Karsten Wenzlaff, Advisor
August 26th, 2025
September 8, 2026 | NCFA Insight | Artificial Intelligence And Data, Wealthtech Investing And Trading, Risk Compliance And Regtech

On September 8, 2026, Focal AI launched agentic AI capabilities for Canadian financial advisors that can read and fill forms, draft emails, prepare client materials and update information across advisor systems. Focal says its platform supports more than 130 integrations and can populate more than 400 fields in Conquest, taking the product well beyond meeting notes into CRM, financial planning, onboarding and administrative workflows.
Data points. Deloitte says 73% of advisory firms already use AI in some capacity, yet only 6% use agentic tools and just 5% have connected AI across systems. Focal is entering the much earlier part of wealth management AI adoption, where the software starts completing work rather than simply generating content.
Focal's Canadian platform handles meeting preparation, notes, follow up emails, CRM updates, planning data, forms and client deliverables. Its current Canadian product page lists Salesforce, Equisoft, Maximizer, Conquest, Microsoft Dynamics 365 and other advisor systems among more than 130 supported integrations, while its browser agent can populate 400+ Conquest fields using structured client information.
The practical difference appears after a client conversation. Instead of an advisor copying the same information into a CRM, planning tool and forms, Focal is designed to carry approved information into those systems and complete related administrative steps. Equisoft's Focal integration provides a concrete example where meeting summaries and selected tasks can flow into Equisoft/connect, while advisors review which extracted tasks are written into the CRM.
Focal's published Canadian Pro price is US$100 a month when billed annually, with enterprise pricing available by consultation. The Pro plan includes unlimited meetings, CRM integrations and the Conquest AI overlay, while custom workflow automations and AI agents for tasks such as onboarding sit in the enterprise offering.
Deloitte estimates advisors spend nearly 70% of their time on work behind the scenes and only about 30% building client relationships. Focal says users can reclaim 10 to 15 hours a week, although that remains a company claim rather than an independently measured productivity result.
Deloitte's 2026 analysis models roughly 30% to 100% higher advisor capacity by 2032 as AI, workflow redesign and technology maturity advance together, potentially freeing 25% to 50% of advisor time from lower value operational work. These are modeled scenarios rather than adoption forecasts, but they put a number on the economic prize if AI can reliably take work out of the advisor's day.
Canada is already producing operating evidence beyond Focal. In March, OneVest launched an AI native wealth operations platform spanning onboarding, account opening, fund movements, billing and document workflows. In a different financial services setting, Scotiabank reported more than 71,000 employees with access to its AI capabilities and 14 million AI actions since March 2026, showing that governed AI can reach substantial operational scale inside a Canadian financial institution.
More automation doesn't remove the advisor or dealer from the regulatory equation. The Canadian Investment Regulatory Organization's 2026 Compliance Report says its examinations will ask dealers about AI use and review the operational controls they have implemented to make sure those systems work as designed. CIRO also tells dealers to consider whether AI or automation of regulatory functions constitutes a material business change requiring advance notification.
Focal says its enterprise agents use human review controls. The Equisoft integration shows one version of that model by allowing advisors to review tasks before information is pushed into the CRM, and available Focal materials don't establish that its agents independently make suitability decisions or provide regulated financial advice. NCFA's recent guide to AI governance for Canadian financial advisors reaches the same practical point.
Firms remain responsible for supervision, privacy, recordkeeping, vendor oversight and the decisions made with the technology they adopt.
Focal also says its Canadian service uses Microsoft Azure infrastructure in Toronto, holds SOC 2 Type II certification, stores no meeting audio or video and doesn't use client conversations to train AI models. Those claims can reduce some vendor review concerns, but Canadian firms still need their own controls over data access, records, approvals and errors.
Focal has already secured useful distribution. The company says it was selected as a trusted AI partner for Financial Horizons' network of 7,000 advisors, while a separate partnership made Focal available to Designed Wealth Management's network of 180 advisors. Those figures describe potential network reach, not 7,180 active users, and shouldn't be treated as adoption numbers.
