Unicorns Are Back
And AI is Accelerating
Unicorns Are Back
I spent this week working hard on various SignalRank projects. The week ended with the good news that Saronic, a SignalRank Index portfolio company, raised $600m to continue its work providing naval and maritime forces with the most intelligent Autonomous Surface Vessels available. More below.
My work this week was to build the first multi-agent chatbot that understands SignalRank's predictive data on venture backed companies, so that investors and others could ask questions and get answers.
The agents have access to SignalRank's data warehouse. You can try asking it anything related to company scores, investor scores or round scores. You can use company names, or ask for all companies at a stage, or with common investors. It's prett flexible in reporting scores for anything, and ranking against them.
You can see the alpha version here - https://agents.signalrank.ai. FYI, this site will be up and down from time to time.
Here is what is happening under the hood:
There are several AI agents that have access to Anthropic, OpenAI, Snowflake and Perplexity. They receive questions via the Orchestrator. Depending on the question it issues jobs to the reasoning agent for analysis and the sql agents for data queries and a visualization agent to build charts. Then a document creation agent that can enable the analysis, data tables and charts to be downloaded. Everything is collected and returned to the user as a single response.
To do this I used Cursor, Anthropic, Open AI, Perplexity, Snowflake, Plotly and did not write a single line of code myself or have help from colleagues. We are in a new world of engineering as Erin Griffith writes this week in the New York Times (See below).
Here is an example of a question and the result, and the document I downloaded at the end:
Signalrank Analysis 1740246977277
349KB ∙ PDF file
Download
Download
The total time to build this agent system was about two weeks. And I was the only builder.
I tell you this to put this week's newsletter into context.
Unicorns are being minted at an accelerated pace because real breakthroughs in what is achievable are happening. And hundreds of millions of people are already using them. I could not have built this system a few months ago.
Crunchbase launched its new platform this week and announced that the future of data is predictive.
CEO Jager McConnell declared historical data dead.
"The historical data industry as we know it is dead. AI has disrupted the status quo. Companies still relying on static data are already obsolete. Crunchbase is not just adapting - we're leading this transformation. Our AI doesn't just capture what happened yesterday; it predicts what's coming tomorrow so customers can stay ahead of the market."
He is right. Congratulations. Crunchbase is shining a light on the next phase of venture data predicting IPOs, Mergers and Funding round likelihood.
SignalRank is an Index builder seeking to decide which companies to invest in. The SignalRank predictive algorithms are designed to predict value growth over time and it invests in the high scoring companies.
SignalRank examines live Series B Rounds and based on a prediction either qualifies them for investment or rejects them. It is really good at saying no.
The algorithm says no 93% of the time. When it says no the company actually fails to meet the definition of success in 87% of cases. Success is defined as a 5x gain in 5 years, or better.
In 7% of cases the algorithm says yes. In those cases, 31% of the time it is right, the company will perform at 5x gain or better in 5 years. All this is done by scoring the funding rounds using a heuristic and an ML approach. A venture fund being right more 30% of the time is very rare. The backtest of the algorithms achieves it for the 2012-2020 timeframe and is consistent.
When the algorithm rejects 93% of company Series B funding rounds it is not surprising that the remaining 7% do very well, averaging over 6x in 5 years. To achieve this predicting failure is as important as predicting success.
SignalRank invests capital behind its predictions. When you combine investing actual cash with predictive algorithms you can de-risk capital deployment and create wealth for investors.
The early stage unicorn renaissance written about by Crunchbase is real. Having capital in those companies at an early stage produces meaningful outcomes.
Saronic, part of the SignalRank Index, and predicted to be in the top 7% by SignalRank's algorithms, raised $600m at its Series C.
As Chris Metinko states below, many of the companies raising funds are doing so at Series A and B at weighty valuations.
It is a good time to be an innovator.
Essays of the Week