I've been building and evaluating EA tools since 2021. In that time, I've watched the market shift dramatically — and I've noticed that most of the "change" is marketing while the actual problems remain stubbornly unsolved.

This isn't a vendor comparison. It's an assessment of where the industry actually is, based on what I've seen work and what I've seen fail.

What's Actually Changed

Cloud-native is now the default

In 2021, most EA tools were either on-prem or hybrid. Now almost every new evaluation starts with cloud-first tools. The legacy on-prem tools still exist, but they're no longer the default choice for teams starting fresh.

AI features are everywhere

Every major vendor has added AI features in the past 18 months. Document import, automated classification, natural language querying — the AI wave hit EA tools like it hit everything else. The quality varies enormously, but the features exist.

Integration expectations have risen

Teams expect EA tools to connect to Jira, ServiceNow, SAP, and the other systems they already use. Tools that can't integrate with at least the major platforms get eliminated early. Integration has become table stakes rather than a differentiator.

What Hasn't Changed

Time to value is still too long

This is the problem that frustrates me most. Despite all the innovation, most EA tool implementations still take 6-12 months to show meaningful value. The tools are easier to set up, but organizations still struggle to populate them with accurate data and get users to adopt them.

Diagrams still go stale

No tool has solved the maintenance problem. Manual update processes still dominate. The AI features added in the past two years help with initial population, but the day-to-day challenge of keeping diagrams current remains largely unsolved.

Executive buy-in is still rare

EA is still often seen as an IT exercise rather than a business capability. Tools have improved, but the organizational challenge of making EA matter to decision-makers hasn't changed much. Teams that succeed at EA implementation typically have an executive champion who understands the value — that's still the exception, not the rule.

The Three Categories of EA Tools

After five years, I've settled on three meaningful categories for EA tools. Most products in the market fit one of these:

1. Catalog-first tools

These tools optimize for tracking applications and their attributes. They're essentially structured databases with visualization layers on top. Good for: organizations that need ITAM-style tracking. Bad for: teams that want help making architectural decisions.

Examples: LeanIX, Clarity (formerly EgPMC), ServiceNow EA

2. Modeling-first tools

These tools optimize for creating diagrams and modeling relationships. They feel more like design tools than databases. Good for: teams that need to produce architecture artifacts. Bad for: organizations that need to track changes over time or maintain historical accuracy.

Examples: Sparx Enterprise Architect, Visual Paradigm, BiZZdesign (partially)

3. Collaboration-first tools

These tools prioritize team usage over modeling depth. They're more approachable for non-architects and often have better UX. Good for: organizations building EA culture. Bad for: teams that need deep modeling capabilities or regulatory compliance features.

Examples: DesignFoundry, Cantara, HOPEX (attempted this)

Most teams pick one category and wish they had features from another. That's the underlying tension in the market — no tool does everything well.

Where I Think We're Going

Three predictions, with appropriate uncertainty:

Maintenance automation becomes the battleground. The vendors who solve the staleness problem — who make diagrams that stay current without constant manual effort — will win the next generation of the market. It's a hard problem, but it's the right one to solve.

EA becomes more embedded in operations. Instead of being a separate practice, EA information shows up in the tools teams already use — project management, service management, compliance tracking. The standalone EA tool doesn't disappear, but its value migrates into operational systems that consume its data.

Regulated industries force a category split. Right now, most EA tools pretend they can serve both general enterprise and regulated environments. Over time, the compliance burden in pharma, medical devices, and healthcare will force vendors to specialize. Tools built for regulated industries will look and feel different from general-purpose tools, and buyers will understand why.

What This Means for Teams Evaluating Tools

If you're evaluating EA tools today:

Ask the hard questions early. Not "do you have compliance features?" but "has your tool been used in a validated FDA environment, and can you show me the documentation?"

Test the maintenance burden, not just the setup. Most demos show you how to build something new. Ask instead: "How do we keep this current in six months? What happens when the person who built it moves to another role?"

Plan for the 90-day failure point. EA tool implementations commonly stall at 90 days — initial enthusiasm fades, daily maintenance becomes a burden, users stop opening the tool. Plan for this. Assign resources specifically for keeping the initiative alive.

Pick a tool you'll actually use. The best EA tool is the one your team opens every week. If the best tool on paper requires so much setup that nobody uses it, it's not the right tool.

The market will keep evolving. The problems that seemed hard in 2021 are still hard, but some are starting to yield to better technology. The teams that get value from EA tools aren't the ones who bought the most comprehensive solution — they're the ones who picked something realistic for their situation and actually used it.

What approaches have worked for you? I'd genuinely like to know — especially from teams that have been through a full EA tool evaluation recently. What would you do differently, and what would you recommend to teams just starting this process?