Reinventing Enterprise AI: Key Takeaways from AWS re:Invent 2025

December 12, 2025

Simon Tindal

By Simon Tindal, CTO, Smart Communications

This year’s AWS re:Invent made one thing clear: We’re entering an era defined not just by AI adoption, but by AI accountability, observability, and real-world value.

From keynotes to analyst sessions to roundtables with fellow CTOs and CIOs, a few clear themes emerged about where enterprise AI is heading next, and what leaders should be thinking about as they plan for 2026 and beyond.

Here are my key takeaways:

1. Agentic AI needs guardrails, not just power

A lot of the buzz this year centered on intelligent agents and their ability to take action, not just generate content. The big question that kept coming up was, “How do we trust agents to act autonomously without losing control?”

AWS introduced new capabilities focused on observability. In practice, that means being able to see what agents did, how they reasoned, and where their behavior drifted from what was expected. This shift from “can we build an agent” to “can we understand and explain what it just did” is a very important one.

For enterprises, especially in regulated industries, this is where AI maturity will be won or lost. It’s not enough for agents to be clever. They need to be auditable, predictable, and accountable.

2. Don’t just automate the status quo. Reimagine the workflow.

A recurring message across the event, especially highlighted in a strong session from Forrester, was that simply recreating your current processes with AI is a missed opportunity.

Most organizations begin with the idea of “let’s take this process and make it faster.” That can be a good starting point, but it’s not where the real value is. The more impactful questions are:

- If we were starting fresh, without the constraints of legacy processes, how would we design this today?
- What could agents do that we would never expect a human to manage manually?

This mindset shift, moving from basic automation to genuine reinvention, is what will transform onboarding, claims, servicing, and other complex workflows. It’s not about shaving a few seconds off a task. It’s about changing the experience entirely.

3. Pricing and ROI are shifting toward outcome-based value

Another theme that surfaced repeatedly was the need for AI pricing that feels predictable. There was a lot of discussion about a long-term goal within the industry: moving from consumption-based pricing toward outcome-based pricing.

That could mean tying value to improved NPS, shorter onboarding times, higher first-contact resolution, or other business metrics that resonate with executives outside of IT.

This evolution will take time, but it reflects an important reality. AI investments are no longer purely experimental. Leaders want to understand:

- What outcomes will this drive?
- When will we see value?
- How predictable is the spend compared to the benefit?

Any AI strategy that cannot answer those questions in straightforward business language will struggle to earn long-term support.

4. AI is now everyone’s job, not just IT’s

One theme that came up again and again, both in roundtables and in hallway conversations, is that AI can no longer sit solely within technical teams.

For AI to succeed at scale, it needs to be:

- Expressed in simple, business-friendly terms
- Built into the tools frontline teams already use
- Governed in a way that is transparent and easy to explain

AI may be technically complex, but the experience of using it should feel straightforward to the people closest to the customer.

5. Empathy still matters: the Kando principle

One session from Sony really stuck with me. It centered on the idea of Kando, a Japanese word that captures the feeling of being deeply moved or inspired. It’s the sense of awe you get when something delivers exceptional quality or significance.

The core message was clear. Even as automation and intelligent agents become more capable, emotion and empathy still matter. This is especially true in customer communications, where moments of clarity, reassurance, and understanding can define the experience.

Whether someone is applying for a loan, updating a policy, submitting a claim, or handling a difficult situation, how an interaction makes them feel is often just as important as the information being exchanged.

AI can help scale responsiveness and personalization, but it should never remove the human element. The goal is to design experiences that are both efficient and emotionally intelligent.

6. SmartIQ Agent at re:Invent: modernizing legacy workflows

One of the highlights of the week was seeing SmartIQ™ Agent featured in the session, “Migrate and Modernize Legacy Workflows to Intelligent Agents.”

The session used SmartIQ Agent to show how organizations can:

- Move away from static, form-based interactions
- Use conversational experiences to collect structured data
- Apply existing business rules and validations in the background
- Modernize legacy workflows without giving up compliance or oversight

In simple terms, SmartIQ Agent turns complex forms into natural, guided conversations, while still ensuring the data is complete, compliant, and ready for downstream systems.

Watch the session

The full PEX304 session is now available on AWS’s YouTube channel. It’s worth watching in full for the broader discussion on intelligent agents. You can also skip to around the 60-minute mark to see the SmartIQ segment:

7. What this means for our collaboration with AWS

re:Invent is always useful for the announcements, but the real value is the alignment it creates.

A few things stood out this year:

- We’re building on the right foundations with AWS and Amazon Bedrock, especially as observability, governance, and model flexibility become more important.
- The industry’s direction of travel aligns closely with our focus on practical, transparent, and scalable AI.
- There is growing interest in packaged, multivendor solutions on AWS Marketplace, which opens up new opportunities for joint innovation.

Our goal remains the same: help organizations modernize customer and citizen interactions with AI that are powerful, compliant, and grounded in real-world workflows.

Looking ahead

I left re:Invent feeling both excited and clear-headed. Excited because the progress around agentic AI, observability, and outcome-based thinking is accelerating quickly. Clear-headed because the path forward for enterprises is becoming more concrete:

- Treat AI as a core business capability
- Build for trust, transparency, and empathy
- Rethink processes rather than simply automating existing ones

We will continue building in that direction, working closely with AWS, our partners, and our customers to turn AI into measurable and sustainable value.