I’m Ian, founder of Memvor.
Memvor is an AI project command center for solo founders. It keeps project plans, AI execution, blockers, and the decisions that still need a founder on one clear path forward.
I have 10 years of product experience, including six years focused on Baidu Maps in-car navigation, intelligent cockpits, and automotive large language models.
I started Memvor because the work kept moving while its context kept breaking.
After leaving my job, I began using Vibe Coding to move several projects forward. Two recurring problems made me stop treating the friction as an isolated inconvenience.
Parallel work made completed work hard to pick up again.
AI coding and research often involved waiting, so I would switch to another project or worktree. When I returned, even a completed task could be difficult to place: where had it reached, what had changed, and why did an apparently relevant result no longer make sense without its context?
New ideas were saved, then effectively lost.
Ideas that appeared while I waited went into drafts, bookmarks, likes, or saved posts. Finding one later meant digging through a long history; validating it meant reconstructing the information and designing the test all over again.
I am not observing this problem from the outside.
I encounter it as a solo founder using Vibe Coding while multiple projects and worktrees move in parallel. Memvor began with the context, progress, and decisions I was personally losing between those threads—not with an imagined user profile.
My own problem came first. Research and real use followed.
The evidence has developed in three distinct stages. Each stage supports a different claim, so I keep them separate.
- 01 · Build
The first-hand problem shaped the first product.
I started by building around the context, progress, and information gaps I experienced myself. Memvor’s foundation and core product direction grew from that work.
- 02 · Validate
Research tested whether the problem was shared.
A targeted concept survey produced 141 valid responses. It confirmed the same symptoms around recovering project context, stalled parallel work, and ideas that were saved but never revisited.
- 03 · Refine
The product entered limited real-world use.
Memvor’s bilingual product site and interactive Demo Tour are live. The first users are now using the product and working with me to refine it while the overall system remains in integration.
- 141
- valid survey responses
- 40
- had to reopen chats, documents, or code to recover project status
- 30
- had projects stall because too many were moving at once
- 56
- saved a new tool or workflow and never opened it again
- 101
- expressed at least some interest in the concept
- 51
- left contact details for the Alpha—research leads, not current usage
Ten years of turning real scenarios into product decisions that can be tested.
I have 10 years of product experience, including six years focused on Baidu Maps in-car navigation, intelligent cockpits, and automotive large language models. At Baidu, I contributed to product definition for Baidu Maps In-Car V20 through V21. Starting from real in-car scenarios, I defined product capabilities for large-model-powered retrieval, failure fallback, and outcome evaluation. I also built a vehicle-map positioning evaluation system from zero to one and used feedback from drivers and customers to help improve satisfaction across positioning, retrieval, and lane-level navigation by 15%–30%.
That work shaped a product method I still use: start with real problems and scenarios, turn ambiguous feedback into priorities, a roadmap, and product capabilities, then validate each iteration through POCs, evaluation metrics, and continuous feedback. Memvor follows the same path—from my own pain, through research and product work, into limited real-world use.
Memvor is in limited use and still being refined.
The first users are actively using Memvor and working with me to improve it. The overall product remains in integration, and access will expand gradually.
I have a long-standing interest in AI agents, Vibe Coding, automation, and exploring new products. Running is how I temporarily clear my head and reorganize a problem.
More about me on ianymu.com