Blog
Insights
Practical guidance on IT, security and technology strategy for growing businesses — written by the people who do the work.
How AI actually automates a workflow, and where it shouldn't
Agentic AI is mostly plumbing: retrieval, business rules, approvals and an audit trail, with a model doing one step. Here is what a real AI workflow looks like.
·Foundry Team
Assistants, workflows, agents: what agentic AI means for a 30-person firm
Agentic AI is the buzzword of the year. Here is the plain-English difference between an assistant, a workflow and an agent, and the order to adopt them in.
·Foundry Team
RAG, explained for people who run a business
Retrieval-augmented generation is how an assistant answers from your documents. What it is, what it fixes, and the permissions problem inside it.
·Foundry Team
Shadow AI: someone in your firm is already pasting client data into a chatbot
Consumer AI tools have different terms from business ones. If you haven't given people a sanctioned option, they've found their own. What to do this month.
·Foundry Team
Before you switch on Copilot: the oversharing problem
Copilot respects the permissions you already have, which is good news and bad news. How to find what is overshared in your tenant before an assistant does.
·Foundry Team
Where does your data go when you use AI? It depends on the product
One vendor can sell a consumer app, an enterprise API and a model in your own cloud, each on different terms. Four deployment models, compared honestly.
·Foundry Team
Prompt injection: the security problem every AI workflow inherits
If an AI system reads untrusted text and can act, someone will put instructions in the text. What prompt injection is and how to design around it.
·Foundry Team
An AI policy for a financial advice firm: five things to write down
The FCA regulates AI through its existing rules, so an AI policy is a governance document, not a technology one. Five sections, and what goes in each.
·Foundry Team
8 articles tagged “ai”