Azimut AI
Illustrative scenario. The company on this page doesn't exist: it's a typical profile for the sector, built to explain how Azimut would be used. It doesn't describe any customer or claim measured results.
The company in this scenario
- 140 people: 110 advisors in tax, accounting and payroll, and 30 in support and front office.
- A client base of SMEs and professionals, with workload peaks at every quarterly close.
- Industry accounting, payroll and tax software; documentation in Drive or SharePoint; email.
- Internal positions scattered across circulars, technical notes and the partners' heads.
The problem
Most of an advisory firm's work is answering: whether an expense is deductible, how a regulatory change applies, which contract makes sense. Each advisor searches their notes, asks a colleague or a partner and writes from scratch, and two advisors can answer the same question differently. Every regulatory change means manually checking which clients it affects. And at the quarterly-close peaks, questions wait.
How it would work with Azimut
Azimut would connect to email and to the repositories where the firm's judgment lives (circulars, technical notes, standard answers) with each area's permissions. The firm would publish four agents.
| Agent | What it does | Who uses it | What gets measured |
|---|---|---|---|
| Client questions | Prepares the answer to a client's question with the firm's internal position and the cited regulation, for the advisor to review. | Advisors | Response time and questions per advisor. |
| Regulatory updates | Summarizes each change published in the Official State Gazette or by the Tax Agency and flags which type of client it affects. | Partners and area leads | Days from publication to client notice. |
| Meeting prep | Pulls the client's email history, documents and open issues onto one page before each meeting. | Advisors and partners | Prep hours. |
| Firm manual | Answers newcomers' questions about internal procedures, templates and the firm's positions. | New hires | Weeks until working independently. |
Client questions call for a mid-tier or top-capability model depending on complexity; regulatory summaries work well with a mid-tier one. Advisors pick the model per task from OpenAI, Anthropic, Google and Mistral.
Where you'd start
You'd start with the firm manual and, right after, client questions. The first forces you to gather in one place the internal positions that are scattered today; without that, the second would answer with generic text. With the firm's judgment on the platform, the client questions agent applies it from the first week.
What's out of scope
There's no connector for the industry's accounting, payroll or tax software: Azimut doesn't read a client's books or file returns. It works with the documents, emails and exports the firm already produces. Every answer is reviewed and signed by an advisor. Integrations are confirmed during rollout and only the tested ones get activated.
How it would be measured
Before turning on any agent, you set the baseline, with the company's own numbers rather than industry averages.
Week 0. Average response time to a question on a recent sample, questions resolved per advisor per week in a normal period and at quarterly close, and current AI use: who, with which tool and with which client data.
Month 3. The same numbers, next to Azimut's usage panel by user, area, model and tool.
Frequently asked questions
Can it look at a client's accounts?
Not directly: there's no connector for industry software. It can work with the documents and exports the firm already keeps in its repositories.
What if the agent's answer is wrong?
The agent prepares a draft with the regulation cited; an advisor reviews and signs it before it reaches the client. Responsibility stays with the firm.
Does it work for the payroll and employment area?
For answering clients' employment questions and preparing documentation, yes. It isn't used to evaluate or select job candidates: that's a high-risk case under the EU AI Act and it's excluded.