Are Financial Advisors Ready for the 100 Most Promising AI Startups List? in Switzerland

Direct answer: Most Swiss financial advisory practices are not yet ready for what a strong Swiss showing on a global AI startup list actually implies — not because they lack interest in AI, but because their client workflows, data infrastructure, and websites were never built to plug into agentic tools. The list itself is a signal, not a product; readiness means having the underlying systems in place before the startups on it start shipping usable enterprise tools.

In August 2026, FintechNews.ch reported on the 100 Most Promising AI Startups of 2026 list, noting a strong Swiss contingent among the companies named. That is the fact on record: Switzerland punched well above its size on a list meant to spotlight the AI companies most likely to shape the next few years of enterprise and financial technology. A precise count of Swiss entrants, or exactly which sectors they cluster in, is not publicly broken out in the coverage available to us, so we won't invent a number. What we can say with confidence, reasoning from the general pattern of these lists and from how AI adoption has moved through financial services generally, is that a national cluster of promising AI startups tends to precede a wave of enterprise tooling built specifically for that market — often before the incumbents in that market have adjusted their own infrastructure to take advantage of it. For financial advisors operating in Switzerland, that gap between "AI startups exist near us" and "our practice can actually use what they build" is the real story here, and it is worth taking seriously now rather than after competitors have closed it.

What the Swiss AI Startup Wave Actually Signals

A list like this is a leading indicator, not a finished product catalogue. When a market produces a disproportionate number of startups recognized on a global "most promising" list, it usually means three things are converging: available technical talent, investor appetite, and — critically — enough friction in existing workflows that founders see a clear wedge to build into. Switzerland's financial services sector, with its dense concentration of private banks, wealth managers, and independent advisory practices, is exactly the kind of environment that produces that friction. Compliance requirements are heavy, client expectations for personalized service are high, and much of the back-office work — portfolio reconciliation, client reporting, meeting prep, regulatory documentation — is still done in ways that have not changed much in a decade.

That friction is what AI startups are built to exploit. It does not mean every Swiss advisory firm needs to adopt a startup's product tomorrow. It means the pipeline of tools aimed at exactly this kind of workflow is thickening, and the practices that have already organized their own data and digital presence to be "AI-ready" will be the ones able to adopt the useful tools quickly when they mature — while everyone else spends that same window doing manual cleanup work just to get to the starting line.

Why This Is Different From Previous AI Hype Cycles

It's fair to be skeptical of any "AI startup list" as marketing noise. But the pattern that matters for advisors isn't the list itself — it's what tends to follow a concentrated startup cluster in a specific geography: localized products, in local languages, aware of local regulatory frameworks (FINMA requirements, Swiss data residency expectations), built by teams who understand Swiss client relationships aren't transactional. That is a meaningfully different starting point than a generic US-built AI tool bolted onto a Swiss practice after the fact.

Previous AI hype cycles in financial services largely centered on generic chatbots and robo-advisory front ends — useful for some segments, but never something a private client relationship in Switzerland was going to be built around. What's different about the current wave, as reflected in a startup list weighted toward a specific national ecosystem, is specificity. These are companies more likely to be solving for reconciliation, document intelligence, compliance drafting, and workflow orchestration — the unglamorous plumbing that determines whether a firm's daily operations run smoothly, not the flashy client-facing demo that gets attention in a press release. That distinction matters because plumbing-level tools are the ones that actually get adopted and kept; demo-ware gets tried once and abandoned.

It's also worth being honest about the limits of what a list like this tells you. It does not tell you which of these startups will still exist in three years, which will pivot away from financial services entirely, or which will get acquired and have their product roadmap absorbed into a larger platform. Advisors reading this kind of coverage should treat it as a weather signal about where investment and technical talent are pointed, not as a vetted shortlist of vendors to sign contracts with today.

Why This Matters Specifically for Financial Advisors in Switzerland

Financial advisors are a trust business first and a technology business second, which is exactly why AI adoption in this sector tends to lag behind e-commerce or media — and exactly why the lag becomes a liability once client-facing AI tools mature enough to be trustworthy. A few dynamics are worth naming directly:

Client expectations are shifting from the top down. Swiss private banking and wealth management clients are, on average, more technically literate and more exposed to AI tools in their own businesses than the general population. When a client sees their portfolio company or their own operations using an AI agent to compress a research or reporting task from days to hours, they start asking why their advisor's practice hasn't done the same for portfolio commentary, meeting notes, or document turnaround.

