Build and test for free. Pay by the agent-month only for agents in production. Add governance when you need it.
For anyone exploring agents.
We want everyone to become an agent builder. So building stays free.
What you get
For teams putting workflows on agents.
Software, plus the people who get your agents working.
What you get
For enterprises with agents everywhere.
Software and people, sized to how many agents you govern.
What you get
Each is priced with you and added to your plan.
Prices in US dollars, billed yearly. We never bill for model tokens.
Agent Test Drive
Bring one real workflow. Build working agents with our engineers, and see them run on your own data before you commit.
Offered to a select set of enterprise teams. If something else fits you better, you'll know that too.
An example run. On your test drive, the workflow and data are yours.
Builders start by taking agents to production. The C-suite starts by governing the agents they already have. The rest joins as you grow, inside one plan we shape with you.
Where developers and AI teams start
ArchitectDescribe an agent in plain language and get a working one.Agent StudioBuild, test and tune agents visually. Test safely against a sandbox of your systems.Cognis · OpenGAP · GitAgentGive agents memory, and build in the open.Where builders and IT start
Agentic OSRun a whole function's workflows on agents.SuperFlowOrchestrate agents into multi-step workflows.Open CoworkA governed AI coworker for every employee.Jazon · Skott · Jeff · DianeStart from ready-made agents for sales, marketing, support, HR.Where the C-suite starts
OpenControllerOne view of every agent, whoever built it and wherever it runs. Policies, audit trails and spend limits per agent.BasecodeControl over the machines where agents are built.Responsible AI modulesHallucination checks, PII masking, bias evaluation.When you want it all in-house
Lyzr OptimusLyzr’s own hardware: run agents, models and governance on an appliance in your data centre, air-gapped. Ships Q4 2026.OpenfinetunerFine-tune open-weight models on your own data.Agent BlocksComposable building blocks you own, to assemble agents on your own stack.Think of three roles on a team. Every role matters and the team needs all of them; each does a different kind of work and draws a different share of one yearly budget, the way salaries do. We call it the agent-month.
Handles one job well: answers questions, finds information, deals with simple requests.
Runs one process end to end, coordinating several agents across your systems.
Works on its own machine, in any application, the way an experienced person does.
| Agent-months drawn per agent, per month | Build & Testnot in production | Quietunder 100 tasks | Standardup to 2,000 tasks | Custom |
|---|---|---|---|---|
| Tier 1AssistantOne agent, one job: answers questions, finds information, handles simple requests. | Free | ¼ | ½ | Talk to usSized with you |
| Tier 2SpecialistSeveral agents in one workflow, running a process end to end. | Free | ½ | 1 | |
| Tier 3ExpertAn agent with its own virtual machine, using applications the way a person does. | Free | 2its VM is reserved | 2 |
More on counting, tasks and which products use agent-months in the FAQ.
Production and governance plans include Reimagineers, who own the business outcome, and forward-deployed engineers, who own the build. Together they take you through RADAR, Lyzr's five-stage method, from problem to measured result.
The detail behind the numbers, for when you want it.
Yes. Building and testing agents is free, with no time limit. Agents start drawing from a budget only when they run in production.
No. Every product is sold on its own. Most teams start with one thing: taking agents to production, or governing the agents they already have. Add more when it makes sense; adding never reprices what you already have.
It depends on how many agents and teams you govern, and across which platforms. We size it with you on a call.
Yes. Production and governance plans include Reimagineers, who own the business outcome, and forward-deployed engineers, who own the build. They work alongside your team through RADAR, from reimagining the work to measuring the result in your own numbers. Talk to us to find out what is included for you.
One agent in production for one month. Each agent draws from your yearly budget according to how it's built: half an agent-month for an Assistant, one for a Specialist, two for an Expert. See how agent-months are counted.
No. An agent handling fewer than 100 tasks in a month draws less: a quarter for an Assistant, half for a Specialist. An Expert always draws two, because its virtual machine is reserved. Up to 2,000 tasks a month is the standard draw, and every further 2,000 tasks starts another agent-month. See the table.
When it's connected to your production identity provider, tools and data. A sandbox pointed at UAT doesn't count.
Agents and SuperFlow workflows built in Agent Studio or Architect and running in production, and Open Cowork team agents shared with colleagues. Building and testing, Open Cowork personal agents, OpenController, Basecode, Openfinetuner and Lyzr Optimus never use agent-months; each has its own price, or is free.
Nothing stops. Your agents keep running and you add agent-months in blocks of 150, at any time.
No. You pay your model provider directly, or run open-weight models on your own hardware, so Lyzr never bills twice for the same work.
No. Hosted by Lyzr or in your own cloud, the price is the same; you cover your own cloud costs. On-premises runs on Lyzr Optimus hardware, priced separately and shipping Q4 2026.
Lyzr holds a SOC 2 Type II report and ISO 27001:2022 and ISO 42001:2023 certifications, and its security program follows GDPR and HIPAA. Reports are in the Lyzr Trust Center.
We'll show you where you'd start and what it would cost, before you sign anything.