The company also raised a US$5 million seed round in October 2025, led by Distributed Ventures and Wischoff Ventures. At that point Focal was already building beyond meeting software into form filling, data entry, onboarding and know your client workflows, so the September 2026 launch is a continuation of a product strategy that predates the current agentic AI push.
Competition is substantial. Jump can take approved actions across meetings, email, CRM and financial planning systems, Zocks automates forms and CRM workflows, and Salesforce is adding AI agents to its financial services platform. As those capabilities become more common, Focal's Canadian case depends less on access to AI itself and more on integrations, local distribution, data infrastructure and how deeply its software becomes part of daily advisor work.
Focal's launch is more proof of where Canadian wealthtech is heading, but the numbers also show how early this market remains. Companies already know how to buy AI assistants. The current evolution is whether agentic products can complete enough real work across existing systems to improve advisor capacity without creating an equal burden of checking, correcting and supervising the automation.
If AI can reclaim a meaningful share of the nearly 70% of advisor time now spent behind the scenes, how much of that work will Canadian wealth firms actually be willing to delegate?
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer to peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: [www.ncfacanada.org](http://www.ncfacanada.org)
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |

On September 3, 2026, Nvidia announced a definitive agreement to acquire Hugging Face for US$12.9303 billion. The deal would put one of the biggest platforms for open source and open weight AI alongside the company that already dominates much of the market for AI computing.
The price includes about US$11.9 billion for Hugging Face stockholders and up to US$1 billion in equity awards for employees joining Nvidia. The transaction hasn't closed. Nvidia says it expects completion in the first half of 2027, subject to regulatory approvals and other closing conditions.
Nvidia is making a very public promise with the deal. Hugging Face will remain open. Developers will still be able to choose their models, clouds, inference providers and computing platforms. Nvidia hardware will not be required.
That promise goes directly to the tension. Hugging Face became valuable because developers, startups, researchers and rival chip companies could all build there. Nvidia can make that ecosystem stronger. Ownership can also make some of those same users wonder whether an open source AI platform can feel as independent once one of the most powerful companies in AI owns it.
Hugging Face has grown into one of the main places developers find, share and use open AI models. Nvidia says more than 18 million developers, researchers and creators use the platform, along with more than 200,000 companies.
The company was worth far less only three years ago. Hugging Face raised US$235 million in 2023 at a US$4.5 billion valuation, with investors including Google, Amazon, Nvidia, Intel, AMD, Qualcomm, IBM and Salesforce.
Nvidia is now paying close to three times that valuation. The premium makes more sense when Hugging Face is viewed as distribution, developer access and influence over how open models get discovered and deployed.
Hugging Face is valuable because millions of people already use it to decide what to build with. Nvidia is buying that relationship as much as the software behind it. The more developers stay, the more valuable the acquisition becomes.
Nvidia already has enormous power in AI computing. Reuters Breakingviews says Nvidia holds more than 80% of the AI accelerator market, while its chips have become a reference point for a growing market in GPU rental pricing.
That power is one reason the acquisition attracts attention. Nvidia will own a major open model platform while selling the hardware many of those models run on.
Hugging Face has also become important to Nvidia's competitors. Its own 2026 data says AMD and Nvidia are the two most active publishers of new open models on the Hub, with each releasing more than 200 model repositories this year.
AMD uses open models to prove its chips can run real workloads. Google, Microsoft, IBM and other companies also publish and distribute models through the platform.
The acquisition does not remove AMD, Google or other hardware and cloud providers from Hugging Face. Nvidia says support for rival silicon will continue. The tension comes from whether those companies remain just as comfortable investing there when the owner also competes with them.
Nvidia says rival chips will stay welcome. Nvidia's CEO Jensen Huang says developers will keep choosing their own models, frameworks, clouds, inference providers and computing platforms. Nvidia compute will not be required to build on or deploy through Hugging Face.
Developers are already debating what ownership could mean in practice. Some community reactions welcome Nvidia because open models create demand for compute. Others worry about future defaults, private repositories, hardware preference and whether another independent open source AI platform will eventually be needed.
