A native macOS app for analyzing FileMaker Server log files. Parse, visualize, and diagnose performance issues across 12 FileMaker Server log types.

Die Macher hinter FileMaker beschäftigen sich mit lokalen KI-Modellen auf dem Mac. Die App soll auch das Zusammenschalten von Rechnern erleichtern.

The team behind FileMaker is working on local AI models for the Mac. The app is also designed to make it easier to connect multiple computers.

Get and update currency rates in your FileMaker solution

When is it helpful to bundle your knowledge into a starter solution?

FileMakerとApache、Mac、Linuxに関するメモ。主にmacOS版およびLinux版のClaris FileMaker ServerによるWeb公開やClaris FileMaker Cloudに関連する情報を少しずつ記録しています。

The team behind FileMaker is working on local AI models for the Mac. The app is also designed to make it easier to connect multiple computers.

It feels quite appropriate that one of Apple’s more interesting AI deployments may not come from the company itself, but from its more business-focused subsidiary, Claris, still better known to many as FileMaker. What we know is limited: It intends to release an app that uses Apple Silicon Macs to create a private AI server. ## **What Claris is promising** The company, famous for its database software for the enterprise, has initiated a private preview of what sounds — at least on the surface — like something worth watching for enterprise users seeking to achieve AI insights from their data. The product is called Claris Harbor, and the company says it “turns the Apple silicon computers you own into a private AI server.” To achieve this, Harbor leans steeply into the powerful performance packed inside Apple Silicon: “Set up an endpoint in minutes on a single Mac.” It also makes use of the existing features of Macs that have already made them so popular among people deploying OpenClaw. “When you want to run larger models, connect multiple machines over Thunderbolt 5 to pool unified memory on hardware you control,” Claris said. It also says you can set up an endpoint on a single Mac — one of your Macs becomes an AI server for everyone else. ## **Your Macs. Your AI** Put this together and you have an Apple solution built around the part of the company with extensive enterprise and database experience that offers one of the emerging holy grails of business-class AI — powerful, scalable AI that can work on confidential and protected data on-premises, on-device, and in confidence. “Data and prompts stay strictly within your physical Mac hardware,” the company promised. While Claris doesn’t claim it, it’s not hard to imagine those banks of Macs already deployed in business being used as AI clusters overnight before returning to normal business use in the morning. That’s what the ability to connect multiple machines makes possible — though it’s not quite what Claris claims. Little more is known about the solution at this point. The company has opened a mailing list to keep potential customers informed, promising to provide early access to the new private AI server as it is built. ## **Why Claris, and why now?** I like the approach to this. Claris has been a consistently profitable entity for Apple. At one point in the history of the company, it was this part of its operation that kept the rest of the firm afloat. It has extensive experience in both data and enterprise and may therefore be the best part of the Apple machine from which to disseminate a wholly Apple approach to local AI. It comes at a good time. Frontier AI has hit a wall of colossal spending combined with relatively feeble returns, even as enterprise customers rapidly recognize the need to protect their data. Enterprise interest in local AI has surged. Businesses are already moving beyond the experimental cloud-only approach, navigating toward deployments guided by the need to control costs, protect data, and provide predictable governance. Deloitte’s State of AI in the Enterprise report, released last month, tells us that 74% of enterprises now prioritize data residency and in-region compute. If, as seems likely, Harbor is designed to integrate tightly with FileMaker databases, that opens a potential market of up to 1.3 million existing FileMaker users for this Apple approach to AI. Claris also has a community of over 30,000 developers, and it would be in keeping with FileMaker’s approach to relational database development for Harbor to be configurable in some way, making it a flexible solution for business users. We shall wait and see the extent to which that potential is realized as Harbor heads toward opening. ## **What we don’t know yet** Claris will have to answer some hard questions to build strong momentum once its new solution ships. Enterprise purchasers will want to know what AI models it runs and will be curious about the performance Thunderbolt 5 pooling provides when it comes to inferencing. They’ll want to learn more about cost and seek performance comparisons against other on-premises AI setups. It’s hard to make predictions about those things, but it does feel fair to point out that AI deployment will not be a zero-sum game. In most cases, I expect people will use on-device AI for most matters, on-premises AI for confidential or slightly more complex tasks, and cloud-based models only when serious AI horsepower is required. Much of this is highly speculative. Right now, Harbor is a mailing list and a promise. But for Mac-heavy enterprises handling sensitive data, it feels likely to be a mailing list worth joining, if only to find out more once it appears. **_Now please subscribe to my daily, human-curated Apple-related news headline feed at_** The Core**_, or follow me on_** BlueSky**_,_** LinkedIn**_, or_** Mastodon**_._**

