Back to vendors
L

Lyzr

Also known as: Lyzr Cognis, Lyzr SuperFlow, Lyzr Architect

Visit site
Agent builderindependentVerified 2026-09-29
Entry priceFrom $19 a month on Starter, or $99 a month on Pro ($79 billed yearly). Production agent runs add $0.08 per run on Lyzr Cloud or $0.03 on VPC, and LLM tokens are billed by the model provider.Full pricing detail

Enterprise platform and control plane for building, evaluating, deploying and governing AI agents, including agents built on other frameworks.

Lyzr is an enterprise platform for building and running AI agents and moving them from a sandbox into governed production. It serves as the infrastructure and control plane for networks of specialized agents operating across a company's systems, letting organizations deploy, monitor and govern those agents at scale.

Teams build on Lyzr through a low-code Agent Studio and SDK, or through Architect, where a user describes a workflow in plain English and the platform assembles the agent, its logic, its integrations, access controls and a usable interface. A library of pre-built agent blueprints spans common enterprise functions, from sales and marketing to HR, finance, banking, insurance and customer service, and a new agent can start from a blueprint or from scratch. For more complex work, an orchestration layer lets multiple specialized agents collaborate on end-to-end, multi-step processes, sharing context and delegating tasks across departments.

A simulation engine lets teams test an agent before it ships, full observability traces every run in real time, and responsible-AI guardrails check outputs for hallucinations and personally identifiable information before they reach a user. Access and governance controls set who can see and modify each agent, and audit logging and compliance features make every action and decision traceable.

Agents can run on models from different providers, and a model can be swapped without rewriting the agent. Lyzr also deploys in a company's own private cloud or on premises, with locally deployable SDKs and private APIs, and the customer keeps ownership of its data and the agents it builds.

Vendor details

Canonical URL

https://www.lyzr.ai

Category

Agent builder

Subcategory

Full-stack agent platform and multi-framework control plane (Agent Studio, Architect, ADK, SuperFlow)

Funding status

Series A. Lyzr's own site carries press badges citing a $100M fundraise at a $500M valuation and a headline about the round being run by its own agent, while other coverage reports an ~$8M Series A in March 2026 led by Rocketship.vc with Accenture among backers at a reported ~$250M valuation. The two accounts do not reconcile, and no Lyzr funding announcement settles them.

Company status

independent

Use cases & customers

Primary use cases

Building enterprise agents three ways: visually in Agent Studio, in code through the Python or TypeScript ADK, or in natural language through ArchitectGoverning agents built on other frameworks and clouds from one control plane, without migrating or rewriting themEvaluating agents against generated test cases and domain scenario suites before production, with regression re-runs when models or prompts changeRunning the platform inside a customer VPC, on-premises or air-gapped for regulated buyers in banking, insurance, healthcare and governmentDeploying voice agents over telephony carriers using the same knowledge bases, tools and guardrails as text agents

Target customers

enterprisesdevelopers

Deployment options

SaaSself-hostedVPCon-premhybridair-gappedregional hosting

Integrations

Lyzr has two distinct integration surfaces. For tool calling, agents use built in connectors such as Slack and CRM systems, custom REST integrations, and native Model Context Protocol support so any MCP compliant tool plugs in. Tools can send email, query databases or call any custom API, and Architect wires in existing ERP, CRM, HRIS and document systems. Tools are metered per tier at 5, 10 and 25 on Community, Starter and Pro, unlimited on Enterprise. Separately, the control plane registers and governs agents built on other frameworks and clouds without migration, named as LangChain, CrewAI, AutoGen, Agentforce, AWS Bedrock, Azure AI, GCP Vertex and on-prem stacks. Knowledge ingestion covers documents, databases, websites, spreadsheets and live data sources, with vector-store selection and an External Knowledge path. Models route per agent across OpenAI, Anthropic, Gemini, Bedrock and DeepSeek, with Bring Your Own Model on Enterprise. Voice agents deploy over Twilio, Telnyx and Plivo. Outbound, every agent exposes a REST endpoint, and a Python and TypeScript SDK plus raw REST APIs make each module callable as a service.

In practice

Your AI agent works in a demo but stalls before production because of governance and compliance. Lyzr adds a simulation engine, observability, guardrails for hallucinations and PII, and audit logging so the agent can be trusted live.

You want to automate work across HR, finance, and support without building each agent from scratch. Lyzr's blueprint library gives you pre-built agents per function, and an orchestration layer lets them collaborate on multi-step processes.

