# AI agents for business

Services · AI agents

Assistants that handle repetitive requests in the channels your company already uses. Your team remains in control of decisions.

## What it is

An AI agent works through your company's WhatsApp, email, phone, or internal systems. It interprets a request, retrieves the required information, and performs only the approved actions.

Before launch, we agree on what can be automated, what needs confirmation, and when a person takes over. Relevant actions are logged for review.

## Who it is for

For companies that receive many similar requests through WhatsApp, email, or phone. It works across healthcare, retail, industry, and services when there are clear rules, reliable data, and a team to handle exceptions.

## Requests an agent can handle

- A WhatsApp front desk that books, reschedules, and manages the waitlist, with immediate handover to the team
- Email triage: quote requests identified, classified, and routed with the data already extracted
- Replies using stock, lead times, and prices retrieved from your system
- An internal assistant that answers the team's questions from your company's technical documentation

## How we start

1. **Choose one request**: We start with a frequent task that has visible impact and clear limits.
2. **Define rules and data**: We map the information sources, approved decisions, and human exceptions.
3. **Test with the team**: The first version is validated against real conversations before it takes on daily work.
4. **Measure and adjust**: After launch, we track errors, interventions, and time saved.

## Frequently asked questions

### Do agents replace people?

No. They take repetitive work off the team's plate: answering the same type of request 30 times a day, copying data from emails into the system, replying to the same questions. Decisions stay with your team.

### What if the agent gets it wrong?

We define clear limits for what the agent may do alone and what requires human confirmation. Relevant actions are logged, the team can intervene, and the behavior is tested before production.

### Which data is used and where does it live?

Your company owns its data. Before we build, we document what information is used, where it is stored, and which providers process it. The architecture and safeguards are part of the proposal.

### How long until it is working?

It depends on the integrations and the risk of each decision. The proposal defines the first scope, schedule, and production-readiness criteria.

## Does your team answer the same request every day?

On the first call, we identify one suitable task and the limits an automated system would need to respect.

[Book a call](https://calendar.app.google/wzfT51cBvNT5i3x86) · <joao.santos@funnysoft.pt>

---

FunnySoft · Forward Deployed Engineering · Leiria, Portugal

João Paulo Santos, sole trader. Trade name FunnySoft. Tax number 239506308. Registered office: Leiria, Portugal.

+351 916 439 451 · Call to the national mobile network

If you have a consumer dispute, you can use [CNIACC](https://www.cniacc.pt).

[Electronic complaints book](https://www.livroreclamacoes.pt/Inicio/)

<https://www.funnysoft.pt/en/services/ai-agents>
