Services

AI agents

AI agent development: agents that work inside your processes, not only in a demo.

Gorilli is an AI agent development company in Milan that designs and builds custom AI agents connected to the documents, data and systems you already use. An AI agent understands a request in plain language, decides which tools to use and takes real actions such as opening tickets, managing bookings or updating a CRM, with permissions, checks and a handoff to a person when needed.

What you get

AI agent development on solid technical foundations.

  • Customer support agents that answer from company documentation, cite sources and hand over to a person when unsure, as Kommander does on websites, WhatsApp Business and Slack in 15+ languages
  • Operational agents that take actions in existing systems: tickets, bookings, CRM updates and internal workflows, with approval for sensitive operations
  • Internal assistants for HR, IT and Sales with central access to company knowledge and each user's permissions
  • MCP tools and integrations that let agents use APIs, databases and SaaS platforms without rewriting the backend

How we work

  1. 01

    Discovery: the tasks the agent must complete, the data and systems involved, the risks and the success metrics

  2. 02

    Design of tools, permissions and agent flow, choosing where an agent is needed and where a workflow with a few well-bounded LLM calls is enough

  3. 03

    Read-only prototype: the agent answers and proposes actions on real data, measured on a set of conversations and edge cases

  4. 04

    Actions are enabled one at a time, with human approval for sensitive ones, a log of every tool call and spending limits

Timeline

  1. Discovery and task selection

    1-3 weeks

  2. Read-only prototype on real data

    2-6 weeks

  3. First actions in your systems, with approval

    1-3 months

  4. New tasks, channels and languages

    Ongoing

Typical budget

  • Simple agent

    Up to about €20,000

    Answers questions from documentation on one channel, without acting in your systems. Typical timeline: 3-6 weeks.

  • Operational agent

    €20,000-80,000

    Reads data and takes actions in 2-3 systems (CRM, ticketing, ERP), with permissions and approvals. Typical timeline: 2-4 months.

  • Multi-agent or enterprise system

    Over €80,000, often over €150,000

    Several agents and departments, many integrations, high security and compliance requirements. Typical timeline: 4-9 months.

  • Running costs after launch

    20-40% of the build cost per year

    Models, vector databases, observability and maintenance. Indicative market ranges: the project price is set after discovery.

Why Gorilli

  • Our own AI agents in production: Kommander builds teams of assistants that answer, act and connect to company systems
  • Storm, a marketplace of MCP tools for agents, took 2nd place at the AI Blueprints with Filecoin Recall hackathon
  • One team for AI, backend, frontend and cloud: we also build the software around the agent
  • Pragmatic: we use an agent only when the problem is truly open-ended, otherwise a simpler and cheaper workflow

Guide: Build an AI product: timeline, costs and FAQ

Comparison

OptionWorks whenMain risk
Gorilli (custom AI agent development)The agent must act in your systems with custom permissions and checks, and you also need the software around itNeeds your team's time for discovery, system access and validating answers
No-code agent platformStandard use cases such as FAQs or basic support, without deep integrationsLimits on permissions, custom integrations and data control
General AI assistant (ChatGPT, Copilot, Gemini)Individual productivity: writing, summarising, searching personal documentsDoes not act in your systems and does not know company permissions and processes
Rule-based automation (RPA or classic workflows)The process is always the same and the data is structuredBreaks on natural-language requests, variable documents and exceptions

Frequently asked questions

How much does it cost to build an AI agent?
As indicative market ranges: up to about €20,000 for a simple agent that answers from documentation, €20,000-80,000 for an operational agent connected to company systems and over €80,000 for multi-agent or enterprise systems. Add running costs of 20-40% of the build cost per year. The first call with Gorilli is free and the price is set after discovery.
How long does it take to build an AI agent?
From 3-6 weeks for a simple agent to 2-4 months for an operational agent with integrations, and 4-9 months for multi-agent systems released in phases. Every project starts with 1-3 weeks of discovery.
How do I choose an AI agent development company?
Check that the team has agents in production, can tell you when an agent is not needed, treats evaluation, permissions and observability as part of the project and can integrate the agent with the APIs, databases and tools you already use.
When do I need an AI agent and when is a workflow enough?
You need an agent when every request is different and the system must decide which tools to use and in what order. If the steps are always the same, a workflow with a few model calls is cheaper, faster and easier to control. We say so in discovery, even when it means a smaller project.
How do you stop an agent from causing damage?
The agent acts only with the permissions of the user it works for and only through tools with validated inputs, never with direct database access. Sensitive actions such as sending emails, changing orders or touching financial data need human approval. Every tool call is logged, and content read from documents or emails is treated as data, not instructions, to limit prompt injection.

Next step

Tell us what you want to build.

We help you understand scope, technical risks and the fastest path to a product people can use. The first call is free.

AI Agent Development Company | Gorilli