Disrupting SignalRank

In February this year, Citrini released their now infamous AI doomer research. The next day there was a mini market meltdown as investors digested the core contention of the report: AI is coming for labor markets much more quickly and much more deeply than originally anticipated.

In particular, Citrini’s report was a critique of the value of human relationships:

Even places we thought insulated by the value of human relationships proved fragile. Real estate, where buyers had tolerated 5-6% commissions for decades because of information asymmetry between agent and consumer, crumbled once AI agents equipped with MLS access and decades of transaction data could replicate the knowledge base instantly. A sell-side piece from March 2027 titled it “agent on agent violence”. The median buy-side commission in major metros had compressed from 2.5-3% to under 1%, and a growing share of transactions were closing with no human agent on the buy side at all.

We had overestimated the value of “human relationships”. Turns out that a lot of what people called relationships was simply friction with a friendly face.

In essence, Critini argued that AI would reduce almost all jobs to Claude skills, tasks reduced to LLM-executable workflows. And, yes, likely including VCs.

Regardless of the credibility of the claims, agentic venture capital is an interesting thought experiment to consider.

VC has historically been somewhat resistant to change. There is a widespread belief that data (and AI) will change every industry in the world except VC. Talent-spotting, coaching & supporting founders is too much of a human activity.

For 10+ years, there have been a smattering of pioneers (such as Correlation Ventures) who have rebuilt workstreams around data for sourcing & selection.

But it is only in recent years “data-driven” VC has become more popularized and VCs have been discussing their own tech stack.

AI agents will likely accelerate this process further.

Our view is that alpha-seeking VCs at the earliest stage will remain a predominantly human endeavor for the foreseeable future. Spotting, selecting & coaching talent has some ineffable qualities that are likely challenging for a model to replicate in the short term. Human qualities if you like. And each investment must have the potential to return the fund. This is a lot of pressure (risk) to put on a model.

But later stage VCs and groups, like SignalRank, who are innovating with smart-beta / index-like models, will likely adopt agentic venture capital at pace. A highly diversified strategy puts less weight on any one individual investment.

Various workstreams, from market mapping to expert network calls, content generation and investment memo generation, are now moving into the agentic realm.

We are already using agents to generate monthly reports on the broader VC ecosystem. You may have noted our disclaimer at the top of these reports: “This report was written by our suite of AI Agents using Claude Code as orchestrator”.

The future disruption of SignalRank’s current workstreams is likely to focus initially on full automation of our investment process (while retaining human board approval for each new investment).

But sometimes humans want human connection. SignalRank itself relies on human relationships with human seed partners for our entire access model.

We had previously built a software tool to enable our seed partners to upload live Series B term sheets to automatically check whether a Series B qualified for our product. We quietly let it deprecate as no-one used it. Our partners prefer email & WhatsApp.

We are still going to try again here. The next iteration we are working on is creating a WhatsApp group with a SignalRank team member, a SignalRank agent and our seed partner. We hope that most of the interaction is between the agent & the partner, with the SignalRank human just there for oversight.

One area where we continue to see human relationships as absolutely critical is in our own fundraising. Fiduciary duty of our shareholders requires trust, not data.

This is particularly true in VC where the expectation is each investment is a multi-decades long relationship across funds.

Public market quant funds are more advanced. Explainability for investment decisions is not easy. Public quants have developed to this level of trust, with no human input into individual investments, over multiple decades.

The quant VC market is not there yet. This trust is still to be earned. Human oversight is still required, at least for now.

The listing of VCs may be a catalyst for change here as the closed-end fund model enables investors to seek liquidity on a daily basis. This potential for daily liquidity in part forces the manager to up their game. An investor can always sell to buy something else.

The real question is not whether AI will enter venture capital. It already has. The question is where humans remain indispensable.

Kyle Harrison previously argued that venture capital doesn’t exist as one monolithic asset class. This line of thinking also applies here.

Venture capital won’t be replaced wholesale by agents. A split will emerge and harden. One one side, artisanal human-crafted investing focused on talent spotting and nurturing. And on the other, higher velocity, systematic, agentic capital allocation.

SignalRank is building the latter.