Regulatory rules mean your data can't leave your environment. Lyzr deploys in your private cloud or on-premises with private APIs, runs on the model of your choice, and lets you keep ownership of your data and agents.

Sources & related URLs

Agentic Index coverage score

12.0 / 14 capabilities · 86%

Integrations & Tool Calling Full

Agents connect through built-in connectors such as Slack and CRM systems or custom REST integrations, with tools able to send email, query databases or call any custom API. Native tool connectors are a built-in module that replaces third-party and custom connectors, and native Model Context Protocol support lets any MCP-compliant tool plug in.

Architect wires in existing ERP, CRM, HRIS and document systems. Separately, agents built on LangChain, CrewAI, AutoGen, Agentforce, Bedrock and Vertex register into the control plane. Tools are capped at 5, 10 and 25 on the Community, Starter and Pro plans and unlimited on Enterprise.

Sourcedocs.lyzr.ai integrations, quickstart and plansread 2026-08-31
Workflow Orchestration Full

Workflows with several agents run with either deterministic orchestration through SuperFlow or dynamic orchestration through a Manager agent, and Lyzr presents Manager Agent plus SuperFlow as replacing external and custom orchestration.

Through the Agent API a manager agent receives a managed_agents array of sub-agents with names and descriptions so it can select the right one for a task, and a Lyzr Agent Orchestrator is part of an AgentMesh. Orchestration is Limited on the Community, Starter and Pro plans and Full on Enterprise. Durable exactly-once execution with clean rollback appears only as a product page claim.

Sourcedocs.lyzr.ai platform intro, manager agent API and plansread 2026-08-31
Knowledge Grounding & RAG Full

Enterprise RAG with secure retrieval is a built-in module that replaces Weaviate, Pinecone and custom pipelines, connecting agents to documents, databases, websites and live data sources.

In Studio a builder attaches PDFs, web URLs, databases or spreadsheets as knowledge sources, the ADK makes RAG a core agent feature with programmatic knowledge-base construction including website ingestion with a page limit, and the REST build path sets a default vector store at project creation.

A Knowledge Graph module sits alongside, and Knowledge Base as a Service is a retrieval layer reachable through the API. Knowledge bases and RAG storage are metered per tier at 5, 10 and 15 bases and 100MB, 100MB and 1GB.

Sourcedocs.lyzr.ai platform intro, ADK, quickstart, API build path and plansread 2026-08-31
Human Oversight & Guardrails Partial

Guardrails for toxicity detection, PII redaction, prompt-injection blocking and hallucination reduction form a Responsible AI block that can block, flag or reroute an output before it reaches a user, alongside a Hallucination Manager, RBAC roles and permissions, and an audit log. These constrain output and access, and Lyzr offers no human approval step, pause for review or escalation queue.

Sourcedocs.lyzr.ai llms.txt index and platform overviewread 2026-09-29
Security, Identity & Governance Full

Teams are managed with role-based access control, audit logs and SSO/SAML, and Access and Governance is a built-in module covering roles and permissions, alongside cost controls with usage and budgets.

Lyzr says the platform is HIPAA and SOC 2 compliant, and adds RBAC scoped at the agent, tool and data level, SSO against the customer's own identity provider, explicit reviewable revocable permissions, immutable timestamped audit logs on every agent decision and tool call with one-click compliance export, an Agent Entitlement Protocol, and badges for SOC 2 Type II, ISO 27001, GDPR, HIPAA and CCPA. Lyzr does not point to a trust portal, named auditor or way to request a report.

Sourcedocs.lyzr.ai platform intro, lyzr.ai control plane and compliance sectionsread 2026-08-31
Observability & Auditability Full

Every agent run is traced step by step with latency metrics for each step, and Full Observability is a built-in module that replaces external tracing tools. Audit logs sit alongside role-based access control and SSO, monitoring and analytics endpoints return usage metrics, log files and performance dashboards through the REST API, and Lyzr adds token usage and cost per interaction with alerting on degradation. Log retention is seven days on Community, Starter and Pro and customizable only on Enterprise.

Sourcedocs.lyzr.ai platform intro, API build path and plansread 2026-08-31
Memory & State Persistence Full

Cognis, Lyzr's recommended memory module, automatically stores interactions, retrieves relevant context, updates stale memories and handles temporal reasoning, with a cross-session toggle so the agent remembers past conversations with the same user across sessions. Session memory keeps the last N messages verbatim and summarizes older ones, retrieval is by session ID or user ID, and Global Context supplies org-level context to every agent. Lyzr does not describe a way to review or delete stored memories.

