Beyond chatbots: the rise of AI workflows

Image

Julien Gruhier, head of AI at Croner Intelligence, explains how AI workflows can handle entire tasks from initial information gathering and template selection to producing a finished draft ready for review

AI first arrived in professional practice as something you asked questions. You entered a query about a treatment or a transaction, received an explanation, and then decided what to do with it.

AI workflows are the next step. Rather than waiting for instructions, the tool follows a defined piece of work through a series of stages and produces the finished document at the end.

Most routine jobs in practice follow a familiar pattern. First, establish what is being asked, then identify the appropriate document, gather the necessary facts, prepare a draft, review it and produce the final version. A workflow makes that sequence explicit, guiding the AI through each stage while allowing the practitioner to review and confirm progress along the way.

A practical example

Correspondence is often the first area accountancy firms choose to automate, and in Croner Intelligence the initial workflows focus on client letters to HMRC.

The process begins with a short description of the client's issue. Before drafting anything, the workflow asks a series of targeted questions. Each answer narrows the options until the workflow identifies the most appropriate document.

Once the correct document has been identified, the workflow completes the draft using the information provided, together with any supporting material that has been uploaded, such as previous correspondence or calculations.

Where information is missing, the workflow asks for it rather than making assumptions. The completed draft is then produced as a Word document.

Building a workflow you can trust

A workflow is only as reliable as the material it draws upon and the standards it is checked against. Croner Intelligence relies on Croner’s in-house tax specialists to ensure the output is correct. The templates are written and maintained by subject specialists, and the structure and content of each letter reflects established professional practice rather than being generated from scratch.

Once completed, the draft is checked against Croner’s own expert content. The workflow then produces a short report highlighting which statements are supported by the available information and where further facts may still be needed.

An AI assistant that answers questions can save you time spent searching for information. A workflow that carries a task through to completion can save an entire afternoon, provided you can see exactly what information it worked from and how it reached its conclusions.

 

*** Find out more about Croner Intelligence, the game changing AI tool for accountants and tax professionals. Book a demo here ***

Julien Gruhier | Director of search and generative AI, Croner

Julien Gruhier is director of search and generative AI at Croner, and is the driving force behind ...

View profile and articles

0
Be the first to vote

Rate this article

Related Articles
Subscribe