Fastest Growing AI Company Ever? CEO Reveals Revenue

In July 2024, to show what the Lyzr agent framework could do, the team built Jazon, an open-source AI SDR agent, as a free alternative to well-funded AI SDR startups like 11x (then charging $70K-$100K a year) and Artisan. They seeded it on Twitter, LinkedIn and Reddit with the message that an open-source version is available to try, so you don't have to pay that much. Demand followed, sold via AWS-funded pilots. In October 2024 they stopped because AI SDRs could mass-email but not reliably book meetings; they gave all signed-up customers the open-source version and went back to the core agent framework.

Lyzr told prospects honestly that it didn't know if the AI SDR would work, but asked them to sign a 12-month contract. Lyzr worked with AWS, which has a program to fund pilots, so the first three months were paid by AWS and the remaining nine by the customer; if it worked, they converted into a paying customer. This let them show the product with no upfront cost to the buyer, booked as contracted ARR.

After seeding the open-source AI SDR on social, co-founder and head of growth Annie's team 'got the SEO in play', bringing inbound leads that were taken to demos. No detail on keywords, content or cadence.

Lyzr pitches itself as a 'sovereign, full-stack' agent platform that runs in the customer's own cloud or on-prem. The argument: when an insurer uploads its underwriting handbook to OpenAI or Anthropic, its intelligence goes to them and can be shared with every other insurer, so 'your secret sauce just went out.' It also claims to replace 20 separate technologies with one platform and to have 50% more agent building blocks than top competitors.

When Lyzr Agent Studio launched in January 2025, the only offer was a $10,000 annual plan. In Q1 they onboarded 25 customers, mostly startups and SMBs. By Q3 2025 they added $50,000 and $100,000 annual plans, then moved on to enterprise deals of $250K-$1M.

Lyzr Optimus is an on-prem appliance (H200/H100 or AMD chips, cut off from the internet) with the full agent studio preinstalled. Four configurations, from Optimus Personal at $10,000 (a device on an executive's desk) to Optimus Max at $1M (14 bundled models, up to 10,000 concurrent users). Pitched as 95% cheaper than frontier API usage and 30 minutes to set up. Hardware BOM is ~48-50% of the price; the margin comes from the software. Pre-orders opened only to existing clients first, not listed publicly.

In large deals Lyzr is usually up against Palantir, LangChain and similar. In two deals against Palantir (the US government and an unnamed critical-infrastructure enterprise) it won on an agent-first product with no legacy software to sell, and by shipping a prototype to the customer in 10 days. The guest frames the underdog tag as an advantage.

Anyone can create a free Lyzr Studio account and gets 500 credits to start building agents. Studio targets developers; a separate product, architect.new, targets business users by planning the agent, building it visually (prompts, model choice) and building the app in one shot.

Instead of acquiring companies, Lyzr treats system integrators as its 'inorganic growth': several invest in Lyzr and partner with it because Lyzr helps them ship agents to production for their end customers. Accenture, the world's largest IT services provider, invested twice in six months (Accenture Ventures joined the Series A). The guest's rule: if you make your partner win, they build the pipeline and bring the demand.

Buy 30 to 40 sending domains on Lemlist or Instantly, schedule sends and use the API to plug in an agent or Claude to write the opening email. Skip the seven-email drip: send two emails that hit the point and include links and screenshots so the prospect can decide. 'If they don't decide in two emails, they're not going to decide in seven emails.' Expect AI SDRs to handle mass emailing, not to book meetings.