Current state audit, vendor selection, transition planning and sequencing, and more— Orium can help you build the plan to move to composable properly.

Consultation & Roadmap

From fulfillment to marketing, agentive systems help your teams automate intelligently, act faster, and scale impact through human + AI collaboration.

Agentive Operations

Cart, checkout, catalog, payments: Our experts will help you with the essential solutions you need to run your business and work efficiently across all channels.

Composable Commerce

Deliver smarter, faster, and more adaptive digital experiences with composable content systems designed to support AI and agents.

Digital Experience Platforms & CMS

Unify in-store and online journeys with composable tools and AI-driven personalization. Deliver fast, connected experiences that drive loyalty and growth.

Store, Mobile, & Unified CX

Solutions

Accelerators

Partners

We help our clients redefine what’s possible in how they work and serve their customers every day.

Customer Cases

Get the information and resources you need to grow, scale, and transform your business from the leaders in composable commerce.

Insights

Fresh thinking, practical takes, and real-world reflections on digital transformation, composability, and the future of commerce.

Blog

Turn AI strategy into action, identify high-impact use cases, plan agentic pilots, and move from idea to execution— fast.

Studio

Our Work

We help enterprises modernize with composable systems, intelligent interfaces, and data platforms designed for seamless collaboration.

About

We’re building next generation digital experiences and we want to work with people that share our passion.

Careers

Our digital platform expertise helps you tackle the challenges your business is facing.

Contact Us

Who We Are

Header Navigation Menu

Stop using messages as your agent's memory. Learn how structured state makes AI agents more reliable, efficient, and production-ready.

Building Agents? Stop Treating messages[] Like a Database

A woman presents in front of a screen displaying code, with a green diamond overlay.

Blog: Build Smarter AI Agents Image

Build Smarter AI Agents: Hero

Blog: Build Smarter AI Agents

But as my agent’s logic became more complex, cracks started to show. The model would repeat tool calls it had already run. Prompt templates ballooned with irrelevant context. I was writing functions to parse the messages[] transcript just to extract the right tool-call result when constructing prompts. It started to feel strange. There had to be a better way.

The deeper problem wasn’t the tools themselves, it was my assumption that the transcript was the state.

[LangGraph](https://www.langchain.com/langgraph) gave me a new lens. Instead of treating the transcript as a catch-all for everything the agent might need to “remember,” I could model structured state explicitly: targeted inputs for each node and isolated fields for tool results, usage history, and other reusable data. Even certain assistant responses—like internal reasoning or chosen strategies—deserve their own state fields if they’ll be reused later. This state lives in the LangGraph thread object and persists across runs, letting me feed the model smaller, more-focused prompts with exactly the context it needs, no transcript archaeology required.

This post walks through that shift and how adopting structured state makes agents more robust, reliable, and ready for real use.

Build Smarter AI Agents: Paragraph 1

Most demos turn the transcript into an accidental database. Instead, lift heavy data into structured state and keep the transcript lightweight:

    // Before
    graphState = {
      messages[] //a mix of user prompts, assistant replies, tool call requests, toolcall results
    }

    //After
      graphState  = {
    messages[], //chat flow only:  user prompts and assistant replies worth showing to the user
    toolResults,
    userPreferences,
     …other reusable data
      }

The messages[] transcript stays readable and short; the agent’s real “working memory” lives in dedicated fields contained in the graph’s state, where it’s structured, inspectable, and easy to reuse. The datatype of these dedicated fields is entirely up to you - for example a string, boolean, number, or an object - whichever best fits your needs.


Build Smarter AI Agents: Paragraph 2

Pattern Snapshot

- Lower token costs — lean prompts skip the full history of tool calls and chat.
- Easier inspection & debugging — state lives in named fields you can log and view instead of being buried in a wall of tokens.
- Predictable, controllable behaviour — explicit state makes each node’s inputs and outputs deterministic, so you direct the LLM with exactly the context it needs rather than hoping it finds the right slice in a giant transcript.
- Higher accuracy — trimming unrelated messages and stale tool results reduces noise so the model focuses on the signal.
- Simpler testing — each LangGraph node can be unit-tested with a small, explicit state payload instead of reconstructing a massive message log.