Regulatory scrutiny raises the bar for how AI gets deployed, not whether it gets deployed. Switzerland's financial regulatory environment means an advisor's practice cannot simply plug an off-the-shelf AI chatbot into client communications and call it done. Any AI-driven workflow — from automated portfolio summaries to client onboarding — needs an audit trail, clear boundaries on what the AI can and cannot represent to a client, and a human decision point before anything client-facing goes out. Firms that wait until they're forced to retrofit this will move slower than firms that build it in from the start.

The advisors who move first on infrastructure, not gimmicks, will compound the advantage. A strong Swiss AI startup cohort means more vendors targeting this exact vertical over the next 18-24 months. The firms whose internal systems — CRM, document management, client portal, website — are already structured in a way that can accept an AI agent layer will onboard those tools in weeks. Firms whose data lives in disconnected spreadsheets and unindexed PDFs will spend that same period just preparing to be ready, while competitors are already live.

Talent and hiring pressure will also shift. A visible AI startup ecosystem in Switzerland tends to pull technically skilled people toward those companies, which makes it harder for a mid-sized advisory practice to hire and retain in-house engineering talent to build this kind of infrastructure alone. That's a practical reason to think in terms of a technology partner relationship rather than trying to staff an internal team for a capability you'll only need to build once and then maintain.

Referral and reputation dynamics are quietly changing too. Independent Swiss advisors have long relied heavily on word-of-mouth and long-standing personal networks. As AI-assisted research tools become a normal part of how prospective clients — particularly younger clients inheriting wealth or founders newly liquid from an exit — vet a potential advisor, a firm's digital presence starts functioning as a reference check that happens before the first phone call. A firm that looks credible and current in that check gets the meeting; one that looks stagnant may not, regardless of how strong its actual track record is.

What Actually Changes for Your Website and Client Systems

This is the part that gets skipped in most commentary about AI startup lists: the list is about the vendors, but the readiness question is about your own practice's infrastructure. Concretely, three things change for a Swiss financial advisory practice thinking seriously about this shift.

From Static Client Portals to Agent-Ready Systems

Most advisory websites and client portals in Switzerland were built as brochure-and-login experiences: a public site describing services, and a secured portal for statements and documents. That structure does not accommodate AI agents well, because agents need structured, permissioned access to client data, documents, and workflows to do anything useful — not just a login screen bolted onto a decade-old CMS. Rebuilding around AI Agents & Automation means designing the client-facing and internal systems so that an agent can, for example, draft a first-pass meeting summary from a call transcript, flag a document that's missing a required disclosure, or pre-populate a compliance checklist — with a human advisor reviewing and approving every output before it reaches a client. Our AI Agents & Automation work is built around exactly this kind of narrow, auditable automation rather than open-ended chatbots making unsupervised representations to clients.

Content and Discoverability Now Include AI Search

A second, quieter shift: as AI search and AI-assisted research tools mature, prospective clients researching a Swiss financial advisor increasingly ask an AI assistant for recommendations rather than searching Google directly. If your firm's website content isn't structured in a way that AI systems can parse and cite confidently, you become invisible in that research path even if your reputation and expertise are strong offline. This is the same dynamic covered in How to Get Your Brand Mentioned by ChatGPT (2026) — clear, well-structured, factual content about your services and expertise is what gets surfaced, not marketing copy optimized for old-style search engines.

Performance and UX Expectations Rise Alongside Automation

As AI-driven features get added to client portals — real-time portfolio insights, automated document retrieval, faster onboarding flows — the underlying technical performance of the platform matters more, not less. A slow, clunky portal undermines the credibility of any AI feature layered on top of it. Firms evaluating whether to build a fully custom client platform or extend an existing one should look closely at the trade-offs laid out in Cross-Platform vs Native Performance: What the Benchmarks Actually Show — the architecture decision made now determines how easily AI features can be added later without a full rebuild. And because these portals are still, fundamentally, interfaces a human advisor and a human client both need to trust and enjoy using, the interface design choices matter as much as the backend — see 10 Best UI/UX Website Examples (2026) for a sense of what a modern, trustworthy financial interface actually looks like next to a dated one.

