Manicule’s newly launched answer engine optimization platform gives Shreyans Jain a broader test of the measurement principle he has applied to documentation: track what AI systems actually recommend, then connect the result to something a company can improve.
Manicule has launched an answer engine optimization platform designed to show companies how AI tools and assistants rank and recommend them to potential customers. For Shreyans Jain, Co-Founder and CTO of Manicule, the launch extends a measurement problem he had already been working through with developer-focused companies: deciding which AI marketing metrics reflect a meaningful outcome and which ones simply record more activity. The platform tracks how companies appear in AI-generated answers and gives users guidance on how to improve those rankings.
Jain mentioned a concrete example of how he applies that principle. When Greptile rewrote its documentation, Mintlify’s built-in AI agent had been successfully answering 60 percent of documentation queries. After the rewrite, that rate rose to 98 percent, a gain of 38 percentage points. For Jain, the result offered a narrow but useful test of whether the work had actually improved the documentation experience.
The metric is limited in what it captures. Mintlify’s “docs success rate” records the percentage of documentation queries its AI agent answers successfully. It does not show revenue growth, customer acquisition, or broader business performance. That limited scope is useful because it keeps the measurement close to the experience the documentation is supposed to improve.
“You have to know what success means before you automate around it,” Jain says. “Otherwise, you can make the process faster without knowing whether the outcome got better.” That principle shapes his role at Manicule, where his contribution includes the shared software platform used across the company’s client work. His measurement question stays practical: what should improve if the system is doing its job?
Documentation makes that question unusually concrete. Developer-focused companies can publish more pages, expand explanations, or use AI to accelerate production, but none of those activities automatically shows whether someone can find and use the information they need. A developer may still encounter the same missing answer after a rewrite, even if the documentation becomes longer or more polished. Jain looks for a performance measure close enough to the user experience that a change in the number represents a specific improvement.
Greptile offered that kind of measure. The documentation was rewritten, and Mintlify analytics provided the same answer-success metric before and after the work. The rate moved from 60 percent to 98 percent. The comparison gave Jain something more useful than a larger page count or faster publishing schedule: evidence that the documentation was answering more questions successfully.
The new AEO platform takes the measurement question into a less stable environment. AI assistants can produce different recommendations across models, prompts, and repeated queries. Manicule’s platform is designed to track how often a company is ranked or recommended in those answers, then give users guidance on changes that may improve their position. The company is entering a category that already includes platforms such as Profound and Peec AI.
That uncertainty is one reason Jain is skeptical of treating a single visibility score as a complete answer. “If the output goes up and the result stays the same, then you have not really solved the problem,” Jain says. “You have just made the process busier.” The easier AI makes content production, the more important it becomes to decide what a company is actually trying to change.
The same caution applies to generative engine optimization and answer engine optimization more broadly. A company can increase publishing volume or generate more AI-oriented content without learning whether assistants are recommending it more often or whether prospective customers are finding more useful answers. Jain’s approach is to keep the measurement as close as possible to the behavior the work is intended to influence.
That discipline has informed Manicule’s work across a broader group of developer-focused companies. Over the last six months, Jain says Manicule has worked with companies worth more than $1 billion combined. The engagements differ, but the question he brings to them remains consistent: what changed after the work, and can that change be measured?
Greptile’s data stops at documentation performance. Manicule has a clear measure of answer success but does not have visibility into downstream changes such as revenue or customer adoption. The documented improvement therefore says something specific about the documentation experience without establishing every business consequence that might follow from it.
“A number only helps if you know what changed because of it,” Jain says. “That is what I want to understand before I decide whether something worked.” Impressions can measure reach, publishing volume can measure activity, and visibility scores can capture one slice of exposure. The useful question for Jain is what those numbers reveal about the task the system was supposed to improve.
That is why the Greptile result remains relevant as Manicule launches a broader AEO product. The documentation metric and its intended job were closely connected: a user asked a question, the agent attempted to answer it, and the success rate recorded how often that attempt worked. Manicule’s new challenge is applying the same discipline to a noisier environment in which AI recommendations can vary from one response to the next.
Manicule has also released Margin, an open-source, agent-first documentation framework for developer-tool companies. The company built it around the view that documentation is becoming basic infrastructure for AI agents and should be easier to host and maintain without adding another paid platform. Margin is Mintlify-compatible, supports features including search, an AI assistant, API documentation, llms.txt, MCP, and agent-readable versions of documentation, and can be hosted for free or self-hosted. Manicule also writes documentation using the same internal harness its team used for Supermemory, Greptile, Reducto, and other developer-focused companies.
For Jain, the launch turns a measurement principle into a product question. AI marketing can generate more content and more dashboards with relatively little effort. The harder task is deciding what should be measured, showing users what the numbers actually represent, and helping them improve the outcome those numbers are meant to capture.