Nvidia doesn't have to close Hugging Face for ownership to change how the platform feels. Developers will notice which hardware gets optimized first, which services are easiest to connect and whether rival products remain equally visible and easy to use.
Open models fit Nvidia's economics surprisingly well. Hugging Face says hardware vendors are publishing open models because a model optimized for their chips is one of the clearest ways to prove the hardware works.
Nvidia can therefore benefit even when the model itself is free to download. More open model use can create more inference and training demand across data centres, enterprises and local machines.
That dependence cuts both ways. Some of Nvidia's biggest customers, including hyperscalers and AI labs, are building their own chips. The Hugging Face deal gives Nvidia a wider developer base at a time when those customers are trying to reduce their own dependence on Nvidia hardware.
Open source AI gives Nvidia access to thousands of smaller users instead of relying only on a few giant buyers.
Nvidia can support open source AI and still benefit commercially from its growth. The company does not need every developer to buy a proprietary Nvidia model. It benefits when more models create more computing demand.
China is pushing hard in the same open model market. Hugging Face data shows Chinese labs released many of the largest open models in 2026. Qwen has become one of the most important model families on the Hub, with more than 151,000 derivative repositories.
Hugging Face says Qwen based models reached more than 2 billion downloads across repositories with declared parameter counts this year.
Chinese open models are also competing on access and cost. Hugging Face found that 59% of Chinese releases above 20 billion parameters used Apache 2.0 licences and another 22% used MIT licences during the period it studied, although some very large releases have begun adding commercial restrictions.
That gives developers another source of capable models as U.S. companies debate how open their own ecosystems should remain.
Open models are part of the technology rivalry between the United States and China. Nvidia's Hugging Face acquisition gives a U.S. company more influence over a global platform at the same time Chinese model families are winning large developer communities of their own.
One possible response to Nvidia ownership is that developers simply stay. Hugging Face already has millions of models, datasets, applications and established workflows. Rebuilding that network somewhere else would be difficult.
Microsoft's GitHub acquisition offers one useful precedent. Microsoft promised GitHub would stay open and independent, and competing developers and platforms continued using it after the acquisition.
Another possibility is that developers begin spreading their work across more places. ModelScope, GitHub, local model tools, cloud registries and private enterprise repositories already give users alternatives for parts of the Hugging Face experience.
A future competitor would not need to copy every Hugging Face feature on day one. It could win users by offering easier migration, open governance, strong model provenance or a clearer commitment to hardware independence.
Network effects make a full replacement difficult, but communities can fragment before platforms collapse. Developers can keep models on Hugging Face while using other tools for discovery, inference, deployment or discussion. Competition may arrive piece by piece rather than through one new platform.
No price increase has been announced. Nvidia says Hugging Face will remain open and hardware choice will continue. That leaves plenty of room for the acquisition to improve reliability, inference tools and enterprise deployment without raising basic access costs.
Costs could still change indirectly. Developers may pay more if the easiest experience ends up depending on premium services, Nvidia optimized infrastructure or harder to replace integrations. The opposite is also possible. Better tooling and stronger open models could lower the cost of running AI compared with closed model APIs.
The acquisition does not automatically mean higher prices. The more interesting cost risk is switching. A service can remain affordable while becoming expensive to leave because models, workflows, integrations and teams are built around it.
Startups could gain from Nvidia's reach. A stronger Hugging Face can give model companies better distribution, more reliable infrastructure and easier access to enterprise customers.
For founders trying to get an open model discovered, being close to a platform used by 18 million developers can be commercially powerful.
Startups may also have less bargaining power if distribution, compute and enterprise access become more concentrated around the same company. A startup can benefit from the platform while still wanting credible ways to deploy elsewhere.
That tension is already visible in competition for cheaper AI inference, where AMD and other hardware companies are trying to give developers alternatives to Nvidia's dominant GPU position.
The upside is distribution. The risk is dependence. Founders will care less about who owns Hugging Face than whether they can still take their models, customers and economics somewhere else when they need to.
Financial institutions face the same ownership question from a different angle. Banks and insurers are already putting AI into governed workflows where data controls, approvals, audit evidence and operational resilience are required.