Build Smarter AI Agents: Paragraph 3

Benefits of Shifting Data Out of messages[]

### The anti-pattern

Most tutorials set up an LLM with tool calling enabled and then append both the tool-call request and its JSON result straight into the messages[] log. Now imagine a user asks, “What’s the forecast for this weekend?” — and then keeps chatting.

1. Every new user message fires the search tool again.
2. Each run appends another request and another JSON blob to messages[].
3. When the user finally says, “So, should I pack an umbrella?”, the LLM must sift through the entire transcript, decide which of many search results is current, and use that data correctly.

That means longer prompts, more tokens, and fragile reasoning.

### A better way

Lift the data into state:

    graphState["searchResults"] = results_json

…then inject only what each node needs:

    Here are the search results:
    {{searchResults}}

Now each user message can reuse or refresh __graphState.searchResults__ without bloating the transcript, and the LLM always sees the right data in a clean prompt.

LangGraph makes this possible, but it doesn’t force it. Making the leap is up to you.
The same logic applies to reusable assistant insights: summaries, recommended actions, parsed conclusions. Store them in state, not in the messages[] array.

Build Smarter AI Agents: Paragraph 4

Case Example: Search Results Bloat

- Treat LangGraph thread state as working memory — structured, inspectable, portable.
- Create clear fields: searchResults, userProfile, pricingData, strategySummary, etc.
- If an assistant message will guide later steps, lift it into state instead of leaving it in the transcript.
- Keep messages[] clean: user text and final assistant replies only.
- Construct prompt templates with just the relevant state needed to best support the LLM in its response.

Build Smarter AI Agents: Paragraph 5

Implementation Tips

Using messages[] for everything works— for a while. But as agents grow, this pattern becomes brittle, opaque, and expensive.

Treat the transcript as a transcript, not a brain. Structured state preserves what matters—tool results, assistant reasoning, decisions—bringing clarity, determinism, and real-world reliability.


Build Smarter AI Agents: Paragraph 6

Final Thought

Build Smarter AI Agents: Use Structured State, Not Just messages

Traditional approaches to change management weren’t working before. AI just makes the gaps impossible to ignore.

Why AI Forces a Rethink of Change Management


A robotic hand interacts with a computer displaying code, with an orange star pattern overlay.

Blog: Why AI Forces a Rethink of Change Management Image

Why AI Forces a Rethink of Change Management

Blog: Why AI Forces a Rethink of Change Management

Teams adopted software without fully using it. Workarounds became the norm. Strategic goals like efficiency, insight, and speed were quietly diluted by partial adoption and quiet resistance. Change happened, but results were never fully realized even as millions were spent on platforms and integrations— not because the tech failed, but because the human transition was underpowered or ignored.

Now, AI is raising the stakes on transformation.

Intelligent agents don’t just streamline tasks, they change how decisions are made, how roles are structured, and where control lives. They act more like semi-autonomous teammates than tools, influencing outcomes in ways that feel personal. Adopting them is less like rolling out software and more like adding a new hire— one who never sleeps, knows more than most, and in the eyes of many, wasn’t invited.

That kind of change hits differently. It needs to be treated like a true team shift, with clarity, shared expectations, and trust. Without that, resistance sets in.

Why AI Forces a Rethink of Change Management: Paragraph 1

Traditional change management breaks down in the face of AI and intelligent systems. It assumes change happens after the system is built: a rollout phase, a round of training, and a handoff to enablement. But with AI and agents, the hard part isn’t getting people to use a system. It’s designing the system so people want to use it, because it fits how they work, earns their trust, and clarifies (not confuses) their role.

That can’t be retrofitted. Change has to shape how the thing is built from the beginning.

AI-led transformation fails when people don’t trust it, don’t understand their role, or can’t see how it improves their work. Those foundations—clarity, confidence, trust—must be baked in early. No training or comms sprint can fix what wasn’t considered upfront.

AI doesn’t just change tools—it changes decisions, roles, and control. That brings fear. And fear shows up long before launch day— in uncomfortable meetings, pronounced silences, and hesitation to commit to decisions. If you don’t design for trust throughout, you’ll spend your time managing resistance after the fact.


