
To move from individual use of tools like ChatGPT to AI that is useful for the company, the AI has to work on data from your business systems, respect each user’s permissions and keep a record of what it does. Today that is possible with native interfaces such as Zoho’s Zia Chat, with connectors based on the MCP protocol, or with your own layer such as IC Agents. The first step is not choosing the tool, but defining which data, permissions and processes will be included.
What the data shows in Panama
Using AI at work is already an everyday practice in Panama, but it happens mostly at the individual level:
- According to the 2026 edition of Konzerta’s AI at Work study, 44% of workers in Panama use AI in their daily tasks, up from 40% in 2025. Even so, the country ranks below Ecuador (45%), Argentina (57%), Chile (61%) and Peru (65%).
- The regional study by CCK Centroamérica and Datalex Latam indicates that 97.7% of Panamanians surveyed are familiar with AI tools and that more than 60% frequently use systems such as ChatGPT, Gemini or Copilot to support their work.
- Among the benefits workers highlight most in the Konzerta study are time savings (54%) and faster task completion (53%).
In other words: personal adoption is moving forward. What many companies have not yet solved is how to turn that individual use into an organizational capability.
The limit of AI that works “outside” your systems
This analysis comes from InterConnecta, based on what we see in projects with companies across the region. When each person uses an AI assistant on their own, three limits appear:
| Limit | What happens in practice |
|---|---|
| No business context | The assistant doesn’t know what stage a customer is in, which invoices are outstanding or what was discussed with support. It answers well in general and poorly on the specifics. |
| Data copied by hand | To get a useful answer, employees paste company information into the tool. The company loses visibility over what data leaves and where it goes. |
| No traceability | There is no record of what was asked, with what data, or what decision was made based on the answer. |
None of this means banning the tools. It means bringing AI to where the company’s data and rules already live.
Three ways to connect AI to your company’s systems
1. Native interfaces inside the platform
On September 23, 2026, Zoho announced Zia Chat, a conversational interface for finding information, analyzing data and starting tasks across the Zoho applications each user can already access. Zoho says it respects app permissions, that answers can be traced back to the source records, and that the data it accesses is not used to train its models. Early access was initially announced in English.
2. Connectors using the MCP protocol
The Model Context Protocol (MCP) lets an external assistant, such as Claude or ChatGPT, use actions in the company’s systems in a controlled way. Zoho offers an MCP server with actions from more than 15 applications that respects its permission structure, and Zoho CRM has pre-built MCP servers, including a read-only one.
3. Your own agent layer on top of the data
When a company needs specific rules, needs to combine several sources or wants to control the channel (text, voice, mobile), it can build an agent layer on top of its Zoho data. That’s the case we describe below.
In an InterConnecta implementation with IC Agents, we made it possible to query company information by text or voice, even from a mobile phone, using real data connected to Zoho: sales and pipeline in the CRM, invoicing and payments, support tickets and SLAs, projects and operations.
What set it apart from an open tool was control: role-based access, full traceability of every interaction, data queried directly from the systems, encrypted transmission through APIs, and processing in AI environments with policies that prevent the data from being used for training.
The main lesson: the value wasn’t in the conversational interface, but in the fact that the conversation happened on reliable data and within the organization’s rules.
A four-step roadmap
This is our recommendation for companies that want to move from individual use to AI at the organizational level:
- Take inventory of current use. Ask your teams which AI tools they use, for which tasks and with what data. That map shows where the real demand is.
- Choose a process with data that’s already structured. Pipeline questions, order status or support tickets are usually good starting points because the information already lives in the CRM or the help desk.
- Define permissions and approvals before connecting. Review who sees what today and which actions will require human approval.
- Measure against a baseline. For example, how long it takes to answer a recurring management question, or how many manual reports a team stops producing.
Frequently asked questions
Is it risky for employees to use ChatGPT with company information?
It can be, if they paste customer data or confidential information without clear rules, because the company loses visibility over what data leaves and for what purpose. The recommended approach is to define a usage policy and offer an alternative connected to your systems, with permissions and logging.
What is Zia Chat?
It’s a conversational interface Zoho announced in September 2026 for finding information, analyzing data and starting tasks across Zoho applications, while respecting each user’s permissions.
What is MCP and what is it used for in a company?
MCP (Model Context Protocol) is a protocol that lets an AI assistant use actions and data from other systems in a controlled way. In Zoho, it allows tools such as Claude or ChatGPT to work with the CRM and other applications while respecting permissions.
Where is the best place to start?
With a process where the data is already in a system, the question comes up often and the result can be measured, such as sales queries or ticket status.
La Estrella de Panamá — Konzerta AI at Work study 2026 (in Spanish)
Revista E&N — CCK Centroamérica and Datalex Latam study (in Spanish)
ThePrint (PTI) — Zia Chat launch
DestinationCRM — Zia Agents and Zoho’s MCP server
Zoho CRM Developer Docs — Zoho CRM MCP Servers
Is your team already using AI on its own?
We help you bring it inside Zoho and the rest of your systems, with permissions, traceability and a measurable first use case.
Let’s talk