How This Plays Out Differently by Firm Size

A one-person or small independent advisory practice and a larger multi-partner firm will experience this shift differently, and it's worth naming that rather than treating "financial advisors in Switzerland" as one homogeneous group.

A small practice typically has an advantage in speed: fewer systems to untangle, fewer stakeholders to align, and the ability to pilot a new workflow within days rather than months. The trade-off is resource constraint — a small firm usually cannot justify a large upfront infrastructure investment without a clear, near-term return, which means the sequencing matters more. Getting the highest-friction, lowest-risk automation working first, and proving its value internally, is usually the right order of operations before touching anything more ambitious.

A larger multi-partner or institutional practice has more resources but also more complexity: legacy systems that have accumulated over years, multiple stakeholders with different risk tolerances, and often a compliance function that needs to sign off before any new tool touches client data. For these firms, the constraint isn't budget — it's coordination. The practices that move fastest here tend to be the ones that create a small, empowered internal group to evaluate and pilot AI tools, rather than trying to get consensus across the entire partnership before taking any first step.

Either way, the underlying lesson from a startup wave like this one is the same: the infrastructure and internal clarity you build now determines how much of the coming wave of AI tooling you can actually use, independent of firm size.

What to Actually Do About It

Readiness is not a single project; it's a sequence of decisions, roughly in this order.

Start with an honest audit of where client data actually lives. Before any AI agent can help draft a report or flag a compliance gap, someone needs to know whether the underlying documents and records are even in a format a system can read reliably. This is unglamorous work, and it is also the single biggest predictor of how fast a firm can move once useful AI tools from this Swiss startup wave become available.

Define the narrow, low-risk automation wins first. Meeting transcription and summarization, first-draft client reporting, document intake and flagging — these are places where an AI agent adds real time savings with a human still firmly in the approval loop. Resist the temptation to automate anything that represents financial advice directly to a client without review; that is where regulatory and trust risk concentrate.

Rebuild the client-facing digital experience around the workflows you actually want to run. A website or portal redesign done well now should assume AI-assisted features are coming within the next year or two, even if you're not building them on day one. Designing for that future state costs little extra today and saves a full rebuild later.

Treat this as an ongoing capability, not a one-time project. The Swiss AI startup landscape referenced by FintechNews.ch will keep producing new tools aimed at this exact market. A practice that builds its own systems to be modular and agent-ready can evaluate and adopt the good ones as they mature, rather than being locked into whatever vendor they picked in a rush.

Assign clear ownership internally before any external work begins. Even when a firm brings in outside expertise to handle the technical build, someone inside the practice — often a partner or operations lead rather than an advisor focused on client work — needs to own the decisions about what data gets touched, what automations are approved, and how outputs get reviewed. Projects that lack that internal owner tend to stall regardless of how good the underlying technology is, because nobody has the authority to say yes to the next step.

Pilot before you commit broadly. Rather than announcing a firm-wide AI initiative, pick one workflow — meeting summarization is a common low-risk starting point — and run it for a defined period with a small group of advisors before deciding whether to expand it. This keeps the initial investment proportional to the uncertainty and gives the firm real evidence, rather than vendor claims, about what actually saves time.

What This Kind of Work Typically Costs

Scope varies by firm size and how much of the existing system needs rebuilding versus extending, but most Swiss advisory practices approaching this fall into one of three tiers:

Tier	Typical scope	Starting price
Essential	Website/portal audit, content restructuring for AI discoverability, foundational fixes	$1,000
Growth	Client portal rebuild or upgrade, initial narrow AI agent automation (reporting, document flagging)	$2,000
Enterprise	Full agent-ready infrastructure, custom automation workflows, ongoing platform development	$4,000+

Key Takeaways

• The strong Swiss showing on the 2026 100 Most Promising AI Startups list, reported by FintechNews.ch, signals a coming wave of AI tools aimed squarely at Swiss financial services workflows.
• Readiness is about your own systems — client data, portal architecture, website content — not about picking a startup's product off the list.
• Regulatory and trust requirements mean AI automation in advisory practices should stay narrow and human-reviewed, especially anything client-facing.
• Website and portal architecture decisions made now determine how easily AI features can be layered in later without a costly rebuild.
• Content structured for AI-assisted research is becoming as important as traditional SEO for being found by prospective clients.
• Treat AI readiness as an ongoing capability to build, not a single project to finish and forget.

