Within a week, a16z invested in two AI agent startups: $7.4 million in Cotool (cybersecurity) and $30 million in Lio (procurement). This is not a coincidence - it is a theory. Why autonomous AI agents will be the next big enterprise market, and what that means for investors and founders.
Two deals, one thesis: agents instead of tools
At the beginning of March 2026, Andreessen Horowitz made two investments in seemingly completely different markets within a few days: $7.4 million seed for the cybersecurity startup Cotool and $30 million Series A for the procurement automation company Lio.
The common denominator? Neither company builds software tools. They build autonomous AI workers – agents that do not assist, but independently take on tasks, make decisions and execute processes end-to-end.
$37.4 millionTotal investment in one week.2 Enterprise verticals. 1 common thesis: Agentic AI.
This investment strategy is not isolated. It reflects a broader paradigm shift: away from software that supports people - towards AI systems that can replace or expand entire departments. For investors, this is a fundamental reassessment of enterprise software markets.
Cotool: AI agents for cyber defense
Cotool is building what the founders call a "distributed cybersecurity factory." The idea: AI agents take over the entire security lifecycle - from detection to analysis to response to threats.
The timing is no coincidence. Cybersecurity teams around the world are struggling with a massive skills shortage. At the same time, the complexity of the threat landscape is increasing exponentially. Cotoool's approach: Agents that continuously monitor systems, detect anomalies and automatically investigate incidents - around the clock, without fatigue, without staff shortages.
The investor list underlines the conviction: In addition to a16z, Y Combinator, WndrCo and angels from Okta, Cloudflare and Ramp are involved - all insiders of the enterprise security market.
Lio: Procurement on autopilot
Lio goes one step further: The startup from the USA and Tel Aviv provides companies with a complete virtual procurement department. The agents triage inquiries, analyze offers, compare suppliers, negotiate, onboard vendors and carry out purchases end-to-end.
The results are impressive: processes that once took weeks are completed in minutes. Dozens of Fortune 500 and Global 2000 companies are already using the system - including Munich Re, Brose and Novozymes. Lio's agents already manage billions in enterprise spend.
Why the agent market is exploding now
The investment thesis behind Agentic AI is based on three converging developments:
Foundation models are mature enough:Only since 2025 have large language models been reliable enough to operate autonomously in business-critical processes. The error rate has fallen below the level at which enterprise customers are willing to delegate decisions to AI.
The ROI is immediate and measurable:Unlike many AI applications that increase productivity in a way that is difficult to quantify, AI agents deliver directly measurable returns: a procurement agent that completes a 3-week process in 15 minutes has an ROI that every CFO understands.
The shortage of skilled workers is forcing us to rethink:Whether cybersecurity analysts or procurement managers: qualified specialists are rare and expensive. AI agents don’t solve an abstract problem – they fill real gaps in real departments.
What Cotool and Lio have in common is a phenomenon that behavioral economists callLoss aversionknow: Companies invest faster in solutions that prevent losses (security gaps, procurement inefficiency) than in those that maximize profits. This makes the agent market particularly attractive for investors – customer acquisition follows the pain principle, not the feature comparison.
What this means for investors and founders
For investors – especiallyFamily officesand institutional investors – Agentic AI is creating a new asset class within the tech sector. The evaluation logic differs fundamentally from classic SaaS software:
Instead of “seats” (workstations), prices are based on “agents” (workforce units). Instead of feature adoption, process automation is measured in hours and dollars. And instead of slow enterprise sales cycles, the best agent startups experience rapid bottom-up adoption – because individual departments see results immediately.
For founders in this segment, the timing has rarely been better. But the bar is rising. Investors like a16z don't just expect a working product - they expect enterprise customers with measurable impact and a clear answer to the question of why this team deserves the trust of Fortune 500 companies.
This is where the right one comes inFinancial feasibility analysisCrucial: Not every agent startup can be financed, but those that are are currently finding a historically favorable financing window.