BIG NEWS: Elements 3.0 is out of beta! šŸŽ‰

Mister @tpbradley has worked his late-night magic and the fix is in and shipped, check for updates :slight_smile:

Can this work in any way with free accounts of the AI models, or is a paid subscription a necessity?

Hi @trinube

You can also use the MCP server to connect an external app, and even run a local AI model, although it’s a bit more complicated to set up.

Is there a particular model or AI app you use?

Cheers

I use both Claude and ChatGPT for some fairly basic things, but currently only the free versions. I just want to know if I can use ā€œElementsAIā€ with the free versions, or if it would necessitate a paid subscription to one, or both, of them? I have no idea how to run a local AI model (and doubt I’d want to just for Elements).

Even after reading the full documentation, I’m struggling to understand the difference between the API and MCP options. I haven’t had time to watch Dan’s videos, but if they don’t cover the above, it would be good to see a video which spells it out for those of us less familiar with AI.

@dan I did not see this mentioned anywhere with regard to using the AI assistant. It says the PRO version of Elements is required to use the AI assistant, is this true?

@tpbradley @dan @ben Could you please point me to the AI MCP documentation? I really, really need to connect Perplexity here. This is what is needed:

  1. Publicly reachable HTTPS endpoint. Perplexity custom connectors require HTTPS and cannot reach http://localhost:9712/mcp . Needed is an endpoint in the form https:///mcp , either hosted by Realmac or exposed as an officially supported, secure tunnel from within Elements.

2. Documented authentication. A token or OAuth mechanism with clear instructions: where it is generated in Elements, the exact header format (e.g. Authorization: Bearer … ), validity period, and revocation. The current docs mention no authentication at all, and an open HTTPS endpoint is not acceptable.

3. Transport specification. An explicit statement whether the server speaks Streamable HTTP, SSE, or both, so the Perplexity transport selection can match.

4. Perplexity listed as a supported client. Inclusion in the compatibility list (currently only Cursor, Codex, Claude Desktop, ChatGPT Work Mode, LM Studio) plus a configuration example with URL, auth, and transport values.

5. Tool reference. A complete list of MCP tools with names, parameters, and return values, so calls can be planned rather than discovered by trial.

6. Permission and scope model. A statement of which actions a connected client is allowed to perform (read, write, publish, switch projects) and whether scopes can be restricted per token.

7. Runtime requirements for remote access. A clear statement whether Elements must be running with a project open, and whether multiple simultaneous client connections are permitted.

Slightly confused..is the AI assistant only available in the Pro version ?

I second that. Some step by step video / guide would be really helpful. Also on free versions, not really in need to subscribe AI before I see how to configure it. But my understanding is that videos are in the pipeline and will be released this week. I am sure it will be covered.

@dan Typo detected. :wink:

Where are you reading that information from, it must be outdated as the AI features are available on ALL Elements plans :tada:

The MCP docs are here, you should only need to give the Elements MCP URL to Perplexity for it to work.

Ensure the MCP server is switched on in the Elements preferences!

Unfortunately I’ve never used Perplexity, so can’t really help you with the setup. Perplexity do have their own guide on connecting to local MCP’s. Hopefully this will help you get things up and running.

@logrunner thanks for the heads-up, typo fixed!

Well that didn’t take long to eat into the Claude tokens BUT really impressed with what you can do. Thanks again to ALL the Elements team this is truly amazing what you continue to build.

I built this custom Claude usage app to keep me in the loop of where I am with tokens and usage.

You have to have a Pro version when you use your own API keys

@dan Thanks, Perplexity will eventually get the update to connect to the local URL. This is in the rollout phase. So I will wait.

@Fuellemann

I think you and me are the only ones using Perplexity here. I only use it because I got 1 year for free with my Internet + Mobile plan. :laughing:

I might keep it after the 1 year is up. I like the fact that Perplexity cites it’s sources very well, so I know if it’s feeding me good information or not from the sources.