Switzerland's AI startup momentum is a preview of what's coming to financial advisory workflows over the next two years, and the firms that prepare their own infrastructure now will be the ones able to move when the useful tools arrive. If you want help figuring out where to start, book a meeting with our team.

Frequently Asked Questions

What is the 100 Most Promising AI Startups list?

It's an annual list, referenced by FintechNews.ch in its August 2026 coverage, that recognizes the AI companies considered most likely to shape enterprise and financial technology in the near term. It functions as an industry signal about where investment and talent are concentrating rather than as a shopping list of finished products.

Why did Switzerland have a strong showing on this list?

The specific reasons behind Switzerland's showing weren't broken down in detail in the available coverage, but it's consistent with the country's dense concentration of financial institutions, technical talent, and a regulatory environment that creates clear problems for AI startups to solve.

Does this mean financial advisors in Switzerland should adopt AI immediately?

Not immediately and not indiscriminately. It means the pipeline of AI tools built for this exact market is thickening, so advisory practices benefit from getting their own data and systems ready now so they can adopt genuinely useful tools quickly when they mature.

What does "AI-ready" mean for a financial advisory practice?

It means client data is stored in structured, accessible formats; the client portal or website can support permissioned access for automated tools; and there are clear internal policies about what AI can and cannot do without human review.

Is it risky for a Swiss financial advisor to use AI agents with client data?

There is real risk if it's done carelessly — regulatory risk, trust risk, and reputational risk. The mitigation is narrow scope: use AI agents for internal, reviewable tasks like drafting or flagging, and keep a human in the loop before anything reaches a client.

What is an "AI agent" in this context, as opposed to a chatbot?

An AI agent is a system that can take a defined action within a workflow — retrieving a document, drafting a summary, flagging an anomaly — rather than just answering a question conversationally. In advisory contexts, agents are most useful when scoped narrowly to a specific task with human approval built in.

How does FINMA's regulatory stance affect AI adoption for advisors?

Swiss regulators expect financial advisors to maintain clear accountability for advice and communications given to clients, which means any AI-assisted output needs an audit trail and a human decision point before it's treated as final. This shapes how automation should be scoped from the outset.

Can AI actually reduce compliance workload for a small advisory practice?

Used narrowly, yes — automating first-pass document checks, flagging missing disclosures, or pre-populating checklists can reduce the manual burden. It doesn't remove the need for a compliance officer or advisor to review and sign off.

What's the biggest blocker to AI adoption for Swiss financial advisors today?

In most cases it's not skepticism about AI itself — it's that client data and documents live in disconnected systems that no automated tool can reliably read. Fixing that data foundation is usually the first real project.

Should a financial advisory firm build a custom client portal or use an off-the-shelf one?

It depends on how much the firm plans to layer AI-driven features on top over time. Off-the-shelf portals are faster to launch but harder to extend with custom automation; a custom-built portal costs more upfront but scales more cleanly as needs grow.

How long does a client portal rebuild typically take?

Timelines vary by scope, but a meaningful portal upgrade with foundational AI-readiness work typically runs several weeks to a few months, depending on how much existing data and infrastructure needs to be restructured first.

What does "agent-ready" infrastructure actually look like technically?

It generally means structured data storage, clear API or integration points between systems (CRM, document management, portal), and defined permission boundaries so an automated tool can access only what it's supposed to, with logging on every action it takes.

Will AI replace financial advisors in Switzerland?

Unlikely in the near term. The advisory relationship in Switzerland's wealth management sector is built substantially on trust and judgment that clients still expect from a human. AI is more likely to take over routine documentation and analysis tasks, freeing advisors for higher-value client interaction.

How does this trend affect smaller independent advisory practices versus large private banks?

Large private banks typically have more resources to build custom AI infrastructure in-house, while smaller independent practices are more likely to depend on external AI vendors — which makes it even more important for smaller firms to have their own systems structured to plug into those vendors cleanly.

What role does website content play in AI readiness?

Increasingly, prospective clients ask AI assistants for advisor recommendations rather than searching manually. If your website's content about your services and expertise isn't structured clearly, AI tools can't surface or cite it accurately, which affects how discoverable your firm is.

How is AI search different from traditional Google search for a financial advisory website?