Governed financial AI workflows become harder when a firm cannot easily change models, clouds or providers without rebuilding controls around them.
Portability can therefore matter more than ownership alone. A bank may be comfortable using Hugging Face under Nvidia if models can still travel across clouds and chips and the institution can keep its own data, controls and audit evidence.
Regulators are also paying more attention to AI vendor concentration and operational dependence as financial firms embed more external technology into critical work.
Financial institutions do not need every AI supplier to be independent. They do need credible ways to change suppliers, hardware and deployment environments without losing control of regulated workflows.
The deal could still produce a strong outcome for open source AI. Nvidia has the engineering resources, compute and enterprise distribution to make Hugging Face faster, more reliable and easier for companies to use.
If AMD, Google, cloud providers, Chinese model labs and independent developers keep contributing, Nvidia can own the platform while the ecosystem remains genuinely competitive.
The harder outcome is quieter. Hugging Face stays open, but developers gradually find Nvidia products easier, cheaper or better supported than alternatives. No door closes. Choice simply becomes less balanced over time.
That is why Nvidia's promise will be judged through product behaviour rather than the announcement itself.
The most valuable version of Hugging Face may be one where Nvidia owns it and its competitors still want to build there. If that happens, Nvidia gets a larger open source AI ecosystem without destroying the trust that made the platform worth almost US$13 billion.
Talking Point
Nvidia does not need to close Hugging Face to gain more influence over open source AI. The deal becomes more valuable if developers, startups and rival chipmakers keep using the platform anyway.
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |
Sep 3, 2026

Some buy now, pay later activity reaches Canadian credit files. Most routine instalment plans currently do not.
That is the short answer, and the qualifications matter more than the answer does. Whether a buy now, pay later arrangement appears on a credit file in Canada depends on the provider, on which credit bureau is involved, on whether a payment was missed, and on the month in which the question is asked. Payments made on time and payments missed follow different routes. Two people financing identical purchases through different providers may find entirely different records.
What follows sets out what is reported, by whom, to which bureau, and with what effect on a score, then explains why the answer is this unsatisfying. The position described reflects Canadian reporting as of September 2026, and it is moving.
The Financial Consumer Agency of Canada describes buy now, pay later plans as arrangements that finance a purchase with credit. Its research characterises the category as covering a wide range of credit arrangements and as generally a type of consumer credit, comparable to instalment lending. That framing is the reason the credit-file question arises at all. The obligation is credit, not merely a payment method.
The agency identifies several distinct payment models sold under the same label: pre-authorized debits, pre-authorized credit card charges, an instalment option applied to an existing credit card, retail credit cards, and financing arranged through a financial institution. The entity behind each model differs. FCAC lists the financial service providers active in this market as including banks, credit unions and caisses populaires, financing companies, and money services businesses such as financial technology firms.
That mix determines oversight. FCAC directs consumers with complaints to different regulators depending on who provided the financing: federally regulated financial institutions must maintain their own complaint-handling processes, while other arrangements fall to provincial and territorial regulators. Oversight therefore follows the provider rather than the product category.
Credit reporting is furnisher-driven. A bureau can hold only what a provider chooses to send it, and furnishing is voluntary. In the United States, four senators on the Senate Banking Committee wrote to the major credit reporting companies in May 2026, reporting that several American providers had told them they were not sharing this data with credit bureaus.
The Canadian position is documented more thinly. The Canadian Lenders Association, an industry body, described the position in late 2025 as one in which inclusion of this data in credit files is voluntary, variably reported, and inconsistently used in underwriting. The same commentary reported that Equifax in Canada had begun incorporating this data, with TransUnion not far behind. Beginning is the accurate word, and it should not be read as complete.
Missed payments follow a different route from payments made on time. An account referred to a collection agency can reach a credit file through that channel even where the on-time payment record never appeared. FCAC states that once a creditor sends a debt to a collection agency, the credit score will go down. An arrangement invisible while it was being paid can become visible once it is not.
Because the position varies by provider and bureau, the only reliable confirmation is an individual file. You can check your credit score and see what each bureau holds in your name.
Three independent variables produce the inconsistency, and naming them is more durable than listing providers whose practices change.