Marc I am not into paying for AI yet as lower end user but just out of curiosity approx cost of this and what did you do to use this amount - Just the usage app. I understand if you dont want to share this but trying to gauge what I need. TIA

When we first started building in AI it was going to be available on the Pro version only, however, we decided to make it available on all plans as this feature is far too amazing to be limited to just Pro users.

Annoyingly, we forget to remove the gate in code (as we all run Pro plans here we didn’t notice). We’ll be releasing a build later today that will remove the Pro plan requirement!

To connect Elements to LMStudio via MCP:

  1. Make sure you’re using the current version. LMStudio has been moving things around.

  2. Open the right-hand panel, click on the hammer icon.

  3. You’ll find the item to add MCP. In the mcp.json area that will show up in the middle pane, type this code: `
    {
    ā€œmcpServersā€: {
    ā€œelementsā€: {
    ā€œurlā€: ā€œhttp://localhost:9712/mcpā€
    }
    }
    }

  4. Install and enable, and you should see a list of mcp/elements.

So, to show how LMStudio works (I just used Gemma-4-12b-qat as the model), here’s my prompt:

Create a single landing page for a new app that I’m launching, called byThom ZConnect. This iOS application connects Nikon Z System cameras to your mobile device, allowing for simple, convenient, and quick downloading of images, which then can be sent on via the app to Photos, Files, Messages, and Mail.

I picked this model because it’s small in size (but that also means it’s more limited in what it can do). It struggled to do what I asked, and kept getting confused about where it was. The difference between this and Claude was a bit startling. Gemma had no real context, while Claude immediately asked me three questions about my goal and then started working through the process, including figuring out how to do dark on light mode and dark mode. Took it quite some time to ā€œget thereā€, but Claude worked itself through the whole process eventually, while the small Gemma model just kept stepping on itself and getting into I’m not sure what the user wants loop.

Just goes to show that Realmac has clearly worked through the Claude integration, Claude is pretty sophisticated in what it can do, and very persistent in doing it. It would have taken me many more hours with Gemma to do the same thing that Claude did in minutes. Here’s Claude’s wrap-up of from that prompt:

Your ZConnect landing page is built in a new Elements project. Here’s what’s on the page, top to bottom: a dark hero with the byThom ZConnect wordmark, an amber ā€œGet the appā€ nav button, the headline ā€œYour Z System images. On your phone. Fast.ā€, and a primary App Store button; a four-step ā€œHow it worksā€ strip (Connect → Browse → Download → Send on); a six-item feature grid (Simple, Quick, To Photos, To Files, Messages and Mail, Made for Nikon Z); a compatibility section with a checklist; an amber closing CTA (ā€œStop emailing yourself photos. Get ZConnect.ā€); and a minimal footer with the Nikon trademark disclaimer.

For the dark, photography-forward look, I didn’t rely on the visitor’s system appearance — every section background is forced to near-black (#0a0a0a) in both light and dark modes, with white/gray text and a single Nikon-ish amber (#fabc00) reserved for the eyebrow label and primary buttons. The copy is written plain and direct, photographer-to-photographer, no marketing fluff.

A few things I deliberately left for you, since I didn’t want to invent facts or links:

  • App Store URL — paste your real link into the four download buttons (hero, compatibility, CTA, and the navbar ā€œGet the appā€). I didn’t want to fabricate one.
  • Images — the hero and compatibility sections still show template placeholder photos; swap in app screenshots or a camera-plus-iPhone shot.
  • Compatibility claims — I kept them generic (ā€œFull-frame Nikon Z bodies,ā€ ā€œiPhone and iPadā€); tighten those to your actual supported bodies and iOS version.
  • Icons — the step and feature icons are template defaults with blue fills; you may want amber replacements.

All of this is logged in the project’s backlog and journal in Elements, so a future session will pick up right where this left off. Want me to tackle any of those next — for example, if you give me the App Store link I can wire up all four buttons?

I should also point out that I’m using both via MCP, not API.