Sourcedocs.lyzr.ai memory-context pageread 2026-09-29
Deployment & Data Residency Full

Lyzr deploys three ways, as Lyzr Cloud managed SaaS, on-premise with the platform running inside the customer's own VPC and data never leaving their environment, and hybrid with orchestration in the cloud while data connectors run on-premise. Regional hosting serves deployments with requirements. Lyzr also offers air-gapped operation with models, agents and data inside the customer's VPC, region and keys, one-click VPC deployment and Bedrock-native running, and the open-source framework can be run locally. Deploying on Lyzr Cloud or On-Prem is an Enterprise entitlement.

Sourcedocs.lyzr.ai platform intro and plans, lyzr.ai sovereign AI and Agent Studio pagesread 2026-08-31
Prebuilt Agents, Templates & Packs Full

Blueprints are a tiered entitlement, Few on Community, Starter and Pro and All on Enterprise. In Studio a new agent starts from a template or from scratch, and a Blueprint Library is the first stage of Lyzr's agent development lifecycle. Named pre-built role agents ship with dedicated pages, including an SDR, marketer, HR, support, competitor-analysis and RFP agent, and blueprints span banking, insurance, HR, procurement and support functions.

Sourcedocs.lyzr.ai plans and quickstart, lyzr.ai product pagesread 2026-08-31
Triggers & Channel Coverage Partial

Agents reach people through voice over telephony providers including Twilio, Telnyx and Plivo, a callable REST endpoint that every agent exposes on creation, a Custom UI Builder for branded interfaces, and built-in connectors to chat systems such as Slack. Lyzr has no scheduled, recurring, cron, event or webhook trigger, and its agent development lifecycle covers design, build, evaluate, deploy, observe and govern without a trigger stage.

Sourcedocs.lyzr.ai platform intro, quickstart and API build pathread 2026-08-31
Model Flexibility & Routing Full

The ADK sets the model as a per-agent provider property in code, and the Agent API takes an llm_config naming provider, model and parameters per environment. Each agent can be routed to OpenAI, Anthropic, Gemini, Bedrock or DeepSeek, the control plane sits above the LLM choice so models swap with a single config change without touching agent logic, and cost comparison across models is built in. Model access is Limited on Community, Starter and Pro, with All Models plus Bring Your Own Model on Enterprise.

Sourcedocs.lyzr.ai ADK, Agent API cookbook and plans, lyzr.ai Agent Studio pageread 2026-08-31
APIs, SDKs & MCP Extensibility Full

A versioned REST API has a full endpoint reference covering authentication, request and response formats and example payloads, and every deployed agent exposes its own endpoint and key through an Agent API surface in Studio.

A Python Agent Development Kit installs from pip as lyzr-adk for creating and running agents, a TypeScript SDK is also named, and the platform spans a visual no-code studio, a Python and TypeScript SDK, and raw REST APIs.

Native Model Context Protocol support lets any MCP-compliant tool plug in, the core framework is open source on GitHub, and an API build path supports CI/CD-driven agent deployment.

Sourcedocs.lyzr.ai API reference, ADK overview and platform introread 2026-08-31
Testing, Debugging & Optimization Full

Agents are evaluated with automated test-case generation and production-readiness scoring, and Agent Eval is a built-in module that replaces external evaluation tooling.

A Simulation Engine runs thousands of domain-specific, adversarial and edge-case scenarios before deployment, with pre-built test suites for regulated domains, regression re-runs of the same simulations when models upgrade or prompts change, and drift detection that feeds production traces back into the next test suite. Studio has a built-in chat simulator for sending sample queries during build. Agent Eval is available only on Enterprise.

Sourcedocs.lyzr.ai platform intro, quickstart and plans, lyzr.ai simulation engineread 2026-08-31
Browser & Computer Use Not documented

Lyzr's product includes no browser, desktop or computer-use capability, and its named platform modules cover agent framework, orchestration, memory, RAG, tool connectors, voice, guardrails, hallucination control, runtime, observability, evaluation, UI building, access governance and cost management. Agents act on outside systems only through connectors, custom REST calls and MCP-compliant tools. Lyzr does not describe two of its product names, Lyzr Nitro and Lyzr Optimus.

