Claris shipped FileMaker 2026.
On June 9, Claris released FileMaker 2026. New paid add-ons for disaster recovery. AI field annotations. Google Gemini support. A promise that agentic development previews are coming "later this summer."
If you run a business on FileMaker, this release surfaces a question that's been building: do you upgrade again, or do you migrate?
I built migrate-filemaker, an open-source FileMaker migration methodology with a DDR parser, a four-phase plan, and a Claude Code skill. Here's what the 2026 release means for the stay-or-migrate decision.
What's in the release
Disaster recovery add-ons. Remote Backup encrypts and ships your data to Apple's cloud every 20 minutes. Standby Server keeps a secondary server synced and ready to take over in minutes. Both are paid add-ons. These are features that should have shipped years ago — every other database platform has had backup and replication for a decade. FileMaker is catching up and charging extra for it.
AI that understands your schema. New field and table annotations let you tell AI models what your fields mean, not just their names. Better natural language queries, better semantic search. Google Gemini joins the LLM lineup alongside OpenAI, Anthropic, and Cohere. This is useful. Claris is treating AI as a platform feature, not a checkbox.
Agentic coding — previews this summer. Claris is building toward making FileMaker a "first-class development target for AI tools." They haven't said exactly what that means, but the intent is obvious: keep you building inside FileMaker, just faster.
You don't have to wait for Claris
While Claris works on agentic previews, the AI-inside-FileMaker path already ships. GenieWave, a tool out of Quebec, runs a local MCP server that connects Claude Desktop directly to your FileMaker environment. Claude can inspect your schema, read your scripts, generate new logic, edit existing code, and build layouts from a screenshot. Every change goes through a human review gate before it touches your solution.
It works now. Not "later this summer." Now.
This changes the stay-or-go math. If AI-assisted FileMaker development solves your actual problems — faster debugging, less boilerplate, quicker layout work — then upgrading to 2026 and adding a tool like Genie might be all you need. You get AI leverage without leaving the platform.
But if your problems are things AI can't fix inside FileMaker — web portals, API integrations, per-user licensing costs, developers aging out of the ecosystem — then AI-assisted FM development just makes you faster at building on a platform that still has the same walls.
The two paths
Every FileMaker shop hits this fork. The 2026 release sharpens it.
Path one: upgrade. You move to FileMaker 2026, maybe add Remote Backup if downtime scares you, maybe add Standby Server if you need uptime. You annotate your fields so AI understands your schema. You connect Claude Desktop through GenieWave or wait for Claris's agentic previews. Your licensing bill goes up. Your stack stays FileMaker.
Path two: migrate. You pull your data, parse your schema, and rebuild on a modern stack. Postgres. A web framework. Cloud hosting that doesn't charge extra for backups. You own your code, your data, your infrastructure. No per-user licensing. No upgrade treadmill.
Three years ago, path two meant a quarter of downtime and six figures in consulting. Today AI handles the boilerplate — schema extraction, data mapping, scaffold generation — but you still have to validate, test, and handle the edge cases. The difference isn't that migration is free. It's that the mechanical work that used to eat weeks now takes hours. The thinking work still takes thinking.
Why the math changed
AI didn't just speed up coding. It collapsed the cost of being wrong.
Before, if you picked the wrong migration approach, you lived with it for months. Now you can fork reality. Build one version. Learn what doesn't work. Spin up a parallel rebuild over a weekend. Keep the version that works until the new one proves itself.
The education loop I wrote about in "The Loop" used to be project-to-project. You'd finish something, realize what you should have done differently, and carry that lesson into the next one. Now the loop is week-to-week. Sometimes day-to-day.
This changes the migration decision. The risk of migrating isn't what it was in 2021. The cost of staying — more licensing, more add-ons, more years on a platform that charges for backups — compounds while you wait.
What staying costs
FileMaker licensing isn't cheap and it isn't getting cheaper. The 2026 release adds paid services on top. Remote Backup. Standby Server. Every new feature that fills a gap you'd get for free on a modern stack becomes another line item.
Then there's the upgrade cycle. Claris ships annually now. Every release means testing your solutions, fixing what broke, deploying to users. Some shops skip versions. That works until it doesn't. Eventually something breaks and you're upgrading through three versions at once.
The hidden cost is the talent pool. FileMaker developers are aging out. New developers don't learn FileMaker. They learn React, Python, Postgres. Every year you stay, the pool of people who can maintain your system shrinks. Claris knows this. It's why they're pushing AI-assisted development so hard. They need AI to fill the developer gap.
Who else is in this
A few people are building businesses around AI and FileMaker. They split into two camps: tools that help you stay, and tools that help you leave.
Stay tools. GenieWave connects Claude Desktop to FileMaker via MCP. Schema inspection, script generation, layout building from screenshots. The most polished AI-inside-FM product I've seen. agentic-fm (64 GitHub stars) is the open-source alternative — XML export parsing, clipboard delivery, active Discord.
Leave tools. Rebased (Belgium) does fixed-price FileMaker to Rails migrations. Solo founder, ex-FM consultant. EUR 5,000-25,000. Clean marketing, no case studies yet. GrowwStacks claims 1-5 day migrations using Claude Code. Light on evidence. FmPro Migrator is the old guard adding AI to an existing conversion tool.
I built migrate-filemaker as an open-source methodology: DDR parser, four-phase migration planner, Claude Code skill. The idea isn't to sell consulting. It's to make the path visible enough that you can walk it yourself.
When to stay
Migration isn't always the right call.
- Your FileMaker system is small, stable, and does exactly what you need. No backlog. No scaling problems.
- You have one or two power users, not a team. The per-user math still works.
- You have zero appetite for technical change right now. The business can't absorb it.
- The AI-inside-FM path actually solves your problems. GenieWave plus Claude Desktop plus FileMaker 2026 covers what you need.
If those are true, upgrade and keep going. FileMaker runs businesses. It's fine.
When to migrate
Here's when you should be looking at the exit:
- You're paying more in licensing than you'd pay for cloud hosting.
- You're adding features FileMaker wasn't designed for. Web portals. External APIs. Mobile beyond basic data entry.
- You can't find developers, or the ones you have are retiring.
- You've been putting off a rewrite because the cost was terrifying. The cost changed. AI handles the grunt work now. You still have to validate and test, but the mechanical part isn't the bottleneck anymore.
- Claris's roadmap doesn't match yours. If you need something FileMaker won't have for two years, waiting is expensive.
AI cuts both ways
Claris positioning FileMaker as an AI development target is smart. It keeps customers in the ecosystem by promising AI will make FM development faster. GenieWave proves the approach works. agentic-fm is building the open-source version. Claris's own agentic previews are coming.
But if AI can build FileMaker solutions faster, it can also build your replacement faster. The same Claude session that annotates your FM fields can write your Postgres schema. The agentic future Claris is selling is real. You just don't have to stay on their platform to get it.
The migration path already works. People have linked to migrate-filemaker from Reddit while migrating their own systems. The methodology is public, the DDR parser is open source, and the four-phase plan handles everything from schema extraction to go-live.
FileMaker 2026 is a good release. The DR add-ons fill real gaps. The AI annotations are useful. The agentic direction is the right one.
But none of it changes the question: do you want to keep paying for features that should have been table stakes a decade ago, or do you want to own your stack?
The decision used to be measured in quarters and budgets. AI shrinks the mechanical work, but you still have to show up and do the thinking. The difference is leverage, not magic.