Lightsage raises $4M as AI agents change how users discover and use software

The rise of AI coding agents is beginning to reshape how software companies attract users, with artificial intelligence increasingly moving beyond recommending products to directly discovering and using them.

San Francisco-based startup Lightsage has raised $4 million in a funding round led by Nexus Venture Partners to develop its approach to what it calls “Agent-Led Growth,” or ALG.

The funding round also attracted a group of technology executives and operators from the developer tools, enterprise software and artificial intelligence industries.

Among the investors are former Salesforce CTO Steven Tamm, Postman CEO Abhinav Asthana, Apollo CEO Matt Curl and DocuSign President and General Manager of Growth Robert Chatwani.

The investor group also includes executives and founders associated with GitLab, Resend, Firecrawl, Daytona, Tinyfish and other technology companies.

Lightsage argues that traditional software growth strategies are changing as AI agents increasingly become involved in discovering, evaluating and integrating software.

Historically, software companies have depended on websites, search engines, product demonstrations and registration pages to convert human visitors into customers.

AI coding agents, however, can bypass much of that process by selecting APIs, databases, authentication services and developer tools and integrating them directly into projects.

For an AI agent to successfully adopt a product, it must be able to discover the service, understand its documentation, select the correct SDK or API and complete authentication and integration steps.

Lightsage CEO and co-founder Jun Liang Lee said the internet is moving toward an environment where AI does not simply tell people which software to choose but increasingly uses the software itself.

The startup believes this shift creates a new category of growth focused on whether AI agents can understand a product and successfully complete tasks with it.

Lightsage’s platform evaluates software products by running simulations across AI answer engines and coding agents.

These simulations allow companies to examine how their products are recommended by AI systems and what happens when agents attempt to use them.

The platform can test workflows involving documentation, APIs, software development kits, command-line interfaces, MCP servers and agent skills.

When an agent fails to complete a task, Lightsage analyzes the workflow to identify possible obstacles.

Potential issues can include poor product discoverability, unclear documentation, authentication problems, API endpoint failures, SDK implementation issues or lack of MCP compatibility.

Companies can then make adjustments to their products or documentation and run the simulations again to determine whether the changes improve agent performance.

Lightsage also monitors real-world agent traffic to customer websites and documentation, providing insight into what AI systems interact with and whether those interactions result in product adoption.

The platform currently supports coding agents such as Claude Code, Codex, Cursor, GitHub Copilot and OpenCode.

The company is initially concentrating on developer-focused software, where AI coding agents are already becoming an important part of the software development workflow.

Lightsage said customers including Firecrawl, Reducto, Daytona, Rime and Tinyfish use its technology to understand how AI agents select products and where integrations break down.

For example, if an AI coding agent consistently recommends a competitor, Lightsage can recreate the same task across different products and agents to investigate whether the difference stems from visibility, documentation or the overall integration experience.

The development could also challenge conventional software analytics, which have traditionally focused on human behavior such as searches, website visits, clicks, registrations and conversions.

AI agents may discover and use software without following those traditional paths, making it harder for companies to understand how customers are actually being acquired.

Agent behavior can also vary significantly between different AI systems, meaning a workflow that works with one coding agent may fail with another.

Lightsage describes this emerging discipline as “Agent Experience,” drawing a parallel with Developer Experience, or DX, which became increasingly important as software companies competed to make their products easier for human developers to adopt.

Nexus Venture Partners partner Abhishek Sharma said AI is shifting more decision-making from people toward agents that can discover, evaluate and act on behalf of customers.

Lightsage plans to use its new funding to expand its evaluation, analytics, attribution and optimization capabilities across APIs, SDKs, CLIs, MCP servers and agent skills.

The company eventually plans to move beyond developer tools into areas such as B2B software, infrastructure and payments as AI agents become more capable of performing tasks on behalf of users.

Founded by Lee and Chief Technology Officer Sean Er, Lightsage is building its business around understanding how AI agents discover, assess and interact with software.

The startup’s funding highlights a broader shift taking place across the software industry, where companies may increasingly need to optimize not only for human users and search engines but also for AI systems capable of choosing and using products autonomously.

As AI agents become more deeply embedded in digital workflows, the ability of software products to communicate with and operate successfully through those agents could become an increasingly important competitive advantage.