CrossCountry Consulting recently hosted a webinar on Coupa AI that drew registrations from 147 companies. Of those, 83% were already live on Coupa. When we polled attendees on where their organization stood on AI maturity, most selected a familiar answer: manual processes.
That response is worth pausing on. Coupa AI and Navi agents lie dormant inside most Coupa environments, available and ready to work. The interest is genuine. The activation is not. Most finance and procurement teams have the capabilities available but aren’t yet putting them to use.
That gap, between AI availability and AI activation, is the real challenge. If your team is live on Coupa but hasn’t engaged the AI features already in your environment, you aren’t behind. You’re in good company, and the path forward is more accessible than it may appear.
What is Coupa AI, and why does it matter now?
Coupa AI is the set of embedded artificial intelligence capabilities built directly into the Coupa platform, branded collectively as Coupa Compose.
It includes Navi agents, Agent Studio for building custom agents, and tools to connect Coupa to external AI solutions like Tonkean. Tonkean serves as an intake and orchestration layer, applying AI to capture procurement needs and orchestrate workflows and systems across the source-to-pay process
This matters now because the features are no longer theoretical. Navi agents and broader set of Compose capabilities were demonstrated broadly at Coupa’s Inspire conference. Most established Coupa customers already have access. The question has shifted from “does this exist” to “why haven’t we turned it on.”
Industry data backs up what CrossCountry sees in the field: McKinsey research on agentic AI in procurement found that organizations can move from prototype to pilot in weeks and from pilot to scale in under a year, once activation actually starts.
A Navi agent in Coupa performs a specific task using a language model, memory, and, as needed, direct access to your Coupa data, knowledge documents and external tools. Conversational agents respond when a user asks something in the chat window, similar to asking a question and getting an answer that may include documents, links, etc.
Autonomous agents run on their own, on a schedule or trigger, with no prompting required.
Why are most Coupa customers not using these features yet?
The barrier isn’t curiosity, and it isn’t the technology. It’s not knowing where to start. Given the multitude of AI options available and the rapidly changing landscape, many companies have been unsure about where to start. Coupa presents a practical way to quickly start the journey and expand in scope and sophistication over time.
Procurement, AP, and IT teams are the ones exploring Coupa AI today, often as small, self-organized groups. Executive sponsorship remains rare, a pattern that lines up with EY’s 2025 Global CPO Survey finding that 80 percent of CPOs plan to deploy generative AI within three years, while only 36 percent have a meaningful implementation today. That gap between intent and execution changes where adoption actually begins: with the people running day-to-day workflows, not a top-down transformation mandate.
A few patterns show up consistently in the field:
- Teams don’t know what’s already available. Navi agents ship inside the Coupa UI, but many admins haven’t explored Agent Studio or the out-of-the-box agent library.
- Governance questions stall momentum. Approval thresholds, audit trails, and “what is this agent allowed to do” are common blockers, and they’re reasonable ones. Gartner research shows that 74 percent of procurement leaders say their data isn’t yet AI-ready, so this hesitation is widespread, not unique to any one organization.
- Early experiments don’t get past the pilot. A team tries one agent, doesn’t see immediate lift, and moves on.
None of these are technology problems. They’re sequencing problems.
How should you sequence Coupa AI adoption?
Treat activation in three stages: crawl, walk, and run. Each stage builds on the trust and familiarity established in the one before it.
- Crawl: Turn on out-of-the-box Navi agents. This is the lowest lift available. Conversational agents like the Analytics Agent, Document Discovery Agent, and Supplier Discovery Agent work entirely inside the Coupa UI with no custom build required. Every agent action generates an activity log, so your team can see exactly what the AI did and why, before trusting it with more.
- Walk: Build one or two custom agents in Agent Studio. Once your team trusts the audit trail, use the low-code Agent Studio platform to build agents tied to a specific workflow. A Sourcing Consolidation Agent can scan open requisitions, group overlapping demand, and draft a sourcing event for buyer review. A Discount Optimizer Agent can evaluate early-payment terms against a defined hurdle rate and flag which invoices are worth paying early.
- Run: Connect to external AI and Orchestrate across workflows. Tonkean chains multiple agents into a single end-to-end process, passing context between steps instead of routing everything through email or manual triage. Tonkean can also extend beyond Coupa to orchestrate across wider eco-system of S2P systems and AI solutions connected via Coupa’s Model Context Protocol that connects Coupa to your broader enterprise AI stack, when that need arises.
- This orchestration discipline is the same approach CrossCountry Consulting applies in its Auto Accrual AI solution, which uses the same Coupa data foundation to automate month-end accruals. This removes manual effort at one of the most time-sensitive points in the financial close.
Most organizations stall at step one, not because step three is too hard, but because nobody took step one.
How does CrossCountry approach Coupa AI activation?
CrossCountry Consulting has been a Coupa Premier Partner since 2014, with practitioners who have worked inside the platform since release six. As the highest-volume Coupa implementer, CrossCountry brings unmatched depth to AI activation: this is a hands-on challenge, not a documentation problem.
The approach starts inside your actual Coupa instance, not a generic sandbox. During a recent client session, a Requisition Creation Agent was built and tested in under a day, a timeline that reflects how quickly these tools move once someone has done it before.
Every agent recommendation ties back to a specific value lever: e.g., sourcing effectiveness, spend control, supplier compliance, or payment timing. That link between the agent and the metric it moves is what turns a demo into a business case.
What should you look for in an activation partner?
Look for a partner who works inside your live Coupa environment, not just around it in slide decks. The right partner should be able to show you a working agent, not just describe one.
Prioritize partners who can map each Navi agent or Agent Studio build to a measurable outcome, whether that’s spend savings, cycle time reduction, or working capital release. This kind of mapping is core to CrossCountry’s Procurement & Cost Transformation approach: activation without a value connection is just a feature tour.
Confirm the partner trains your internal team as part of the engagement. The goal isn’t a one-time build. It’s an admin team that can maintain and expand the agent library after the engagement ends.
Frequently asked questions
What Coupa AI features are available right now?
Existing Coupa customers already have access to Navi agents, both conversational (like the Analytics Agent and Supplier Discovery Agent) and autonomous (like the Discount Optimizer Agent). Agent Studio, for building custom agents, and Tonkean, for orchestration, are also generally available..
What is a Navi agent, and how do I use it?
A Navi agent is a task-specific AI tool built into Coupa. Conversational agents respond when you ask a question in the chat window. Autonomous agents run automatically on a schedule or trigger, such as creating a payment batch every Monday, with no prompting needed.
Do I need IT to enable Coupa AI features, or can procurement and AP do it ourselves?
Out-of-the-box Navi agents typically require no IT involvement and can be activated directly by procurement or AP admins inside the Coupa UI. Business users can build custom agents using the low code Agent Studio, though more complex builds (e.g., incorporating API connections) benefit from technical support
Which Coupa AI features are worth turning on first?
Start with out-of-the-box conversational Navi agents, since they carry the lowest implementation lift and build trust through visible audit trails. From there, move to one or two custom agents tied to a specific, high-value workflow like sourcing consolidation or discount optimization.
How do we address governance concerns before scaling adoption?
Start with agents that generate a full activity log, so approvals, audit trails, and agent permissions are visible from day one. Governance is easier to build into a small, contained pilot than to retrofit later.
To activate the Coupa AI already sitting in your instance, contact CrossCountry today.