---
**Daily Launch** · [https://dailylaunch.news](https://dailylaunch.news) · [RSS](https://dailylaunch.news/feed.xml)
---

# From portal-hopping to instant answers: HEMA's journey with MCP and Amazon Bedrock
**Enterprise AI** · Sep 26, 2026 · 3 min read
Source: AWS ML Blog — https://aws.amazon.com/blogs/machine-learning/from-portal-hopping-to-instant-answers-hemas-journey-with-mcp-and-amazon-bedrock/
### The Gist

HEMA, a 100-year-old Dutch retailer, just killed dev portal-hopping by building HAL, an internal AI assistant. They used Model Context Protocol (MCP) to pull data directly into existing workflows without forcing users to jump between tabs or manage complex AWS credentials.

### Why It Matters

For builders, this is proof that context and integration matter more than raw model intelligence. For operators, it shows how to solve tool fatigue by bringing AI to where employees already live, rather than forcing them into a new dashboard.

### Market Impact

This shifts the competitive focus from model performance to the integration layer. Companies that master seamless data connectivity via protocols like MCP will win over those just selling raw LLM access.

- Build specialized middleware that uses MCP to connect legacy enterprise silos to modern AI agents without breaking security protocols.
- Target non-tech-native industries like retail or manufacturing that have massive data gaps and high tool friction.
- Focus on 'invisible AI' that functions as a plugin or layer within existing tools like Slack or Teams rather than standalone apps.- The 'Feature vs. Product' trap: AWS or Microsoft could easily bake this specific integration logic into their core offerings, wiping out niche startups.
- Security complexity: Using MCP to bridge internal data to AI agents creates new attack surfaces if identity management isn't perfectly synced across clouds.### ELI5

Imagine you have ten different textbooks spread across ten different rooms. Instead of running around looking for answers, you get a smart friend who has already read all of them and just whispers the answer to you while you're sitting at your desk.

### Deep Dive

{"sections":[{"heading":"The Death of Portal Hopping","body":"Developers hate context switching. HEMA's HAL isn't just a chatbot, it's an efficiency layer that sits where the work actually happens. By eliminating the need to jump between portals, they're reclaiming lost productivity that usually dies in the transition between tabs."},{"heading":"The MCP Advantage","body":"The real hero here is the Model Context Protocol. It solves the massive headache of giving an AI access to specific, governed data without requiring every user to have complex AWS credentials. It makes the data accessible but keeps the security anchored in tools like Microsoft Entra ID."},{"heading":"Plumbing Over Intelligence","body":"A smart model is becoming a commodity. The real moat is in the plumbing, which is the ability to securely connect that model to specific, messy, real-world company data. HEMA is betting on connectivity, not just compute."},{"heading":"What to Watch","body":"Watch for a surge in 'internal-only' agent deployments across other legacy industries. If we see more companies moving away from standalone AI assistants toward MCP-based integrations, the integration play is officially the winner."}]}

### Key Takeaways

- **Integration beats intelligence every time** A decent model with perfect context and zero friction will beat a god-tier model that requires manual data uploading.
- **Stop building new destinations** For builders, the real money is in making AI part of existing workflows, not asking users to change their habits.
- **Watch the connectivity layer** Investors should look for companies building the 'connective tissue' between models and enterprise data, rather than just the models themselves.


[View on website](https://dailylaunch.news/articles/from-portal-hopping-to-instant-answers-hema-s-journey-with-m)