AI & Machine Learning · Lyzr + Claude
Lyzr MCP with Claude: Build, Test and Manage AI Agents Through Natural Language
The short answer
Claude works with Lyzr through the Model Context Protocol (MCP), an open standard that lets Claude securely use software on your behalf. Once connected with an API key, Claude can help build agents, run evaluations, manage knowledge bases and coordinate workflows through natural language, with your team approving key changes.
What is Lyzr?
Lyzr is an AI agent platform for building, running, evaluating and managing AI agents. It includes Studio services for memory, voice, simulations and knowledge bases, helping teams ship AI workflows more safely. Product, engineering and operations teams use it to create chatbots and task-focused agents, test them before release and keep them running well. Lyzr offers both an MCP server and a direct API for connection.
How Claude connects to Lyzr through MCP
The Model Context Protocol, or MCP, is an open standard that acts like a universal adapter between Claude and other software. It lets Claude read information and take actions in a tool, using permissions your organisation controls. Instead of copying data by hand, your team can describe what they need in ordinary language.
Once Claude is connected to Lyzr, it can typically look at your existing agents, review how they are set up, start evaluation runs, and check knowledge bases for gaps. It can also help organise a sequence of steps, such as updating an agent and then running a simulation to test the change.
Permissions and approvals keep people in charge. We connect Claude using your API key with least-privilege access, so it only reaches the parts of Lyzr you allow. Consequential actions, such as publishing an agent or changing a live knowledge base, can require a named person to approve them first, and every action is logged.
What you can automate
Agent Build Assistance
Product teams can describe a new assistant in plain language, such as a support bot for billing questions. Claude can help draft the agent configuration in Lyzr, suggest the tools and knowledge it needs and prepare it for review. A developer or product owner then checks the setup before anything goes live.
Automated Evaluation Runs
Quality teams often test agents by hand, which is slow and inconsistent. Claude can start evaluation runs against a set of realistic questions, summarise the results and highlight answers that failed or drifted. This gives reviewers a clear list of problems to fix before each release.
Knowledge Base Upkeep
When policies, price lists or product manuals change, the knowledge behind an agent can fall out of date. Claude can review the knowledge base, flag outdated documents and prepare updates for approval. This helps keep answers accurate without a manual audit every month.
Simulation and Scenario Testing
Before launching a voice or chat agent to customers, teams need to see how it handles difficult conversations. Claude can help set up simulations covering common scenarios and edge cases, then report where the agent struggled. Results feed directly into the next round of improvements.
Operations Reporting
Operations leaders want to know how their agents are performing without logging into several screens. Claude can gather run results, usage patterns and evaluation scores into a short weekly summary. Managers receive a readable report they can act on in minutes.
An end-to-end workflow: Customer Support Agent Release Cycle
TriggerA support lead requests an update to the returns-policy agent after a policy change.
Claude reads the current returns-policy agent and its linked knowledge base in Lyzr.
Claude drafts the knowledge changes and flags any documents that still describe the old policy.
The support lead approves the changes, and Claude runs an evaluation set of past customer questions.
Claude compares the new results with the previous scores and lists any answers that now fall short.
Claude posts the summary to the support team channel in Microsoft 365 and creates a review task in the ticketing system for remaining issues.
The agent is published only after a named reviewer signs off.
OutcomePolicy updates reach customers faster, with a documented test record and a clear approval trail for every release.
How we set it up
Free Automation Audit
We review your current agents, support processes and reporting needs to find the workflows worth automating with Lyzr. You receive a short plan showing where Claude adds the most value, at no cost.
Secure API Key Connection
We connect Claude to Lyzr using an API key stored in your secure credential manager. Access is limited to the permissions your team agrees to, and the key can be rotated or revoked at any time.
Workflow Design
We design the end-to-end workflow across Lyzr and your other systems, such as Microsoft 365, CRM, ERP or ticketing. Each step that changes something live is marked for human approval.
Testing and Launch
We test the workflow against real past cases, fix weak points and launch with monitoring and audit logs turned on. Results are reviewed with your team before the workflow is trusted with live work.
Team Training
We train your staff to use Claude with Lyzr confidently, covering how to give clear instructions, review outputs and handle approvals. We also provide written guidance so the workflow can be maintained after handover.
Who benefits
- Head of Product
- Ships new agents faster, with each release supported by a documented evaluation record.
- Quality Assurance Lead
- Replaces slow manual testing with consistent evaluation runs and clear reports on failures.
- Customer Support Manager
- Keeps agent answers aligned with current policies without waiting on a monthly audit.
- Chief Information Officer
- Gains a controlled, logged way for AI agents to be managed without widening system access.
Security and control
- Claude connects to Lyzr with an API key and least-privilege permissions, limited to what your team approves.
- Consequential changes, such as publishing agents or editing live knowledge bases, require named human approval.
- Every action Claude takes is recorded in audit logs that your team can review.
- Evaluation results are checked by people before any agent is trusted with customer-facing work.
- Keys can be rotated or revoked at any time, and access can be withdrawn without affecting other systems.
For teams in the US, UK, India and the Middle East
Organisations in the United States and the United Kingdom can host their Claude workflows in the region that suits their compliance needs. In India, we design workflows to respect the Digital Personal Data Protection Act, 2023. For organisations in the Middle East, including the UAE and Saudi Arabia, we align data handling with local data protection laws. Hosting region is chosen to meet your specific requirements.
Learn more about Claude and MCP
Anthropic's AI model and assistant.
Secure links between Claude and business tools.
How engineers build and change software with Claude.
Hosted agents that run multi-step work.
The open standard Claude uses to connect to tools.
Frequently asked questions
What does the Lyzr MCP server do?
The Lyzr MCP server acts as a bridge between Lyzr and tools like Claude. It lets an AI assistant read agent configurations, start evaluations and manage knowledge bases through standard requests. Your team keeps control over which actions are permitted, and the server does not replace the Lyzr platform itself.
Is connecting Claude to Lyzr secure?
Security depends on how the connection is set up. We use an API key with least-privilege access, store it in a secure credential manager and limit Claude to the actions you approve. Consequential changes require human sign-off, and audit logs record every action for review.
What can Claude actually do in Lyzr?
Typically, Claude can review existing agents, help draft agent setups, start evaluation runs, summarise results and check knowledge bases for outdated content. It can also coordinate multi-step workflows. Final decisions on publishing or changing live content remain with your people.
How long does setup take?
Timelines vary with the size of your agent estate and the workflows involved. A focused single workflow is often connected and tested within a few weeks, following the free audit. Larger programmes covering several agents and systems take longer and are scheduled in phases.
What plan or permissions do I need?
You need an API key from your Lyzr account with permissions that match the actions you want Claude to take. Generally, we recommend starting with read access and adding write permissions only for specific, approved steps. Confirm your account's options with Lyzr, and we will advise on the right permission set during the audit.