Cordrey The Journal

Writing on AI,
automation, and operations

Showing 43 articles

ChatGPT Work vs Claude Duet vs Microsoft Copilot: which work assistant fits a regulated firm

ChatGPT Work and Claude Duet now rival Microsoft Copilot. For a regulated advice firm, where client data already lives matters more than the model.

Read article

Which AI agent actions need human sign-off? Three questions that decide it

Which AI agent actions can run alone and which need a human? Three questions decide: can it be undone, whose reputation is at stake, does a regulator care?

Read article

The one question to ask an AI vendor: what is still worth paying for when the model improves?

Microsoft's own study found the base model did most of the work. One procurement question shows which AI tools keep their value, and how to price the rest.

Read article

Your IT supplier was breached by AI agents: five questions regulated firms should ask

Attackers used AI agents to breach a managed IT provider and thousands of its clients at once. Five questions to put to your IT supplier, mapped to DORA.

Read article

Open-source or enterprise AI: which should handle your clients' sensitive data?

Self-hosted open models keep client data in-house but make you the support desk; enterprise APIs bring indemnity but send data off-site. How to choose.

Read article

How to audit your advice firm for AI hallucinations before they reach a client file

AI tools report usage, never error rates. A four-step hallucination audit for advice firms: map touchpoints, set checks, keep a log, and name an owner.

Read article

Why small advice firms should automate admin now, and where to start

Intelliflo puts admin at 40 adviser hours per 100 clients. What automation can realistically take off a small UK advice firm, and the first five steps.

Read article

Can an advice firm rely on AI vendors to self-regulate safety? Nvidia says yes, the FCA won't

Jensen Huang says AI safety belongs to tech firms, not regulators. Why vendor self-governance leaves FCA-regulated advice firms holding the risk.

Read article

Using an AI assistant on WhatsApp Business in an advice firm: what clients get

An AI agent on WhatsApp can answer routine client queries in seconds. What it should and shouldn't do, what Consumer Duty expects, and the data question.

Read article

What the AI slowdown petition means for an adviser's Consumer Duty evidence

AI lab staff asked the US government to slow frontier AI. Why that weakens vendor safety assurances under Consumer Duty, and three controls to add.

Read article

How to run an AI proof of concept that reaches production

Only 25% of AI agent projects reach full production. How to design a pilot in a regulated firm that ends in a clear decision, not a forgotten prototype.

Read article

What AI vendor contracts say about your client data, and what to check before signing

Training rights, retention, sub-processors and data location: the four AI contract clauses that matter to a regulated advice firm, and five checks.

Read article

Should an advice firm switch to Claude Fable 5.1? Cheaper inference, looser guardrails

Fable 5.1 promises 25 to 45% cheaper agentic workloads by relaxing safety filters. Which advice-firm workflows can take that trade, and six checks first.

Read article

OpenAI slows Astra over security risks: what UK advice firms should check in their own AI

OpenAI slowed its Astra agent over cyberattack risks it couldn't control. Why advice firms' real exposure sits in the data layer, and three checks to run.

Read article

Will OpenAI's Jalapeño chip make AI cheaper for advice firms?

OpenAI's Jalapeño inference chip should push API prices down. What cheaper inference changes for advice firms, and the data question it leaves open.

Read article

What ESMA's €2.1m Moody's fine means for AI-driven decisions in financial services

ESMA fined Moody's Germany €2.1m for misreporting. What the EU AI Act asks of high-risk AI, and four governance steps for advice and wealth firms.

Read article

How to manage AI agents in an advice firm: briefs, review and accountability

When AI agents do the drafting, the manager's job becomes the brief and the review. Five controls an advice firm needs before agents touch regulated work.

Read article

How financial advice firms can get value from AI without the governance risk

Most advice firms adopt AI in the wrong order. Where the value is in 2026, the governance mistakes to avoid, and a 90-day plan with tools you already have.

Read article

How to build a DORA vendor risk assessment for third-party AI providers

Standard supplier due diligence misses what makes AI providers risky. The seven areas a DORA-aligned assessment of an AI vendor should cover.

Read article

Best AI tools for a financial advice firm: where the gains are beyond note-taking

87% of advisers use AI for note-taking, but bigger savings sit in reports, onboarding and connected systems. Four questions to ask before choosing a tool.

Read article

What a month of building software with AI agents looks like: inside the software factory

Five launches in 23 days, one person, AI agents doing the building. The six habits behind a software factory, and what they mean for regulated firms.