Sourcedocs.lyzr.ai platform intro, ADK, quickstart and plansread 2026-08-31

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Product updates

·Security / enterprisePartially Verified

Lyzr launched Sovereign AI, a solution providing enterprise-grade governance for regulated industries. It enables organizations to build, deploy, and govern AI agents for mission-critical operations while maintaining strict data sovereignty and compliance.

Bears on: Deployment / data residency

View source
·Agent capabilityPublicly Claimed

Lyzr open-sourced SivaClaw, the AI agent it initially built to support its own Series B fundraising process. The agent runs on GitAgent using the OpenGAP protocol and handles automated workflows like investor communication, diligence material organization, and company briefings.

Bears on: Agent capability

View source
·Observability / auditabilityVerified

Lyzr released the Agent Improvement Engine, which analyzes live production traces to detect quality issues in deployed agents. Its Agent Hardening layer spots failure patterns and auto-generates recommendations to revise an agent's instructions or goals.

Bears on: Observability / auditability

View source
View all 6 Lyzr updates →Tracked since Feb 2025 · Verified from public vendor sources

Pricing

From $19 a month on Starter, or $99 a month on Pro ($79 billed yearly). Production agent runs add $0.08 per run on Lyzr Cloud or $0.03 on VPC, and LLM tokens are billed by the model provider.

A flat monthly subscription with bundled credits, plus a charge per production agent run and LLM tokens passed through at provider rates.

Free tier

Included quota

Community is free, with 10 agents, 500 credits a month, 1 builder license, 5 knowledge bases, 100MB of RAG storage, 5 tools, limited orchestration, limited models and blueprints, 7 day logs and email support. Starter, at $19 a month and billed monthly only, has 15 agents, 2,000 credits a month, 1 builder license, 10 knowledge bases, 100MB of RAG storage, 10 tools, limited orchestration, limited models and blueprints, 7 day logs and email support. Pro, at $99 a month or $79 a month billed yearly with 120K credits a year, has 25 agents, 10,000 credits a month, 1 builder license, 15 knowledge bases, 1GB of RAG storage, 25 tools, limited orchestration, limited models and blueprints, 7 day logs, a few Super Agents and email support. Enterprise is custom, with unlimited agents and knowledge bases, custom credits and storage, unlimited tools, full orchestration, all models plus Bring Your Own Model, all blueprints, unlimited Super Agents, team collaboration, Agent Entitlement Protocol, Responsible AI, Hallucination Manager, Agent Eval, customizable log retention, deployment on Lyzr Cloud or on premises, Lyzr Agent Build Services, a 48 hour custom integration SLA and 24/7 support. Yearly billing gives two months free.

What is public

Lyzr (lyzr.ai) lists its Agent Studio subscription tiers in its documentation rather than on its pricing page. The four tiers are Community, free, Starter at $19 a month, Pro at $99 a month or $79 a month billed yearly, and Enterprise, custom. Tiers scale by agents (10, 15, 25 or unlimited), monthly credits (500, 2,000, 10,000 or custom), knowledge bases (5, 10, 15 or unlimited), RAG storage (100MB, 100MB, 1GB or custom), tools (5, 10, 25 or unlimited) and log retention (7 days on all three self serve tiers, custom on Enterprise). All three self serve tiers carry one builder license, limited orchestration and limited model and blueprint access. Production usage is billed per agent run on top of the subscription, at a lower rate for deployments the customer hosts, with LLM tokens passed through separately.

Billing mechanics

Billing has three layers. The first is a flat subscription tier with a bundled monthly credit allowance (500 on Community, 2,000 on Starter, 10,000 on Pro, custom on Enterprise), and credits meter build time and premium operations such as knowledge base operations, tool calls and specific features. The second is a charge per production agent run on top, $0.08 per run on Lyzr Cloud and $0.03 per run on Lyzr VPC or on premises. The third is LLM token costs passed through at provider rates (OpenAI, Anthropic, Google, Bedrock), paid directly to the model provider. Annual billing gives two months free (Pro $99 to $79 a month). Extra credits top up one time at $0.01 a credit, from $10 for 1,000 up to $5,000 for 500,000.

Cost watchouts

The subscription is the small part of the bill. Production agent runs ($0.08 per run on Cloud) and LLM pass-through dominate at scale. At 1,000 runs a day on Cloud, runs cost about $80 a day, or about $2,400 a month, against the $99 Pro subscription, and a 20+ agent workflow runs about $1.02 per successful run on premises. LLM token costs frequently exceed the platform fee. VPC or on premises deployment cuts the run rate to $0.03 but requires Enterprise.