Why AI Forces a Rethink of Change Management: Paragraph 2

Change Management is More Than a Phase, it’s a Core Requirement

Treating change as a core constraint doesn’t mean adding extra meetings or documentation. It means asking better questions at key moments, ones that shape the build itself. The earlier these questions surface, the fewer surprises downstream.

__What’s changing for the people this system touches, and how materially?__
If the answer is “a lot” (and it probably is) treat those changes like features: scoped, validated, and tracked. Map old vs. new workflows. Run role clarification workshops before development starts. Build in readiness checkpoints. Create frontline feedback loops that shape the system in real time.

__Who’s losing control, visibility, or decision rights?__
These people are often gatekeepers and their quiet resistance can sink adoption. Identify friction early: who used to approve something that the agent now handles? Bring them into plan reviews. Document what’s shifting, and clarify how governance will work going forward. Don’t gloss over the power dynamics, deal with them directly.

__Where is the team skewing negative and how do we rebalance?__
With AI and agents, teams often fixate on what might break, who might lose, or why it won’t work. That caution has a place, but it can drown out possibility. Run pre-mortems and “pre-successes.” Document risks and unrealized benefits. Make sure the loudest voices in the room aren’t just the most skeptical.

__How will this land in day-to-day work?__
Deploying agents isn’t just a technical implementation, it’s a form of org design. You’re introducing a new actor into the system, and that requires the same rigor you'd apply to defining a human role. What’s the agent accountable for? Where does it hand off? What does collaboration look like? Use tools like RACI models or workflow diagrams and adapt them to include agents. When expectations are explicit, teams are far more likely to engage rather than resist. Uncertainty is often scarier than the change itself.

__How transparent are we being about what’s changing and what’s still uncertain?__
People don’t need false certainty. They need honesty. Leaders should acknowledge that AI-driven change is complex, unpredictable, and often uncomfortable. That includes naming the fears, recognizing the workload, and admitting when all the answers aren’t there yet. What builds trust isn’t perfection, it’s a clear commitment to communicating openly, often, and early.


Why AI Forces a Rethink of Change Management: Paragraph 3

Leading Change in the Age of AI

At Orium, we don’t treat change management as a separate service. We treat it as a mindset that informs how we build. That mindset starts early: during discovery, during architecture, during delivery. We’re not layering change on top of the system. We’re shaping the system so change can succeed.

We surface role impacts early and define them explicitly— not just for humans, but for agents too. We document what’s changing, for whom, and how success will be measured from a people perspective, not just a platform one. And we make space for transparent, ongoing communication. We acknowledge the fear, the friction, and the uncertainty, and we show how feedback is actively shaping the build.

We stay alert for early friction—unclear ownership, quiet pushback, or workflow cracks that get glossed over in planning but surface during execution. These aren’t annoyances to smooth over later, they’re signals. And we treat them that way, adjusting course before small problems become cultural resistance.

Most importantly, we partner directly with leaders as they navigate this change. Sometimes that means offering an outside perspective. Sometimes it’s helping execs make sense of what they’re hearing from their teams. And sometimes, it’s just being a sounding board. Change is hard—even when it’s wanted—and we help leaders carry that load.

AI projects come with big expectations and a degree of uncertainty. At Orium, we help leaders manage both by making change part of how the work gets done, not an afterthought.


Why AI Forces a Rethink of Change Management: Paragraph 4

How Orium Enables Adoption from the Start

AI brings out a lot of feelings: good feelings like enthusiasm and curiosity, but also negative feelings like fear, uncertainty, and even grief. That’s always been part of change management, but with AI, it’s amplified, because the change isn’t limited to a new tool or a single workflow. It’s all around us. People are navigating it in their jobs, in the news, and in their careers. They’re not just reacting to this project, they’re reacting to everything AI represents in their lives.

As AI-supported development makes it faster to build and deploy solutions, the work of transformation shifts even more heavily to people, workflows, and adoption. The tech can move quickly, but it’s the human system that needs the most attention— and the most intention.

Change management in the age of AI isn’t a checklist. And it’s not just a comms plan or a training module. It’s a mindset that has to shape how we scope projects, define roles, make decisions, and measure success. Teams that treat it that way move faster, adopt more fully, and avoid the quiet resistance that stalls transformation.