Traditional search ranks pages by keywords and backlinks; AI search tools synthesize an answer by parsing and understanding page content directly, so clarity, factual accuracy, and clean structure matter more than keyword density.

What's a realistic first AI automation project for a Swiss advisory practice?

Meeting transcription and first-draft summarization is a common starting point — it saves real time, carries low client-facing risk since a human reviews the output, and doesn't require touching sensitive portfolio data directly.

How does data residency affect AI tool choices for Swiss financial firms?

Swiss clients and regulators often expect sensitive financial data to stay within specific jurisdictions or under specific handling standards, which narrows which AI vendors and hosting arrangements are appropriate — this should be a filtering criterion when evaluating any new AI tool.

What's the difference between the Essential, Growth, and Enterprise tiers for this kind of work?

Essential covers foundational fixes like auditing and restructuring a website or portal for AI readiness. Growth adds a client portal rebuild and initial narrow automation. Enterprise covers full agent-ready infrastructure and ongoing custom development for firms with more complex workflows.

Can an existing website be upgraded incrementally, or does it need a full rebuild?

Many practices can upgrade incrementally — restructuring content, adding permissioned access points, and layering in narrow automations — without a full rebuild, as long as the underlying platform isn't too outdated to extend.

How do I know if my current website platform can support AI agent integrations?

The key questions are whether it has an accessible API or integration layer, whether client data is structured rather than locked in flat documents, and whether the platform vendor supports modern authentication and permissioning standards.

What happens if a Swiss advisory firm does nothing about this trend?

Nothing happens immediately, but the gap widens: competitors who prepare their systems now will be able to adopt genuinely useful AI tools from this startup wave within weeks, while firms that haven't organized their data will spend that same window just catching up.

Are AI startups from this list going to sell directly to individual advisory practices?

Some may, particularly if they build to a specific vertical need; others will focus on enterprise clients like large banks first. Either way, tools built for this market tend to filter down to smaller practices over time as they mature and become more affordable.

What's the risk of adopting an AI tool too early, before it's mature?

Early tools can be unreliable, may not meet Swiss data handling expectations, and can create client trust issues if outputs are wrong or presented without adequate human review. Testing internally before any client-facing use is the safer path.

How should client consent be handled when AI is used in advisory workflows?

Best practice is to be transparent that certain internal processes — like drafting a first-pass summary — may involve AI assistance, while making clear that a human advisor reviews and takes responsibility for anything communicated to the client.

Does using AI agents change liability if something goes wrong?

The advisor and the firm remain responsible for advice given to clients regardless of what tools were used internally to prepare it. This is exactly why human review before anything client-facing goes out is a non-negotiable design principle, not a nice-to-have.

What kind of ongoing maintenance does an AI-integrated client portal need?

Beyond standard software maintenance, it needs periodic review of automated outputs for accuracy, updates as underlying AI models or vendor tools change, and ongoing monitoring of what data the automation is accessing and why.

How does this trend intersect with cybersecurity for financial advisors?

Any AI agent given access to client data expands the practice's attack surface, so permissioning, logging, and secure integration design matter as much as the AI functionality itself. This should be part of the initial infrastructure planning, not an afterthought.

Is there a difference between AI tools built for Swiss financial services versus generic international AI tools?

Tools built with awareness of Swiss regulatory requirements, language needs, and client relationship norms tend to fit more naturally into local advisory workflows than generic tools adapted after the fact, which is part of why a Swiss-heavy startup cohort matters for this market specifically.

How can a financial advisory firm evaluate whether a new AI vendor is trustworthy?

Look at how the vendor handles data residency and security, whether they support clear audit trails, whether their tool allows narrow scoping rather than broad autonomous access, and whether they have any track record with financial services clients specifically.

What's a reasonable timeline to expect for seeing return on an AI automation investment?

For narrow automations like meeting summarization or document flagging, time savings are usually visible within the first few weeks of use. Broader infrastructure investments, like a full portal rebuild, tend to pay off over a longer horizon as more automations get layered on top.

Should advisory firms wait for AI standards or regulation to settle before acting?

Waiting entirely carries its own cost, since competitors moving on foundational data and systems work now will be positioned to adopt whatever standards emerge more quickly. The safer approach is building flexible infrastructure rather than waiting for perfect regulatory clarity.

