sivasurend

Our Anthropic bill started zooming toward $1 million. So we built a system to manage coding agents.

Meet BaseCode - MDM for coding agents.

Here is how it works:

1. Engineers run Claude Code, Codex, Copilot, Gemini, and opencode on their machines. 2. BaseCode centrally injects agent identities, coding policies, skills, MCP servers, and models into all of them. 3. Then it observes every prompt, project, and token, and enforces guardrails and budgets across the fleet.

Runs locally in your own cloud (air-gapped). Supports all model providers and frameworks.

Starts at $20/dev/month.

You're right, the name 'Claw' creates a lot of hesitation among enterprises. But it's also the best way to convey what you're trying to build, especially if you're building a file system-powered agent. I think the surname 'claw' will work for some more time. :)

We just hit $3M in Contracted ARR in under 8 months at Lyzr AI!

Here’s our journey

Lyzr began as an AI data analysis app(PowerBI+ ChatGPT). The name Lyzr comes from the last four letters of "Data Analyzr."

However, while building the app on Langchain, we encountered several challenges related to deployment, troubleshooting, bug fixing, and monitoring.

Our initial idea didn’t gain traction, and we faced rejections from YC, AIGrant, HF0. But this led us to pivot and create a Langchain alternative for enterprises, using an agentic approach.

Lyzr was reborn as an Enterprise Agent Framework for building reliable and fully private AI agents.

We gained some decent traction from January to April 2024, with customers building single-agent apps like chatbots, knowledge search tools, and data analysis apps.

As a hack in the early days, we used to search for job postings in Upwork where customers asked for 'Langchain' support. We used to meet with those customers, understand their problems, build the prototype on Lyzr in <24 hours, and convert them.

We went from $0 to $100k in ARR in the first 4 months.

Then, we launched our multi-agent orchestration capability—Lyzr HybridFlow. Around this time, I came across a tweet by Jason Lemkin looking for AI agents to automate research, outreach, and sponsorship sales for the SaaStr Europa event.

We took on the challenge, building and launching a 'downloadable AI SDR agent' template in just 4 days. While the tweet gained some traction, we saw even more interest on LinkedIn and Instagram as influencers picked up our story and shared it.

In about 10 days, we received about 250 inbound leads. We decided to productize the AI SDR agent, naming it Jazon, and launched it with a world-class UI. Paired with Skott—The AI Marketer (Copy AI+Surfer SEO) - these two agents started bringing us a steady stream of leads.

I personally handled most of the initial calls, demoed the product, and onboarded our early customers onto a 3-month pilot. Interestingly, more than 50% of the agents built on the Lyzr platform today by enterprises are custom agents designed to automate complex business workflows (think UIPath, but built with self-learning AI agents).

We have a strong pipeline and are yet to start proper outbound. We’re aiming for $6M ARR by the end of this year and $30M next year.

Can we get to $100M in 3 years?

Lyzr is a low-code agent framework, building fully-integrated private agent SDKs for the 99% of the business who do not have the liberty of time, talent, need to build on complex open-source frameworks like Langchain or LlamaIndex.

We debuted our low-code multi-agent workflow automation framework 3 weeks back. We are building fast, learning actively from other agent communities like Autogen, LangGraph, CrewAI.

Check our library and let us know what you think?

We have Agents, Tasks, Tools (as Functions) up.