FileMaker Server Checklist Critical: Do Today ☐ Limit Who Can Reach Your Server. This is the most effective step. Restrict access to known IP addresses with a firewall or VPN. If the public uses WebDirect or a website, restrict access only to the Admin Console and APIs under Administration > Restrict Access. Keep Remote Desktop and SFTP or FTPS backup downloads off the open internet. ☐ Turn Off Services You Do Not Use. Under Connectors in the Admin Console, turn off WebDirect, the Data API, OData, ODBC/JDBC, and XML publishing unless they are in use. In Configuration > FileMaker Clients, turn on Filter Databases so outsiders cannot list your files. ☐ Keep FileMaker Server on a Supported Version. Run a version of FileMaker Server that Claris still supports, and install security updates as soon as they’re released. If you run PHP or Custom Web Publishing, patch PHP first. Keep FileMaker Pro, the operating system, web server, plug-ins, and admin tools current too. ☐ Secure Your Admin Console. In Administration > Administrator, change the Admin Console password to a strong, unique one, and change it again whenever someone with access leaves or you suspect it was exposed. Replace your Admin API keys at the same time. If you can, limit Admin Console access to known IP addresses under Administration > Restrict Access. ☐ Check Your Logs for Signs of Access. In the Admin Console, go to Logs > Log Viewer and review the Access log, plus the Admin Console, Data API, and OData logs for any service you run. Bursts of failed logins or repeated attempts from unfamiliar IP addresses mean someone is testing your server, even if no one got in. Then look for anything you didn’t create: Admin API keys under Administration > Administrator and schedules under Configuration > Schedules. Something you didn’t create means someone may have gotten in. ☐ Confirm Your Backups. Keep a current, encrypted backup that an attacker on the server cannot change or delete, and test a restore. ☐ Install a Trusted SSL Certificate. The default FileMaker certificate is for testing only, and without a custom certificate, FileMaker Pro and Go connections to the server aren’t encrypted. In the Admin Console, go to Configuration > SSL Certificate and import a certificate from a trusted authority, or set up Let’s Encrypt for FileMaker Server for a free certificate that renews automatically. Next: Plan These ☐ Set Up Single Sign-On (SSO) with MFA. Sign in through an identity provider such as Microsoft Entra ID, Okta, or Google to require multi-factor authentication. When someone leaves, disabling one account removes their access everywhere. ☐ Review Older Web Integrations. Older PHP and XML integrations often run on one shared Full Access account. Plan a move to a maintained, least-privilege API connection. FileMaker Application Configuration Checklist Critical: Do Today ☐ Turn Off Guest Access and Auto-Login. In File > Manage > Security, uncheck the Guest account. Then, in File > File Options > Open, deselect Log in using. Files are often set to sign in automatically with a default Admin account, the first door attackers try. Need Guest access? Keep public data in a separate file. ☐ Lock Down Every Full Access Account. In File > Manage > Security, remove any Admin account with a blank or default password. Give every Full Access account a strong, unique passphrase, including accounts used for Data API or OData keys, and limit Full Access to a few named people. ☐ Remove Extended Privileges Accounts Do Not Use. In File > Manage > Security > Advanced Security Settings > Extended Privileges, turn off fmrest, fmodata, fmxml, fmxdbc, fmphp, and fmwebdirect wherever they are not needed. Keep fmapp on for FileMaker Pro and Go users. ☐ Set a Minimum Password Length. In File > Manage > Security > Advanced Security Settings > Privilege Sets, edit each custom privilege set and set a minimum password length. Built-in sets like [Full Access] can’t be edited, so move regular users onto custom sets and keep [Full Access] for a few named developers. ☐ Remove Old and Shared Accounts. In File > Manage > Security, delete accounts for former employees, contractors, and vendors, and disable any account no one can identify. Replace shared logins with individual accounts so every user has their own. Each unused account is one more password an attacker can guess, and accounts for people who’ve left are rarely watched. Next: Plan These ☐ Give Every Account Only the Access It Needs. Build privilege sets on a least-privilege basis, with separate accounts for FileMaker Pro, WebDirect, and each integration. ☐ Enable Encryption at Rest. Encrypted files can’t be opened outside FileMaker, even if someone copies them from the server or a backup. ☐ Keep Secrets Out of Your Files and Logs. Do not store passwords in FileMaker fields or build login systems that look up users in a table. Keep API keys and personal data out of logs and exports. Short on Time? Start Here If you only have time for a few things this week, do these five. They close the doors automated attacks try first: ☐ Turn Off Guest Access and Auto-Login in every file. ☐ Lock Down Every Full Access Account with a strong, unique passphrase. ☐ Turn Off Services You Do Not Use in the Admin Console. ☐ Secure Your Admin Console with a new password and new Admin API keys. ☐ Keep FileMaker Server on a supported version, patching PHP first if you run it. Think Your Server Is Under Attack? In the Admin Console, enable Block New Users under Configuration > General Settings, disconnect current users, and close the server to external traffic. Then change your passwords, remove any keys, schedules, or accounts you didn’t create, review your logs, and contact your host or our team. How DB Services Can Help FileMaker provides strong security controls. Most attacks succeed only when those controls go unused, and the same habits protect every system with a login. Want a second set of eyes? DB Services can review your servers and applications against these checklists and make the changes for you. Schedule a FileMaker Health Check or contact us for further FileMaker safety assistance.

Apple arbeitet an einer neuen Mac-Anwendung, die den Betrieb lokaler KI-Modelle deutlich vereinfachen soll. Die Software trägt den Namen „Harbor“ und wird von Apples Tochterunternehmen Claris entwickelt, das vor allem für die Datenbanksoftware FileMaker bekannt ist. Harbor soll Macs mit Apple-Prozessoren in private KI-Server verwandeln. Dabei lassen sich die Modelle nicht nur auf einem einzelnen […] ★ ifun.de bei Google bevorzugen ★ Der Beitrag Mehrere Macs als KI-Rechner: Apple arbeitet privaten KI-Servern wurde zuerst auf ifun.de veröffentlicht.