dx0 now uses a stronger thinking model when it researches your systems, whether that’s your Salesforce org or software connected through GitHub, GitLab, or Bitbucket.

It traces automations, Apex, code, dependencies, and relationships with more care than before.

You can also choose how much thinking to spend on a conversation from the dropdown next to the composer:

• Fast — quicker answers for simple questions
• More thinking — a bit slower and more thorough (this is the default)
• Extra thinking — takes longer; best when the system is messy or the answer really matters

Pick it before you send the first message. That choice stays for the rest of the conversation, so later follow-ups keep the same depth.

Use More thinking or Extra thinking when you’re mapping an inherited Salesforce org, understanding an unfamiliar codebase, tracing dependencies across systems, checking what a change could affect, or asking something you really don’t want dx0 to miss.

Your Salesforce metadata only tells part of the story. Now dx0 understands your business context too.

With Context, you can bring the knowledge that lives in documents, spreadsheets, screenshots directly into every conversation.

You can now:

• Add Context Notes for acronyms, business rules, historical decisions, naming conventions, or reusable user story templates.
• Upload files such as user stories, requirements, CSVs, screenshots, and other supporting documents.
• Combine Salesforce metadata + business context + uploaded files in a single prompt for dramatically better answers.

Need to know whether a new set of requirements impacts a field? dx0 can analyze the requirements document against your Salesforce metadata. Want user stories generated in your company's standard format? Just add your template as Context and dx0 will follow it automatically.

And yes, attachments are finally here. Simply drag and drop your files into the conversation.

Watch the video below to see Context in action.

You can now choose whether dx0 can see CRM data

Metadata got us far. Dependency maps, impact analysis, org intelligence — all built on metadata alone. But there's a whole category of Salesforce problems that metadata can't solve, and we've been hearing about them.

Why isn't this lead assignment rule firing? Why is the duplicate rule letting records through? What's actually happening in production?

Now the agent can go find out.

How to get it

Data access is off by default — reach out to us and we'll enable it on our end. From there, you switch it on in your account settings and you're live. The agent decides when a data query is actually needed; you don't have to prompt it.

On privacy

Before anything reaches the LLM, it goes through the dx0 trust layer. PII gets stripped on our servers — emails, dates of birth, ID numbers, gone before the model ever sees them. The agent knows this and won't even attempt to surface that data if you ask. It's not a setting you configure. It's just how it works.

If you want it enabled, get in touch.