Different answers to the same question
The policy is in a PDF, the exception in an email and the new rule in the supervisor's head. Each agent answers with whatever they found.
Memory for Customer Service
Kluco brings the service team's policies, procedures and answers into a reviewed knowledge base, with an owner and a review deadline. AI agents serve customers from it, and the team checks the same source.
The problem
When the answer depends on who picks up, the customer notices.
The policy is in a PDF, the exception in an email and the new rule in the supervisor's head. Each agent answers with whatever they found.
New joiners take a long time to answer on their own, and experienced people stop what they are doing to help.
A bot that answers without the company's knowledge base invents deadlines, promises what it cannot, and nobody knows where the answer came from.
Before and after
From the first contact to the day's summary.
| Without the company's memory | With Kluco |
|---|---|
| Without the company's memoryThe right policy is somewhere in the drive | With KlucoThe current policy is in the knowledge base, with an owner and the date of its last verification |
| Without the company's memoryThe bot answers without saying where it got it | With KlucoThe agent answers from the knowledge base, and the passages consulted are recorded |
| Without the company's memoryA new question gets lost in the chat | With KlucoAn unanswered question goes into a queue and becomes an article |
| Without the company's memoryNew joiners learn by asking a colleague | With KlucoNew joiners ask the knowledge base and get the same answer as the team |
| Without the company's memoryThe leadership builds the day's summary by hand | With KlucoAn agent delivers the day's summary with topics, open items and conversations at risk |
What goes into memory
Documents come in from Google Drive, OneDrive, SharePoint, file upload, API or webhook, and resolved customer service conversations also teach the knowledge base, with personal data masked and review before publishing.
Use cases
The most common use cases in customer service, with the part of the platform that does each one.
The agent serves customers on WhatsApp and the other channels with what is in the reviewed knowledge base, and hands over to a person what it does not know.
AgentsWhat the knowledge base could not answer goes into a queue. Someone answers it once, and the answer becomes an article for everyone.
BrainAt the end of the day, an agent puts together the summary for the leadership: volume, topics, open items and conversations at risk.
AgentsTemplates, clauses and guidance sit in the knowledge base with an owner and a review deadline. AI uses the current version and shows where it came from.
BrainWhen the company turns it on, the agent masks personal data in its logs and answers.
AgentsChatGPT, Claude, Gemini and other MCP-compatible tools query the same knowledge base, each person with their own access.
BrainAgents · AI agents
The service agent answers customers with what is in the knowledge base. You set what it can do, who approves and how much it can spend.
Brain · the memory
The Brain turns the service team's documents into articles with an owner and a review deadline. When the deadline passes, the article is flagged.
LLM Management and Chat
The team uses the best models in the Chat, which draws on the company's memory and saves to the Brain what is worth keeping, after review. In LLM Management, you track limits and cost.
Control
The company sets up each control below in the console.
Related
FAQ
Yes. WhatsApp is one of the agents' channels, along with Telegram, Slack, Microsoft Teams and Google Chat. The agent answers with what is in the knowledge base, and you set the rules and approvals.
The question goes into the unanswered questions queue. When someone answers it, the answer can become an article and applies to the team and to the agents from then on.
Every article can have an owner and a review deadline. When the deadline passes, the article is flagged for review. AI also reads the connected sources and proposes updates, which can wait for a person's approval.
Yes. The team asks the knowledge base in the console or in the Chat and gets the answer with its source passages, from the same knowledge base the agents use.
With access profiles, data isolated per company, an audit log and the option to mask personal data in the logs and answers of each agent. The Security page covers each control.
Next step
In the demo, we show the knowledge base, the agent and the unanswered questions queue with customer service examples.