diff --git a/AI_AGENT_INTEGRATION.md b/AI_AGENT_INTEGRATION.md
index e9538ec42..daac6cbfb 100644
--- a/AI_AGENT_INTEGRATION.md
+++ b/AI_AGENT_INTEGRATION.md
@@ -162,7 +162,7 @@ REFLEXIO_API_KEY="..."
 ```
 
 Implementation rule: if Reflexio is unavailable, the agent must continue
-normally. Treat Reflexio as a best-effort learning layer, not as a dependency
+normally. Treat Reflexio as a best-effort learning platform, not as a dependency
 that can break user work.
 
 ## Configure an LLM Provider
diff --git a/benchmark/gdpval/RESULTS.md b/benchmark/gdpval/RESULTS.md
index ab5acd31b..8653c1db6 100644
--- a/benchmark/gdpval/RESULTS.md
+++ b/benchmark/gdpval/RESULTS.md
@@ -5,10 +5,10 @@
 Modern AI agents already try to improve themselves between runs: they
 cache skills, remember prior plans, and get faster on tasks they have
 seen before. This raises a pointed question: **is there still room
-for an external self-improvement layer to add measurable value on top
+for an external learning platform to add measurable value on top
 of an agent that is already self-improving?**
 
-We test **Reflexio**, a layer that reads a completed run of a task
+We test **Reflexio**, a learning platform that reads a completed run of a task
 and produces a concrete, copy-pasteable recipe for the next run of
 the same task, on **five real knowledge-work tasks** from the public
 OpenAI GDPVal benchmark. These five tasks were selected because the
diff --git a/client_dist/README.md b/client_dist/README.md
index 192a24f3f..76693ad40 100644
--- a/client_dist/README.md
+++ b/client_dist/README.md
@@ -5,7 +5,7 @@
 [![License: Apache 2.0](https://img.shields.io/badge/license-Apache%202.0-green)](https://github.com/ReflexioAI/reflexio/blob/main/LICENSE)
 [![Downloads](https://static.pepy.tech/badge/reflexio-ai/month)](https://pepy.tech/project/reflexio-ai)
 
-The official Python SDK for [Reflexio](https://www.reflexio.ai/) — the adaptive memory layer for AI agents. Reflexio automatically extracts user profiles, generates playbooks, and evaluates agent performance from conversation data. This client provides type-safe, sync-first access to the full Reflexio API. For source code and contributions, see the [GitHub repository](https://github.com/ReflexioAI/reflexio).
+The official Python SDK for [Reflexio](https://www.reflexio.ai/) — the learning platform for AI agents. Reflexio automatically extracts user profiles, generates playbooks, and evaluates agent performance from conversation data. This client provides type-safe, sync-first access to the full Reflexio API. For source code and contributions, see the [GitHub repository](https://github.com/ReflexioAI/reflexio).
 
 ## Table of Contents