The first is whether the provider furnishes at all. This is voluntary, and it varies both between providers and by product.
The second is what the receiving bureau does with it. In a 2022 post it has since archived, the United States Consumer Financial Protection Bureau noted diverging approaches: one credit reporting company implemented a business industry code while letting furnishers supply data in their preferred format, and others planned to hold it in specialty files kept apart from the core files behind traditional reports. That account is American and several years old. Canadian bureau practice is not documented publicly in comparable detail. That gap is part of the answer.
The third is whether the scoring model uses the data. TransUnion Canada stated in a 2024 paper that it was analysing alternative data, including buy now, pay later, without initially affecting its scores. Data can sit on a file while remaining absent from the decision. Presence on a record and effect on a score are separate things, a distinction that governs which financial activity does and does not build a credit file.
The foundational federal research on buy now, pay later in Canada is a pilot study, and the agency says so itself.
FCAC surveyed 1,034 Canadians aged 18 and over. The sub-sample of actual users was 66 people, of whom 20 took part in follow-up interviews. The agency states that most findings are drawn from that sub-sample, that these early findings should not be generalised to Canadians at large, and that unweighted percentages are used throughout. Those are appropriate disclosures on a pilot. The difficulty lies with how often it is cited as settled evidence.
Two details matter. The survey reference period ran from September 2019 to March 2021; the report was published in November 2021. And 44 percent of the users surveyed found the potential effect on their credit score difficult to understand: the confusion this article addresses was documented at the outset. Interview participants described using these plans to bridge a timing gap, wanting to purchase immediately while knowing funds would arrive later.
FCAC identified risks of over-borrowing and over-indebtedness but stopped short of recommending regulation, committing instead to monitor the market, conduct follow-up research, coordinate with provincial and territorial authorities, and provide consumer education. As of September 2026, the agency's published research index lists no further study.
Where these obligations are not furnished, or are furnished into files that scoring models do not read, they are absent from any assessment built on bureau data. A household carrying several concurrent instalment plans can present on a credit file as a household carrying none.
The omission runs in both directions. A consumer reviewing their own file may conclude they carry less than they do. And every party that assesses affordability from bureau data, from banks and credit unions to licensed Canadian lenders, works from a record that omits a category of live obligation. TransUnion Canada listed this as a market concern in 2024, noting that limited reporting constrains the ability of other lenders to conduct credit checks and assess affordability.
This is neither new nor specific to one product. Rent, utilities and telecommunications payments are largely unreported in Canada as well. Buy now, pay later is a recent addition to a longer list of obligations that credit files do not capture. The observation concerns what the record contains, not what any party ought to do about it.
On 23 June 2025, FICO announced two scoring models, FICO Score 10 BNPL and FICO Score 10 T BNPL, built to incorporate buy now, pay later data. The announcement was framed explicitly around the United States credit ecosystem, and FICO stated the models would initially be offered alongside its existing scores rather than replacing them, leaving adoption to individual lenders. No equivalent Canadian scoring model has been announced.
The Canadian Lenders Association, an industry body representing lenders, has argued that the sector needs a consistent framework so that this data supports credit inclusion rather than working against it. The position is reasonable and worth reporting. It is not a neutral one.
The effect of fuller reporting would run in two directions. For a consumer with a thin file, a furnished record of payments made on time would constitute history where none existed. For a consumer carrying several concurrent plans, the same reporting would make visible an obligation load that had gone unobserved. Which effect applies is a matter of individual circumstance.
The question a reader arrives with is whether buy now, pay later touches their credit file. The accurate answer is that it depends on the provider, on the bureau, on whether a payment was missed, and on the month in which the question is asked.
That is unsatisfying, and it is not a hedge. Furnishing is voluntary and partial. Bureau treatment differs and is not documented publicly in Canada at the level of detail the question deserves. Scoring treatment is a separate matter again. The Canadian federal evidence base remains a pilot study of 66 users describing behaviour from a period that ended in March 2021.
Each of those conditions can change without announcement. This article describes the position as of September 2026. A reader returning to the question in a year should expect a different answer.
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |