Two Google alumni raise $11.3M to back AI startups that enterprises will actually pay for

BAG Ventures, founded by former Google executives Bontia Stewart and Jackson Georges Jr., closed an $11.3 million early-stage AI fund after already investing in 10 companies. The firm will write $100,000-$500,000 checks over the next two years, targeting AI infrastructure, agentic systems, security, governance and vertical SaaS. Its thesis is that enterprise AI spending is shifting from experimental chatbots toward deterministic, workflow-integrated products with measurable unit economics, proprietary data and clear monetization.

Analysis

This is a modest but useful read-through on the next phase of enterprise AI budgets: procurement is likely to concentrate in applications with measurable labor substitution, auditability, and integration into systems of record. That favors platforms controlling data, workflow distribution, and security primitives—MSFT, NOW, CRM, SNOW and PANW—over public software names whose AI monetization remains primarily seat-price expansion or generic copilots. The near-term implication is multiple dispersion rather than a broad AI-software rerating: vendors unable to quantify ROI at renewal risk facing longer sales cycles and AI-feature commoditization over the next 1-3 quarters.

The less obvious beneficiary is identity and security infrastructure. Autonomous agents expand the number of machine identities, permissions and data-exfiltration paths faster than traditional endpoint security architectures were designed to handle; OKTA, CRWD, PANW and CyberArk (CYBR) have credible routes to capture this spend, though agent-specific revenue is unlikely to be material before 2027. Hyperscalers also benefit indirectly because regulated deployments tend to require private connectivity, governance, inference capacity and managed security, supporting GOOG, AMZN and NVDA demand even if application-layer startups capture limited durable economics.

Contrary to the bullish venture framing, outcome-based pricing can pressure software gross margins and cash-flow visibility. If customers pay for completed work rather than licenses, the vendor absorbs model-inference costs, exception handling and implementation risk; only companies with proprietary workflow data or low-cost distribution should retain SaaS-like margins. The key falsifier is evidence that enterprise AI projects produce net-new budget rather than displace existing software seats: watch FY2027 guidance and remaining-performance-obligation growth at NOW, CRM and SNOW, alongside cloud AI consumption disclosures.

Key Decisions for Investors

• No standalone trade on the fund close; treat it as a thematic confirmation, not a price-moving catalyst.
• Overweight CYBR and PANW versus a basket of high-multiple application software through the next 6-12 months. Agent identity, policy enforcement and data-governance spend has a clearer budget owner than broad productivity copilots; reassess if enterprise security bookings decelerate below low-teens growth or valuation premiums expand materially.
• Pair trade: long NOW / short ADBE over 3-6 months, sized modestly. NOW has deeper workflow ownership and can tie automation to ticket-resolution economics, while ADBE remains more exposed to AI-driven feature parity and pricing pressure; exit if NOW subscription growth misses guidance or ADBE demonstrates sustained AI-driven net-new ARR.
• Maintain core GOOG and AMZN exposure rather than chase smaller AI application names. Regulated enterprise deployments should increase consumption of cloud, governance and private-data tooling, but reduce exposure if cloud growth fails to accelerate despite rising AI capex—evidence that workloads are cannibalizing rather than adding spend.
• Set an earnings-monitor alert for SNOW: sustained consumption growth above guidance coupled with improving gross margin would validate proprietary-data platforms as AI toll collectors; weak consumption or rising compute costs would argue against the thesis and favor security over data-platform exposure.