Andreessen Horowitz (a16z) has identified a massive but unglamorous opportunity: using artificial intelligence to improve how organizations use the legacy software systems they already depend on, rather than replacing them entirely. While startups and investors have focused on flashy new AI capabilities like voice agents and workflow automation, a16z argues the real value lies in helping enterprises get more from systems like SAP, ServiceNow, and Salesforce that encode decades of business logic and institutional knowledge. Why Is Legacy Enterprise Software So Hard to Replace? At first glance, systems like SAP seem outdated and painful to use. The interfaces are clunky, navigation is confusing, and customization is a nightmare. But beneath that ugly exterior lies something far more valuable than most people realize: these systems capture the canonical data model of the business, the permissions and controls that keep it compliant, the workflows that make it operable at scale, and the integrations that connect dozens or hundreds of downstream processes. The real problem is switching costs. Upgrading from SAP ECC to SAP S4HANA, for example, can cost $700 million, take three years, and require a team of 50 consultants from Accenture. Even then, the software becomes mostly useful only for generating read-only reports that are nearly impossible to manipulate. The more a company has invested in custom fields, workflows, pricing rules, and reporting logic, the more the system becomes a competitive moat. This is why roughly 70% of large-scale enterprise transformations fail to meet their objectives. What Problem Are AI Startups Actually Solving? Rather than trying to rip and replace these systems, a16z-backed startups are building AI tools that make legacy software faster to implement, easier to use, and simpler to extend. The opportunity breaks down into three phases of the software lifecycle: implementation, daily operations, and building on top of the system as the business changes. In the implementation phase, which is the riskiest and most budget-sensitive, AI can turn messy discovery meetings and documentation into structured requirements, then auto-produce the implementation workstream including process mappings, configuration code, test scripts, cutover plans, and migration playbooks. This is where the clearest payback exists. Companies are building copilots, project management tools, and assurance layers to de-risk and accelerate these transformations. How Are a16z Portfolio Companies Tackling This Opportunity? a16z has invested in several startups working across different parts of the legacy software transformation problem: - Axiamatic: An AI assurance layer for ERP systems that builds a knowledge graph from project artifacts and flags hidden failures in requirements and change management via Slack or Teams, partnered with SAP Build and baked into KPMG, EY, and IBM consulting motions. - Conduct: A code and process-mapping copilot that generates a semantic layer and technical documentation across ECC to S4 migrations, with question-answering capabilities over custom tables and APIs to speed internal takeover. - Auctor: An agentic implementation delivery platform for systems integrators and professional services that auto-captures discovery into structured requirements before becoming a system of record for statements of work, design documents, user stories, configurations, and test plans. - Supersonik: An AI-powered product enablement platform for channels and resellers that uses vision and voice agents to teach inside the real user interface, reducing systems engineer headcount needs and enabling reseller-led implementations. - Tessera: An AI-native systems integrator that manages enterprise transformations end-to-end by connecting into a customer's existing ERP instance, evaluating how it's implemented, and automatically remediating what needs to be changed during migration. These companies create value by making transformations faster, cheaper, and less risky. They catch problems early in requirements and change management before they snowball, compress timelines where a single slipped month can cost millions, turn messy project data into structured knowledge so internal teams can take ownership faster, and reduce reliance on large systems integrator teams through automation of mapping, documentation, testing, and enablement. How to Build AI Tools That Work With Legacy Enterprise Software a16z sees room for more startups building tools that work alongside existing partners rather than against them. The winning approach focuses on three key areas: - Implementation Agents: Tools that share in outcomes and risk by handling requirements tracking, configuration comparison, cutover simulation, code generation, and drift detection to ensure transformations stay on track. - Semantic Documentation Tools: Platforms that keep knowledge current and accessible across the organization, turning institutional memory into structured, queryable information that survives staff turnover. - Enablement Agents: Systems that turn training and channel rollout into repeatable, scalable processes, reducing the need for expensive human trainers and enabling faster adoption across the organization. The destination with AI might not be to "replace SAP, ServiceNow, or Salesforce," but to make them more programmable and approachable. The winners will be platforms that plug into transformation budgets with measurable risk and timeline reduction, then expand into day-to-day operations as the trusted control plane for work, gradually unbundling the legacy user interface into composable, governed, AI-assisted actions and thin applications. What Does This Mean for the Broader AI Investment Landscape? This thesis represents a significant shift in how a16z thinks about AI's impact on enterprise software. While the firm continues to invest in frontier AI models and new AI-native applications, the recognition that legacy systems represent a $380 billion software implementation and systems integration market suggests that the real value may lie in making existing infrastructure smarter rather than replacing it. The broader context shows a16z is diversifying its AI bets across multiple categories. The firm recently raised a $6.75 billion Growth fund, bringing its total assets under management to over $22 billion across five Growth funds. The firm's portfolio includes companies like Waymo, ElevenLabs, Kalshi, and Coinbase, reflecting investments across autonomous vehicles, voice AI, prediction markets, and fintech. Additionally, a16z has made significant bets on Latin American startups, including a $6 million seed investment in Handle, an AI agent platform for enterprise operations founded by Alfonso de los Ríos, the former CEO of Mexican unicorn Nowports. The core insight remains consistent across all these investments: the fastest-growing opportunities in technology are increasingly happening in private markets, and AI is forcing investors to rethink traditional metrics for evaluating software companies. For legacy enterprise software, that means the next wave of value creation won't come from wholesale replacement, but from intelligent layers that make existing systems work better for the humans who depend on them every day.