If you want adoption, build for it. If you want people to embrace agents, help them see where they still belong. If you want real, lasting change, put the people in the system at the center. Leave them out, and the system fails. It’s as simple as that.


Why AI Forces a Rethink of Change Management: Paragraph 5

Change Isn’t a Phase. It’s the Work.

Traditional change management no longer works. Learn why AI raises the stakes and how to design adoption into every stage of your transformation.

Why AI Demands a New Approach to Change Management

How smart companies are evolving with agent-powered delivery models, and what it takes to lead in the new era of intelligent services.


AI Isn’t Killing Services, It’s Redefining Them


Three people have a discussion in a modern meeting room, with a large monitor in the background and a gold star pattern overlay.

Blog: AI Isn’t Killing Services, It’s Redefining Them Image

AI Isn’t Killing Services, It’s Redefining Them

Blog: AI Isn’t Killing Services, It’s Redefining Them

There’s no getting around it: routine, low-value work is disappearing. Tasks such as setting up code components, tweaking UIs, or writing QA tickets no longer demand hours of manual effort. AI and agentic tools now handle image validation, merchandising automations, and reporting with minimal oversight. It’s not that the volume of this work has gone away, but rather the human effort required to do it has dropped dramatically.

As these rote tasks fade, services teams will become leaner and more focused. Large, junior-heavy teams are being supplanted by skilled specialists paired with intelligent agents and client-side experts to deliver higher-quality outcomes, faster.

That means velocity and precision are the new competitive edge. Rather than throwing bodies at problems, modern service providers are expected to bring smart systems, accelerators, and agent-powered workflows that cut down on cycle times and raise the quality bar. Clients aren’t interested in buying hours anymore, but they will buy speed, certainty, and outcomes.

The firms that thrive in this new reality won’t be the ones that resist change or try to protect the old margins. They’ll be the ones that embrace AI as a co-pilot, invest in upskilling their teams, and reorient their delivery models around value, not volume. 


How Services Are Changing and What the Future Looks Like

Clients want accelerated development velocity. They ask for end-to-end automation of repetitive tasks so their teams can focus on strategy and innovation, and they look for smarter project management routines, AI-assisted QA, and sales proposals that reduce overhead. So we recalibrated our routines, our offers, and our approach to be able to give them what they need.

By integrating custom accelerators, automated testing frameworks, and agentic workflows directly into project delivery, we help customers reach their goals without the large teams (and larger billed hours) of a bygone era. 

By co-delivering with client teams, we’re able to keep everyone aligned and moving forward, harnessing specialized talent and modern project governance that can flex for rapid innovation. 

And by building on a heritage of Agile methodologies and composable technology foundations, we continue to refine our approach and offerings to suit the dynamically evolving needs of enterprise organizations.

In short, we’re doing what we’ve always done: taking a customer-centric approach to how we operate. And in the age of AI, that means embracing automation, assistive technology, and agents.


Where we are changing the status quo and what value we deliver to clients

We recognized early that AI adoption requires intentional change. And it’s not just a change in tooling; this is an end-to-end change management lift that cuts deep into habits formed over decades of services work. To tackle this, we’ve built a whole-systems structure for our teams.

Internally, there are three core components:

__1. Tool training and testing__: We held monthly leadership workshops and trained teams in new tooling. We gave our whole organization access to the best AI assistants with guardrails that allowed for safe experimentation, streamlining our security assessment criteria for faster testing of new tools.

__2. Embedding AI and new routines__: We built internal agents for everything from daily project routines and QA support to reporting and company engagement, ensuring AI became a part of the fabric of our daily work lives so everyone was familiar and comfortable with the new technology.

__3. Org structure and role redefinition__: We invested time, energy and care into mapping the roles and functions we need, redefining what key roles will look like in 2-3 years time as these tools and processes become standard, and creating rubrics that offer clarity and a roadmap for long-term success so team members can see and understand the changes and work with managers to upskill along the way.

Services, in the end, is all about people— even as AI-powered agents enter the collaborative workforce. It’s why intentional planning and real change management are essential, ensuring people are at the core of the technology evolution, and not the other way around.

