G2

Mohit S.

Frontend Developer

Small-Business (50 or fewer emp.)

11/1/2026

Business partner of the seller or seller's competitor, not included in G2 scores.

"Useful, editable shop memory for our voice-first bookkeeping prototype"

5/5

What do you like best about Meko?

Meko gives us a dedicated memory layer for Munim, our voice-first bookkeeping assistant for Indian shopkeepers. We use it to retain shop terminology and conventions across conversations, retrieve context before the model interprets a message, and let users inspect, correct, or forget remembered facts. The MCP interface fits our TypeScript backend, and separate runtime datapacks let us keep each account's shop context separate. What I like most is being able to make memory visible and editable in the product. We keep the financial ledger in YugabyteDB and use Meko for context, with clear boundaries around amounts, payment status, and authorization. Our experience so far is from building a hackathon prototype rather than operating at production scale. Review collected by and hosted on G2.com.

What do you dislike about Meko?

The main area I would improve is the developer experience for custom applications. In Munim, we handle asynchronous memory delivery, retries, bounded retrieval, and cache invalidation around corrections in our own backend. Clearer extraction-status visibility and end-to-end TypeScript examples would help. I would also welcome a data-operations CLI, typed Code Mode support for composing retrieval workflows, and broader REST coverage for memory and conversation operations alongside the existing Datapack API. These are integration improvements I would value; we have not yet evaluated long-term reliability or production-scale performance. Review collected by and hosted on G2.com.

What problems is Meko solving and how is that benefiting you?

Munim needs continuity between conversations: a shopkeeper should be able to teach the assistant a local product name or shop convention and have that context available later. A financial ledger alone does not capture this conversational knowledge. Meko provides persistent memories, extraction from eligible exchanges, and retrieval through MCP, so we can build on a dedicated memory service instead of implementing that whole layer ourselves. We use separate runtime datapacks for account-specific context and expose memory search, correction, and forgetting in the app. The benefit for our prototype is a clear way to connect remembered shop context to the interpretation flow while keeping exact balances and financial writes in YugabyteDB. We have not measured time savings or validated outcomes with real shops yet. Review collected by and hosted on G2.com.