diff --git a/apps/web/src/app/about/page.tsx b/apps/web/src/app/about/page.tsx
index 78ed3185..8d8cde40 100644
--- a/apps/web/src/app/about/page.tsx
+++ b/apps/web/src/app/about/page.tsx
@@ -17,28 +17,25 @@ export default function AboutPage() {
         <Markdown>
           <h2>Setting the stage</h2>
 
+          <p>Building reliable agents is hard.</p>
+
+          <p>But building them without the right tools is even harder.</p>
+
           <p>
-            It's 2025. LLMs are still scaling. AI agents are just starting to
-            take off. MCP is exploding. The singularity looms. Shoggoth looks
-            back on us from the near future and gives a sly smile that doesn't
-            quite reach His inhuman eyes.
+            MCP is really promising, but it's still early and the ecosystem's a
+            bit of a mess.
           </p>
 
           <p>
-            In this increasingly AI-native world, what's the best way for
-            "normal" software engineers like us to stay relevant and provide
-            value for both our families as well as our future AI edgelords?
+            Add to that the fact that most MCPs are just thin wrappers around
+            REST APIs – which works, but is far from ideal in terms of context
+            efficiency and instruction following.
           </p>
 
           <p>
-            Well, I don't know about you, but contributing to foundation models,
-            AI alignment, and AGI research all seem incredibly sexy but are also
-            a lil outside my areas of expertise.
-          </p>
-
-          <p>
-            That's why Agentic is focused solely on building value at the LLM
-            tool calling layer. MCPs are a great example here.
+            The best LLM tools require a fundamentally different UX than REST
+            The best agents require their tools to be optimized for LLM usage
+            with a fundamentally different UX than REST APIs.
           </p>
 
           <p>
@@ -48,7 +45,7 @@ export default function AboutPage() {
             <span className='font-semibold'>
               providing access to high quality tools that are specifically
               designed and optimized for agents will become increasingly
-              important over time
+              important
             </span>
             .
           </p>
diff --git a/apps/web/src/components/example-usage-section.tsx b/apps/web/src/components/example-usage-section.tsx
index 6a72b495..3e84bbe9 100644
--- a/apps/web/src/components/example-usage-section.tsx
+++ b/apps/web/src/components/example-usage-section.tsx
@@ -59,8 +59,7 @@ export async function ExampleUsageSection() {
           <span className='font-semibold'>MCP servers</span> as well as simple{' '}
           <span className='font-semibold'>HTTP APIs</span>. MCP is important for
           interop and future-proofing, whereas simple HTTP POST requests make
-          tool use easy to debug and simplifies integration with existing LLM
-          tool calling.
+          tool use easy to debug and simplifies usage with LLM tool calling.
         </p>
       </div>
     </section>
diff --git a/apps/web/src/components/mcp-gateway-features.tsx b/apps/web/src/components/mcp-gateway-features.tsx
index 9d9e3e29..0110b37a 100644
--- a/apps/web/src/components/mcp-gateway-features.tsx
+++ b/apps/web/src/components/mcp-gateway-features.tsx
@@ -66,7 +66,7 @@ const mcpGatewayFeatures: FeatureData[] = [
         <span className='font-semibold'>MCP servers</span> as well as simple{' '}
         <span className='font-semibold'>HTTP APIs</span>. MCP is important for
         interop and future-proofing, whereas simple HTTP POST requests make
-        tools easy to debug and simplifies integrating with LLM SDKs.
+        tools easy to debug and simplifies usage with LLM tool calling.
       </>
     ),
     icon: CheckCheckIcon,