion-dollar profit pool, about 15% of the total, and is where investors are most worried about AI disruption, mostly data analytics and insights.

That is an even smaller slice, since things like its CapIQ Excel add-in, which pulls proprietary S&P Global data, are not easily replaced: a firm like Goldman Sachs putting together a presentation does not want to risk the underlying data being wrong, so paying S&P Global to ensure accuracy still makes sense.

S&P Global's main business, though, is its credit rating agency, which is not at risk from AI.

A better AI-generated credit rating would not matter, because what buyers are really purchasing from S&P Global, Moody's, or Fitch, is a rubber stamp: corporations need that stamp to sell bonds, since bonds rated by one of the more than a dozen smaller agencies do not get the same yield, and pricing a bond 10 basis points worse because of a cheaper rater is far more expensive than paying S&P Global.

If this business survived the financial crisis rating junk mortgage-backed securities as triple-A and came out the other end intact, it is hard to see AI displacing it.

S&P Global also owns advantaged assets like the Platts commodity indexes and the S&P 500 index itself, which hedge funds pay a licensing fee just to reference in performance reporting.

Constellation Software.

Constellation Software owns a large portfolio of small, vertical market software companies.

It looks more insulated than most software businesses because its end customers are not the type to vibe-code a brand-new piece of software for themselves: the software is mission-critical, runs their entire operations, is usually not that expensive relative to the value it provides, and is custom-built by Constellation to the specific end user.

There is some risk across all software companies if AI keeps getting better and better at writing great software, since in pure behavioral terms, it does not seem like something a typical business owner is eager to rip out and replace with homegrown code.

A useful way to think about why: building a little piece of software for a personal task can be easy and genuinely helpful, but there is a real difference between that and moving an actual business onto something self-built.

A business doing $5 million in revenue is unlikely to put its core operations on software it created in-house, because the people running it generally do not deeply understand how that software works under the hood, and if something needs a consultant to make sure the cybersecurity is up to date, it just makes more sense to outsource the whole thing to a vendor like Constellation.

This one sits at the edge between A and B.

ServiceNow.

ServiceNow was hammered by the market until fairly recently and has had a nice recovery since, up 70% from its lows.

Part of that recovery reflects the market starting to recognize that ServiceNow is not quite the threat-to-AI story some people feared.

ServiceNow has something called a configuration management database that tracks every device across an organization, how all of it is connected, along with permissioning and governance, which turns out to be a genuinely strong position to be in once AI enters the picture, since no company wants AI roaming freely through its systems without guardrails around what it can and cannot access and do.

ServiceNow is trying to become that AI control center: the idea is that whichever AI models a company wants to use, it has to use them through ServiceNow or in coordination with it, which helps protect the business even if that protection is easy to overlook sitting quietly in the background.

It also is not a service a company would want to replace, much of it tied to IT service management, meaning that when something breaks, ServiceNow is the software companies use to manage and resolve the problem.

If a business replaced the very software it uses when other software has problems, then what happens when the replacement itself has a problem?

The company would need an entire internal team just to handle whatever it swapped in instead of outsourcing that function to ServiceNow, which already uses AI to make the process faster and easier.

It is hard to see that as something businesses are eager to give up; it feels too mission-critical to replace.

AppFolio.

AppFolio is vertical market software for property managers, and it is just a really good product that property managers genuinely like.

It started out serving smaller, SMB-sized property managers and has been moving up-market over time, picking up larger customers as it goes.

It has always had a cleaner tech stack than its larger competitor Yardi, with better interface design and a track record of rolling out new products more quickly, and with all of the recent advances in AI, it has been rolling a lot of that out too, including features that help facilitate payments.

Property managers are unlikely to want to vibe-code a replacement for this themselves: the software is mission-critical, they are running their entire business on it, they want the latest features, and they generally do not want to take on responsibility for maintaining their own software, which is a hard, annoying job that defeats the entire point of outsourcing it to a vendor in the first place.

It also fits a broader pattern worth naming: new companies are not likely to go build their own vertical market software from scratch, but it is a different story for a brand-new startup that builds this kind of software itself from day one, since the risk there is low from the beginning and the trust in it builds gradually as it gets iterated on over time, which feels different from a mature business ripping out AppFolio and attempting its own migration.

There is a small amount of edge-case risk here, where a very small property manager might simply never start using vertical market software like this at all rather than ever switching away from it, which is part of why this sits a little on the edge of its tier, though that looks unlikely to become a meaningful trend.

