Memorable GTM brain

Memorable gives AI agent builders procedural memory that captures successful workflows and retrieves them for similar tasks.

category

AI agent procedural memory

market profile

  • text: Agents repeatedly solving similar tasks can waste tool calls and tokens reconstructing steps; procedural recall targets this repetition.
  • text: Dedicated memory layers also compete with general fact memory, graph frameworks and agent-native persistence.

personas

  • title: AI Agent Engineer
    pains:
    • Agents rediscover known workflows across runs
    • Must debug whether recalled steps were actually successful
  • title: Engineering Manager, Developer Productivity
    pains:
    • Repeated coding agent work drives token and execution cost
    • Needs to evaluate memory improvements across real tasks and existing harnesses
  • title: AI Platform Lead
    pains:
    • Agent traces may contain sensitive information
    • Needs explicit control over what is recorded and sent

icp

  • text: Software teams deploying coding or workflow agents that repeatedly execute similar multi-step tasks.
  • text: Agent platform builders with traces and a measurable baseline for tool calls, task completion and token cost.

message pillars

  • text: Reuse the procedures behind successful agent runs.
  • text: Turn traces into compact workflows without unsuccessful branches.
  • text: Connect recalled workflows into plans for multi-part tasks.

objections

  • objection: Will it work with our agent harness?
    response: The site lists curated integrations and a JSON trace path for other harnesses; test your own traces.
  • objection: Does memory expose complete agent transcripts?
    response: The site says transcripts are not sent and recording requires explicit enablement; verify the actual data flow for your deployment.
  • objection: Do savings hold for our workload?
    response: Benchmark similar tasks with and without recall, including failures and memory overhead.

competitors

  • name: Mem0
    why: Provides agent memory extraction and retrieval through an SDK and hosted service.
  • name: Zep
    why: Offers graph-based memory and retrieval for agents.
  • name: Letta
    why: Provides stateful agents with long-term memory capabilities.

keywords

  • agent procedural memory
  • AI agent workflow memory
  • coding agent memory
  • agent trace reuse
  • graph memory for AI agents
  • reduce agent token usage

voice

  • developer-oriented
  • measurable
  • technical and direct

signals

  • text: An engineering team runs repeated coding or support workflows through agents and tracks token costs.
  • text: A team is evaluating persistent agent memory or instrumenting traces for agent evaluations.

stack

  • Claude Code
  • Codex
  • OpenCode
  • gbrain

pricing

model: unknown

motion

hybrid

channels

  • developer documentation
  • YC Launch
  • website

confidence

medium

Founders

  • name: Nikhil Krishnaswamy
    title: Founder/CEO
    role: CEO
    bio: CS @ Stanford Building navigation for agents @ Memorable nikhil@memorable.sh
  • name: Advaiyt Sane
    title: Founder/CTO
    role: CTO
    bio: CS @ CMU '29 Creating Memorable, turning agent reasoning into graph search.

Back to YC companies