Read article

The future of automation in financial services under the FCA's existing rules

The FCA won't write AI-specific rules. Here's what that means for advice firms automating onboarding and client work, and where the real risk now sits.

Read article

What the move from rented AI chatbots to AI agents means for wealth firms

The latest AI releases push agents that act across your systems while some firms bring models in-house. What that changes for wealth and advice firms.

Read article

How to implement AI agents in an advice firm without losing 6.4 hours a week to them

Staff at firms using AI agents lose 6.4 hours a week supervising them. Six steps to roll out agents in an advice firm so they cut work, not add to it.

Read article

Why CEO-led AI is three times more likely to show ROI, and what it means for advice firms

KPMG found firms whose CEO owns AI strategy are 3x as likely to report established ROI. What that does and doesn't mean for IFA and wealth firm principals.

Read article

Why modular infrastructure is a trap for fintechs

Parker Card's collapse shows what happens when every layer of a fintech is rented from vendors with no duty to stay. The lesson for any regulated firm.

Read article

What Anthropic's Fable access dispute shows about AI vendor risk for regulated firms

Claude Fable was cut off for weeks in a US government dispute. Why relying on one AI provider is now a third-party risk for financial firms.

Read article

What custom AI chips like OpenAI's Jalapeno mean for small regulated firms

OpenAI's own chip should make AI cheaper to run within a few years. It also puts more client data in one company's hands. What that means under UK GDPR.

Read article

The new rules of fintech marketing, and what they mean for advice firms

Fintech products have converged, so brand and trust decide who gets remembered. What the Fintech Insider panel said, applied to advice and wealth firms.

Read article

What agentic AI means for financial services firms in 2026

AI agents that act on their own are reaching advice and wealth firms faster than governance. Why most pilots stall, and where the human sign-off must sit.

Read article

Should financial services firms run AI locally? Lessons from the Fable restriction

A government pulled a live AI model from use overnight. The continuity risk for regulated firms, and four levels of local AI deployment compared.

Read article

Why automating a bad process makes everything worse

Automation copies a broken workflow faster and at scale. Why regulated firms should fix the process first, and the five answers to have before building.

Read article

What the next wave of enterprise AI means for professional services firms

OpenAI and Microsoft's June 2026 news points to dearer compute and task-tuned models. What that changes for advice firms choosing AI tools and vendors.

Read article

The AI token shortage begins: why flat-rate AI subscriptions are ending

AI providers are swapping flat fees for usage-based billing. Why that makes AI a variable cost for advice firms, and why it also hands them an audit trail.

Read article

Why UK financial services firms are reluctant to adopt AI

Regulatory uncertainty, cyber liability and a skills gap keep many UK advice firms cautious about AI. Which concerns are fair, and what waiting now costs.

Read article

How AI is used in managing your investments, and what to ask your adviser

Where AI shows up in portfolio management, its real benefits and risks, and four plain questions to ask your financial adviser at your next review meeting.

Read article

What the FCA and Bank of England joint statement on frontier AI means for advice firms

The FCA, Bank of England and HM Treasury say frontier AI has outpaced cyber defences. Why that reaches small advice firms through their suppliers.

Read article

Why most AI pilots in advice firms never make it to production

AI pilots in advice firms usually stall for reasons unrelated to the technology. The sequencing mistake behind it, and a three-part test for a first build.

Read article

Why only 12 percent of firms using AI tools see real business impact

KPMG research on 1.4 million AI interactions found only 12% of firms see real impact. Three decisions set them apart, from whole workflows to one owner.

Read article

What deepfake fraud means for FCA-regulated firms and how to protect client verification workflows

Deepfake voice fraud attempts rose 300% in a year. How advice firms can redesign phone verification with a second gate for high-risk client instructions.

Read article

What does an automation audit involve? What gets reviewed, what you get, and a DIY version

What happens in an automation audit, what the report should contain, how long it takes, and a version you can run yourself in under an hour.

Read article

Why AI pilots stall at scale: the data is the bottleneck, and the model rarely is

AI pilots work because someone cleans the data by hand. Why fragmented records, not immature models, stop AI projects in advice and professional firms.

Read article

Do you need a custom AI build? Usually you already have the tool

Most AI problems a firm brings in are solved by tools it already pays for. How to tell whether you need training, an integration or a real custom build.

Read article

Schedule an obligation-free call

The best first step is a free 30-minute call. I will ask about your operations, you can ask how I work, and we will both know if it makes sense to go further.

Book a discovery call

Get the weekly journal

One email a week on AI and automation for financial services firms. No spam, unsubscribe anytime.

Got a question instead?