Variable cost rationale

Production agent runs are metered, at $0.08 per run on Lyzr Cloud and $0.03 on VPC or on premises, so the bill grows with invocation volume and the deployment choice changes the cost of each run. LLM tokens are passed through at provider rates and are often the largest line. Credits are also drawn down by knowledge base, tool and premium operations. The flat subscription is a small part of the bill. No cap applies to run charges. Credits are bounded by the monthly allowance and one time top ups at $0.01 a credit, and the lower VPC rate needs an Enterprise deployment.

Additional watchouts

Charges per agent run and LLM pass-through make up the real bill, and the $19 or $99 subscription is a small share of it at production scale.

Overage / add-ons

One time credit top ups cost a flat $0.01 per credit, from $10 for 1,000 credits through $50 for 5,000, $100 for 10,000 and $1,000 for 100,000 to $5,000 for 500,000. They are added instantly, need no subscription and are available on all plans. Production agent runs are billed separately as you go, at $0.08 per run on Lyzr Cloud and $0.03 per run on VPC or on premises, with no cap, and neither run rate appears in the plans documentation. LLM token usage is passed through and billed directly by the model provider at market rates, and it frequently exceeds the platform fee at scale.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free Community plan with 500 credits a month, no card needed

Lowest paid plan

Starter at $19 a month, billed monthly only

Commercial notes

Lyzr positions Agent Studio as the first agent framework with safe and responsible AI built into the core. It focuses on regulated verticals (BFSI, insurance, healthcare, IT services), with prebuilt agentic blueprints and multi agent orchestration. 75% of customers run 2+ agents, and an NTT Data case study cites 70,000+ hours saved. Lyzr is based in New York. It raised an $8M Series A in 2025 (Rocketship.vc and Accenture Ventures), then a $14.5M round led by Accenture in March 2026, and it has 400+ enterprise customers (Accenture, AWS, Hitachi Energy, Publicis, AirAsia). Lyzr builds agents and does not provide contact data, so B2B data needs a separate enrichment layer.

Key ambiguities

Tier gating matters more than price ambiguity. Agent Eval, Responsible AI, Hallucination Manager, Agent Entitlement, team collaboration and on premises deployment are absent from Community, Starter and Pro entirely, and orchestration is limited on all three tiers and full only on Enterprise, so the platform's differentiating capabilities require a custom quote. Production is billed per agent run, with Cloud and VPC or on premises tiers, and a complex multi agent workflow is illustrated at roughly $1 per successful run. The $0.08 Cloud and $0.03 VPC rates per run are no longer restated by Lyzr. Starter has no yearly equivalent. What consumes a credit, as against what makes a billable agent run, is not itemized.

Cancellation / refund

Community is free with no card. Starter is monthly only, and Pro is monthly or yearly with two months free ($99 to $79 a month). Credit top ups are one time at $0.01 a credit, and Enterprise and on premises deployments are custom contracts (50+ licenses, build services).

Support SLA / resale

Lower tiers get email, help desk, chat and knowledge base support. Enterprise adds 24/7 support, a 48 hour custom integration SLA and dedicated build services.

Missing data

Exact rules for credit consumption against agent runs are not fully itemized. The subscription tiers (Community free with 500 credits, Starter $19 a month with 2,000, Pro $99 a month with 10,000 or $79 a month billed annually, Enterprise custom) are not the main cost driver. Production is billed per agent run ($0.08 on Cloud, $0.03 on VPC or on premises) on top, plus LLM token pass-through and $0.01 per credit top ups.

Agentic Index verified 2026-08-31
Official pricing page Docs

Related vendors

Alternatives to Lyzr

The closest documented capability profiles to Lyzr among agent builders tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Moveo.AI12.0 / 14Fuller documented coverage on Triggers & Channel Coverage
  • SnapLogic12.0 / 14Fuller documented coverage on Triggers & Channel Coverage
  • Vellum11.0 / 14A lighter documented profile than Lyzr
  • Mendix12.5 / 14Fuller documented coverage on Human Oversight & Guardrails and Triggers & Channel Coverage
  • OutSystems12.5 / 14Fuller documented coverage on Human Oversight & Guardrails and Triggers & Channel Coverage
  • AgentX12.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Triggers & Channel Coverage

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare Lyzr with other vendors →Compare all agent builders →See where Lyzr ranks →
Data confidence: high

Head to head

Lyzr vs AiriaLyzr vs StackAI

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.