How does mobile experience factor into this shift for financial advisory clients?

Clients increasingly expect to review portfolio updates or communicate with their advisor from mobile devices, so any AI-enhanced portal or reporting feature needs to perform well on mobile, not just desktop.

What's the relationship between UI/UX quality and trust in an AI-enhanced advisory platform?

A confusing or dated interface undermines confidence in the accuracy and professionalism of whatever AI features sit behind it, so interface quality functions as a trust signal even when the AI itself is working correctly.

Can AI help with multilingual client communication in Switzerland?

Yes, this is one of the more practical near-term applications — drafting or translating client communications across Switzerland's multiple national languages, with human review before anything is sent, can meaningfully reduce turnaround time.

What kind of internal training does staff need before adopting AI agents?

Staff need clear guidelines on what the AI tool is allowed to do, how to review its outputs critically rather than accepting them at face value, and what to do when an output looks wrong or incomplete.

Is this trend specific to wealth management, or does it apply to all types of financial advisory practices?

The pattern applies broadly across financial advisory types — wealth management, retirement planning, corporate advisory — since the underlying friction points (documentation, reporting, compliance) are common across all of them, even if the specific tools differ.

How does client data quality affect AI project timelines?

Poor data quality is usually the single biggest factor that extends timelines, since automation built on unstructured or inconsistent data requires additional cleanup work before it can function reliably.

What's the first practical step a firm should take this quarter?

An honest internal audit of where client data currently lives and how accessible it is to any future automated tool — this determines almost everything else about how fast subsequent AI adoption can move.

How does AI Agents & Automation differ from simply hiring more administrative staff?

Automation handles repetitive, well-defined tasks consistently and at lower marginal cost over time, freeing administrative staff and advisors for judgment-intensive work that genuinely requires a person, rather than replacing headcount outright.

What ongoing costs should a firm expect after the initial AI automation build?

Beyond the initial development tier, expect modest ongoing costs for vendor tools, occasional adjustments as workflows evolve, and periodic review of automated outputs to ensure continued accuracy.

How does this Swiss AI startup trend compare to what's happening in other European financial hubs?

Detailed comparative figures weren't available in the source coverage, but the general pattern — a concentrated startup cluster preceding a wave of vertical-specific enterprise tools — tends to repeat across financial hubs, with local regulatory and language factors shaping which tools gain traction fastest.

Can a financial advisor test AI automation without touching live client data?

Yes, testing with anonymized or sample data before rolling out to live client workflows is a sound practice, and it lets a firm evaluate reliability without introducing unnecessary risk.

What's the risk of vendor lock-in with AI tools from this new startup wave?

Choosing tools that integrate through open, well-documented interfaces rather than proprietary closed systems reduces the risk of being stuck if a startup pivots, gets acquired, or shuts down.

How should a firm budget for AI readiness work relative to other technology spending?

It's reasonable to treat this as part of ongoing digital infrastructure investment rather than a separate line item, since much of the foundational work — data structure, website architecture — pays off for reasons beyond AI as well.

Does this trend affect how financial advisors should think about hiring technical staff?

Firms without in-house technical capacity typically rely on an external partner for this kind of infrastructure and automation work rather than building a full internal engineering team, which is usually the more practical approach for small and mid-sized practices.

What's the danger of over-automating client communication?

Clients in Switzerland's advisory market generally expect a personal relationship, and over-automating communication risks making interactions feel generic or impersonal, which undermines the exact trust that differentiates a good advisor from a low-cost alternative.

How quickly should a Swiss advisory firm expect to see AI tools from this startup wave become commercially available?

A precise timeline wasn't specified in the source coverage, but based on typical startup-to-enterprise-product cycles, usable, market-ready tools from a cohort like this generally take twelve to twenty-four months to mature into something a smaller advisory practice can adopt confidently.

How does firm size change the right AI adoption strategy in Switzerland?

Smaller practices generally benefit from moving fast on a single narrow pilot, while larger multi-partner firms need more coordination up front but often have more resources to build broader infrastructure once a pilot proves itself. Neither situation should default to inaction.

What should a firm do if it's unsure where to start on any of this?

Starting with a straightforward conversation about current systems, client workflows, and realistic near-term goals is usually more productive than jumping straight into a large project — book a meeting with our team to talk through what makes sense for your practice specifically.