Amazon.

The real AI risk to Amazon is AI agents.

A lot of Amazon's monetization now happens through sponsored ads, which are increasingly how the business makes money, and that revenue has been coming in and growing very quickly lately; it is high-margin revenue, worth about $76 billion within its advertising segment, which also includes Amazon Prime.

That is a business Amazon should largely be able to defend, though not without some real risk: Muse tried to search Amazon directly, and Amazon ended up restricting it, which suggests Amazon will likely keep doing that with AI agents generally unless it finds some way to monetize that access instead, perhaps by charging agents to connect through an API or some similar arrangement.

What makes Amazon resilient is less the fast delivery itself and more the trust built up over years: drop-off points to return an item across the country, and the accumulated confidence that if something is wrong with a product, Amazon will just take care of it.

That makes buying from an unfamiliar website feel riskier by comparison, and it makes people nervous to shop anywhere else for purchases they would rather not think too hard about.

This comes back to a framework worth applying broadly here, the idea of consumer surplus and a consumer hierarchy of preferences: consumers very much prefer shopping on Amazon because of the trust, the delivery, and the habit, which together feel completely natural.

Switching to an AI agent might mean more selection and the benefit of being price-checked automatically, but it also adds work: does the buyer trust wherever the item is actually coming from, and will it be easy to return if something goes wrong?

For most of those reasons, most people are still going to prefer buying on Amazon directly.

Where Amazon has a real gap is discovery purchases, the category where a shopper does not know exactly what they want yet, something like browsing for a new office plant, where Amazon has never been particularly good and AI could plausibly do better.

But most Amazon purchases are highly intentional: if someone knows exactly what they want, going straight to Amazon and buying it is still quicker than waiting for a page to load somewhere else, and that includes scrolling past the sponsored ads at the top of the results.

Sponsored ads function as a kind of tax on the shopping experience, but a fairly fair one: it is not a big deal to scroll one extra second to reach the organic listings, even for someone who has never clicked a sponsored result and bought from it in their life.

Sea Limited.

Sea Limited sits in a similar position, with a similar sponsored-ad dynamic to Amazon's, though its sponsored ad business is not as large a share of the total.

The bigger difference is that Sea has built in a lot more discovery functionality: spend any time in the app and a lot of it is gamified, with spinning wheels, countdown clocks, and ways to save money by engaging with those mechanics, all of which give someone a reason to keep opening the app directly rather than routing a purchase through an AI intermediary.

Sea is also building out its own logistics layer more and more, which adds further insulation.

The real difference between Amazon and Sea is that Amazon is driven more by intentional purchases, while Sea leans more heavily on discovery: people genuinely want to go into the app to find things, and discovery is a good trait to have here, because it means it will be a long time before someone goes to an AI agent and says something like, "surprise me with things I did not even know I wanted, and make it fun with countdown clocks," since that kind of experience is not something an AI agent currently replicates.

That keeps people going directly to the app rather than around it, even as Sea also competes against TikTok, which is pushing into some of that same discovery-driven commerce territory.

Mercado Libre.

Mercado Libre gets a similar ranking for similar reasoning, leaning even more toward intentionality-driven shopping than Sea does, while also owning its own logistics layer and its own payments business, giving it the same kind of habit, delivery, and trust advantages that protect Amazon.

Advertising is a bigger piece of MercadoLibre's business than it is for Sea, but it should still be able to command direct traffic to its own site, and, much like Amazon, it will likely cut AI agents off wherever it does not make sense to let them operate inside the marketplace.

Moving up to our final and highest tier, S.

S Tier.

Palantir.

Setting valuation aside, even though LTM earnings sit around 167x and forward earnings around 104x, which is definitely up there, there is not much AI risk to see here.

Palantir's key advantage is what it calls the ontological layer, essentially a digital twin, a digital footprint of a business that exists alongside the physical one.

In practical terms: a company shipping different items, running warehouses, tracking different logistics trucks and inventory, has all of that tracked on Palantir, and AI lets it do a lot of new things with that data, like rerouting a shipment that is being delayed because of weather, or moving raw materials to a different warehouse to get them somewhere else faster.

These are mostly business use cases, though Palantir is also used heavily in military operations, since it tracks where all the air fleets are, all the drones, and all the infantry personnel, and AI is effectively enabling the military to do things it was not able to do before, making it a more effective operation overall.

