DT 1.5 reads your public site, keeps the evidence, explains what it found, and drafts supported fixes. Eligible Shopify changes move only after your approval and are checked against the public result; unsupported work remains a review-ready patch.
Most AI products blur this line on purpose. We refuse to. Green means you can use it today. Amber means it is in development and does not exist for you yet, no matter how good it sounds.
A real crawl of any URL you give it, on demand: page metadata, headings, robots.txt and AI-crawler rules, schema.org structured data, llms.txt, and sitemap coverage. Findings ranked by severity, written in plain language.
Where the audit finds missing structured data, DT 1.5 drafts the exact patch from your real page content. The suggestions are deterministic, built from what is actually on your site, not language-model guesswork.
Every drafted fix lands in a queue where you approve or reject it. The workspace, task states, and approval flow are real in the dashboard today.
Sign in without payment during the current open beta and connect a Shopify store, Google Search Console property, or optional AI key. Eligible approved Shopify fixes move through the executor and are checked against the public storefront; other supported fixes remain local or paste-ready patches.
Checking how your business appears inside ChatGPT, Claude, and Perplexity answers. Live in the CLI (mgo ai-scan, bring your own key) and in the dashboard once you connect an AI key.
An interactive agent (run bare mgo and type plain language), drafted titles and descriptions, llms.txt (informational only, not a ranking or citation factor), ad copy, review replies, and a brand voice guide, plus an advisory weekly plan from mgo director. Drafts only: everything is filed for your approval and nothing is ever auto-published.
Budget and margin checks, campaign structure, creative drafting, exact ad-account discovery, pixel or tracking-source selection, identity selection, location validation, and creative upload. Nothing spends during setup.
For supported connected Google Ads, Meta, and TikTok accounts, DT 1.5 can create an approved campaign paused and read the exact provider state back. Activation and later budget or status changes require separate confirmation. Google Business Profile remains unavailable while MGO awaits Google's separate API approval.
Consistency monitoring across your listings, publishing, and the dashboard versions of the drafting lanes. In development. DT 1.5 does not manage or publish your brand today; on your own key the CLI already drafts on-voice responses, see the row above.
DT 1.5 never acts on its own. It is built around one rule: the system proposes, you decide.
Give it your URL. It crawls your site live, the way Google's and OpenAI's crawlers do.
Findings come back ranked and translated. Not "missing JSON-LD", but "AI assistants cannot read your product pages, here is why it matters".
Drafted fixes wait in your queue. You see exactly what would change before it does. Reject anything, no questions.
Approved fixes apply through the executor to your connected Shopify store, or come as a patch ready to paste anywhere else. Either way, it happens after your yes, never before.
Using Shopify or another ecommerce platform? Read the exact SEO automation and verification workflow.
AI models earn version numbers by getting measurably better between releases. We hold DT to the same standard: DT is the agent and system, 1.5 is the version you are using, and the capability board above is its honest spec sheet. When the ads, branding, and content lanes clear our bar, the version number moves and the board moves with it.
To be precise about what DT 1.5 is not: it is not a foundation model MGO trained. It is the agent that does the work, and where that work needs a language model it runs on the leading models you bring with your own key, the way Claude Code and Codex do.
What never changes between versions: the approval gate, the plain language, and the refusal to claim capabilities that do not exist yet.
No. DT 1.5 is MGO Data's agent and system. Its audits and schema drafts are deterministic. The dashboard includes an assistant with no user API key required, and you can optionally connect your own model. In the CLI, your keys stay in your environment and go provider-direct. A key connected in the dashboard is encrypted server-side, is never shown to the browser again, and is used only for the scans and drafts you request.
Real. The dashboard runs a live crawl of whatever URL you enter and the findings come from your actual pages. When the live service is unreachable it falls back to a recorded sample of our own store, and it is labeled as such on screen.
No, and that is the point. Every finding is written for a founder, not a developer. Where a fix is technical, DT 1.5 drafts it for you so you can hand it to whoever touches your site, or paste it yourself.
It can plan campaigns and verify the exact connected account, tracking source, identity, location, and creative before approval. For supported Google Ads, Meta, and TikTok connections, an approved campaign can be created paused and read back from the provider. Activation is always a separate confirmed action. Google Business Profile remains unavailable while MGO awaits Google's separate API approval. If you want people to own the whole process, that is what the human team is for.
They are two different things. The team is people doing the work for you, by application. DT 1.5 is software you drive yourself. The team uses DT 1.5 as a tool; DT 1.5 does not pretend to be the team.
A live crawl of the URL you give it: page titles and meta descriptions, heading structure, robots.txt rules for AI crawlers like GPTBot, ClaudeBot, and PerplexityBot, schema.org structured data, Open Graph tags, sitemap coverage, and llms.txt. Findings come back ranked by severity in plain language, and where structured data is missing DT 1.5 drafts the exact patch from your real page content. One honest note: llms.txt is reported as informational only, because per Google guidance it is not a ranking or citation factor, so a missing one is never flagged as a problem. The exact commands are on the get started page.
ChatGPT (OpenAI), Claude (Anthropic), and Perplexity, on your own API key. The Perplexity lane is web-grounded and cites sources; the OpenAI and Anthropic lanes answer from model knowledge in this build. Honest caveat: API answers approximate the consumer apps rather than exactly equal them, so treat the scans as directional.
The audit takes about a minute and does not ask for your card.
Try DT 1.5 free