Ask most brokers what worries them about AI, and cost will not be the first answer you get. Data protection, compliance, regulatory scrutiny – those are the big concerns, and rightly so. But if you ask me which of the four pillars in our AI Charter is going to bite brokers and broker firms hardest, and soonest, cost is at the top of the list.
It is also the pillar almost nobody is discussing honestly with brokers... until now.
What brokers need to understand about how these tools are actually priced
A significant volume of the "AI-powered" tools being sold to brokers right now are not running on their provider's own AI. They are running on very large, general purpose models built by OpenAI, Google or Anthropic, with a layer of mortgage-type branding on top. The vendor you are paying is, in turn, paying a third party for every query, every token, every output your firm generates.
What does this mean for brokers? It means you are sitting at the end of a pricing chain you cannot see or control.
There are two cost risks brokers need to understand.
The first is obvious. As your firm's usage grows, the cost grows with it. If your provider is dependent on a third-party model, they have no way of controlling those costs. You have no way of controlling them either. You are both exposed.
The second is more fundamental, and almost nobody is telling brokers about it. The price firms are paying for AI today is artificial. The major model providers are running at significant losses. The token economics the tech industry is building its business cases around are not sustainable. Some analysts forecast price increases of 10 to 30 times current rates as providers are pushed toward commercial viability. That is not a fringe view. It is simple arithmetic.
How could this impact brokers? Very directly. A tool your firm signs up to today at a comfortable monthly fee could cost several times that within twelve to twenty-four months. Not because the tool has improved. Not because you are using more of it. But because the provider behind your provider has changed its pricing.
Broker firms committing to third-party AI infrastructure today are making a long-term bet on a short-term price.
Why this matters more for brokers than for other industries
Every business has to manage its cost base. Brokers and broker firms are managing theirs under particular pressure.
Consumer Duty has raised the bar on demonstrating fair value to clients. Margins across intermediary businesses remain tight. And unlike other sectors, brokers cannot simply pass unexpected technology cost increases through to clients. Regulatory expectations on fair pricing and good consumer outcomes sit squarely in the way.
What does this mean for broker firms in practice? If a core technology in your advice journey suddenly costs materially more, your options are not good ones:
• Absorb the cost and take the margin hit.
• Pass it on to clients and invite a Consumer Duty challenge.
• Rip the tool out and re-engineer your process mid-flight.
None of those is a position a valued broker firm should find itself in.
The brokers best insulated from that scenario are the ones asking hard questions of their providers today, before the pricing environment shifts.
The jet engine and the bicycle: why brokers are paying for power they don't need
There is a common assumption that more powerful AI is better AI. For brokers, the opposite is often true.
Most mortgage tasks do not need the horsepower of a general purpose frontier model. They need a well-designed, focused system built for the job. Running a structured mortgage task through a huge general purpose AI model is, bluntly, a jet engine powering a bicycle. Smaller, purpose-built systems are more accurate, more consistent, and significantly cheaper to run.
What does this mean for brokers? It means that every time your AI tool fires up a large general purpose model to do a task that a smaller, focused system could handle, you are paying premium prices for premium computation you do not need. That cost is being baked into what your firm pays, every month.
Because Mortgage Brain builds and runs its own AI, we can make those trade-offs deliberately for our broker customers. We use proven foundations from the global AI community, then tailor them specifically for mortgage use, building and deploying smaller, focused systems for the specific jobs they perform.
The commercial consequences for brokers are straightforward: lower running costs, predictable pricing, and no exposure to sudden usage charge increases from an external provider.
Companies routing everything through OpenAI or Google are exposed to pricing decisions entirely outside their control, and by extension, outside their broker customers' control. Today and in the not too distant future.
What brokers should be asking their providers
If cost predictability matters to your firm, and it should, here are the questions every broker and broker firm should be putting to their AI or technology provider before signing, or at your next renewal conversation.
Is your AI built and run in-house, or are you reselling access to a third-party model? This is the foundational question every broker should ask first. If the answer is "we use OpenAI, Anthropic or Google under the bonnet", everything that follows about pricing stability needs to be tested against that reality. A provider reselling third-party AI is, by definition, a pricing middleman, and your firm is paying for that middle layer.
If a third-party model provider increases its prices, how will that affect what my firm pays? Get this in writing. Ask specifically whether increases are passed through in full, in part, or absorbed. Many contracts contain pass-through clauses brokers do not realise they have agreed to.
Can you commit to predictable pricing over the term of our contract? A provider with genuine control over its infrastructure can commit to stable pricing for its broker customers. A provider reliant on third-party AI typically cannot. Honestly, it should not.
What happens to our costs if our firm's usage increases? A lot of AI tools are priced on a per-query or per-token basis under the hood, even when the topline pricing looks like a flat monthly fee. Ask where the ceiling is, and whether there are usage-based escalators in the contract that could catch your firm out as your business grows.
Are you using the right size of AI for the job, or defaulting to the most powerful model available? A thoughtful provider will be able to explain which tasks use AI, what type of AI, and why. A provider that cannot articulate this is almost certainly running everything through a large general purpose model by default. Your firm is paying premium prices for tasks that do not require premium computation.
Can you show me how your pricing has moved over the past twelve to twenty four months? Historical pricing stability is the best predictor of future pricing stability. A provider who has had to adjust pricing repeatedly because of upstream changes is telling brokers something important about the resilience of their model.
The bottom line for brokers
AI in mortgage technology should not be a cost time bomb sitting inside your broker firm's P&L. It should be infrastructure. Predictable, sustainable, and under the control of the people ultimately accountable for the client journey, which, in our industry, is you.
The broker firms treating cost as a serious, fundamental question in their AI procurement process, alongside data security and regulatory fit, will be the ones best positioned when the wider pricing environment shifts. And it will shift. The only question is whether your firm sees it coming or finds out at your next renewal.
Cost is the first pillar of our AI Charter for a reason. Brokers deserve to know what they are really signing up to.