This is hard to disrupt because once a business builds out its ontological layer on Palantir, it is very unlikely to switch away from it.

There is no real reason to, since no one else is operating at that level currently, making Palantir something close to the only real option in that regard, and switching out would be a total pain regardless.

Very importantly, AI does not actually replace what Palantir does, it enables it: a lot of what the ontological layer creates is essentially a real-world footprint that is deterministic, with permissions and guardrails built around it, exactly the kind of structure AI needs wrapped around it to be used safely and effectively.

AI is more of an engine for Palantir here, not a real threat to the business.

CrowdStrike.

CrowdStrike is a quick one: AI looks like an enabler here rather than a risk, not a threat to the business model.

One real concern was that a company like Anthropic, with frontier AI capabilities, could move into cybersecurity directly and stop sharing its models with third parties, since cybersecurity defense is one of the few areas that really needs access to the single best model available at any given time, because an attacker who gets access to that same best model can find holes that a lesser model would miss entirely.

For most other tasks, fewer and fewer things actually need the most intelligent frontier model in the world; something like basic bookkeeping just needs a model that can reliably reconcile a set of statements, nothing close to frontier-level intelligence.

Cybersecurity is different, which is why a lab like Anthropic going direct looked like a real threat to a business like CrowdStrike.

Instead, the leading AI labs ended up partnering with cybersecurity companies, giving them early access to their models that is now built directly into their products.

The remaining risk is that a frontier lab eventually decides it needs new sources of revenue and tells the market, in effect, that anyone who wants the best cybersecurity functionality now has to come directly to them rather than going through a third party like CrowdStrike.

Even in that scenario, CrowdStrike has built up other cybersecurity assets that are not trivial to quickly recreate, like endpoint detection and response, along with a large base of data from past attacks that it can learn from and abstract out to protect other customers when it catches something new.

For that reason, this is not something to be particularly worried about in the near term.

Shopify.

Shopify is a company that arguably should not have been caught up in the broader AI narrative at all; there is essentially no AI risk here.

Shopify helps merchants run the back end of their business, and it does not really matter how customers end up purchasing from them, since Shopify is fundamentally back-end software and infrastructure.

It provides the Shopify website itself, but that has never been the reason merchants choose it; they choose Shopify for everything else it handles, organizing catalogs, tracking inventory, and facilitating payments and returns, so even if a merchant's website were eventually built with AI, that would not be disruptive, since the website is just one of many services it provides.

The point of software is that it is deterministic, doing something reliably, and Shopify is well positioned to keep incorporating AI where it is helpful rather than being replaced by it.

Plans often start around $30 to $50 a month, and on the purchasing side, it does not matter to Shopify whether a shopper visits a merchant's site directly or buys through an AI agent, since Shopify gets paid either way.

Closing Thoughts.

Looking at all 18 of these rankings together, a few patterns show up more than once.

The businesses that land near the top, Palantir, CrowdStrike, Shopify, ServiceNow, and Constellation Software among them, tend to share a common trait: AI does not replace what they do, it runs on top of infrastructure a customer is unlikely to rip out, whether that is an ontological layer, a history of past cybersecurity attacks, or the back-end plumbing behind a merchant's catalog and payments.

The businesses sitting lower down, Chegg most of all, but also pieces of Atlassian, CoStar, and Intuit's Credit Karma, tend to share the opposite trait: a product built around a task a capable model can now approximate well enough, often for free, that the old reason to pay for it starts to fade.

In between sits a wide band of sponsored-ad and marketplace businesses, Amazon, Sea, Mercado Libre, Booking, and Airbnb, where the real question was never whether AI kills the business outright, but whether intentional, habit-driven buying keeps winning out over a shopper handing the task to an AI agent instead.

For now, trust, delivery, and habit are still doing most of that work, though that edge looks more durable for some of these names than others.

None of this is a stock recommendation, and for several of these companies the valuation already assumes a lot has to go right, with or without AI in the picture.

But the AI risk itself, at least from this vantage point, looks far more concentrated in a handful of specific business models than spread evenly across software and e-commerce as a whole.

For more on AI Risk Ranking, check out this video below.

YouTube FAQ

Nothing in this newsletter is investment advice nor should be construed as such. Contributors to the newsletter may own securities discussed. Furthermore, accounts contributors advise on may also have positions in companies discussed. Please see our full disclaimers here.