But internal changes are only half of the equation (and in some ways, the easier half). 

On the client side, timelines, attitudes, and expectations are much less within your control, but the shift here needs to be as intentional and focused as internal all the same. With a foundation in composable, we had an advantage: our core skillset and offerings were built around the ability to adjust and adapt as markets evolve. But our commercial models needed to evolve to reflect those changes, so we’ve shifted to fixed-scope, outcome-driven engagements instead of buckets of hours with uncertain results. By tying fees directly to outcomes, we’re able to give clients the path to ROI they need when investing in new technologies. 

My advice for any company on this path is to plan for company-wide transformation. Invest in training, give teams time to experiment, and align incentives with outcomes rather than hours.


How Orium Embraced the Agentic and AI World and Advice for Companies

This is a key moment in our industry. The change is happening, it’s happening fast, and it is big. How you approach this moment will determine your long-term success, there’s no way around it. But services firms are built on the ability to navigate big, transformational change, so treat your firm the way you would any client.

Begin by defining your North Star and establish clear success metrics for AI adoption. Put the tools in your team’s hands with as much liberty as you can safely grant them, testing agents and AI tools on real tasks. Explore demos, read case studies, and immerse yourself in what AI can do. Most importantly, talk to your teams and bring them along on the journey. Ask for their input, listen to their concerns, and work with them to co-create your future.

When in doubt, speak to others further ahead of you. Call Orium, talk to peers, go listen to some AI software pitches. Once you see AI in action, you can’t unsee its impact. The key is hands-on experience.

AI is not killing services. It is the catalyst for a new era of high-value, agent-powered offerings. At Orium, we’re leading that transformation. If you’re ready to reshape your services for the AI age, start now and partner with experts who have already walked the path.


Where to Start

Discover how agent-powered services are evolving with AI and why smart firms are shifting from hours to outcomes in the new services era.

AI Is Redefining Services, Not Replacing Them

Popular Blogs

Popular Blogs - Blog Landing Page

Recognized among 40 leading providers for composable expertise, industry depth, and proven business outcomes.

Orium Named to Forrester’s Q3 2025 Commerce Services Landscape Report


Orium Named in Forrester Commerce Services Landscape Q3 2025

Orium Named to Forrester’s Q3 2025 Commerce Services Landscape Report

Blog: Orium Named to Forrester’s Q3 2025 Commerce Services Landscape Report

Not every provider makes the cut. Orium’s inclusion underscores our role as a specialist partner, helping enterprises move beyond legacy systems and embrace modern, composable architectures built for change.

Where Orium stands out:
- __Composable commerce expertise__. We’ve helped some of the most complex enterprises transition to flexible, future-ready systems.
- __Depth in B2B2C and healthcare__. Few partners can match our record in industries where scale, regulation, and multi-channel experiences collide.
- __Commitment to outcomes__. We measure our success by our clients’ results—growth, efficiency, and long-term adaptability.

“As the market evolves, enterprises need more than implementation support—they need a partner who can de-risk transformation and build systems designed to adapt,” said [Jason Cottrell](https://www.linkedin.com/in/jasoncottrell/), CEO of Orium. “This recognition reinforces our belief that adaptability will define the future of commerce, and we’re proud to be at the forefront of that shift.”

[Download the full Forrester Commerce Services Landscape report here.](https://www.forrester.com/report/the-commerce-services-landscape-q3-2025/RES186235)

*Forrester does not endorse any company, product, brand, or service included in its research publications and does not advise any person to select the products or services of any company or brand based on the ratings included in such publications. Information is based on the best available resources. Opinions reflect judgment at the time and are subject to change. For more information, read about Forrester’s objectivity here .*

Article continued

Orium is recognized among the top 40 providers in Forrester’s Q3 2025 Commerce Services Landscape for composable expertise and business outcomes.

2025-09-16

Orium Named to Forrester’s Q3 2025 Commerce Services Landscape Report

Popular Articles

Building Agents? Stop Treating messages[] Like a Database

Why AI Forces a Rethink of Change Management

AI Isn’t Killing Services, It’s Redefining Them

Solutions

Our Work

Our Team