Benny
The in-app assistant, scoped to what you are allowed to see, answering from your knowledge base and your live records.
Everything you can do with ai & assistant
An assistant that has read your workspace, not the internet.
Inside ai & assistant
Every piece that makes this work, and what each one is actually for.
Retrieval
Your articles, notes, and recordings indexed so an answer can cite the paragraph it came from.
Setup generation
Describe the form, the workflow, the calculated field, or the import mapping, and Benny drafts it. You review and refine before anything goes live.
Action plans
Turn a goal into a sequenced plan and a vague ask into a structured query with a working view behind it. Focus plans bring the next block of work forward.
Meeting intelligence
Transcription, summaries, and action items extracted from recordings, filed against the account or project they concern.
Agents
Scheduled and triggered passes that summarize, score, and tidy without waiting for someone to ask.
MCP access
An authenticated bridge so external tools and assistants can read and act on the workspace under real permissions.
AI credits
Usage visible and budgeted per organization, so the bill is never a surprise.
Finished work becomes documentation
A meeting recording, a browser capture of a process, or a note all carry their provenance forward, so you can always see where a task or an article came from. Action items pulled out of a note keep the link back to the sentence that produced them. The same article then feeds the assistant, the client portal, and the public help center, from one source rather than three copies.
What this removesDocumentation as a separate chore. The assistant answering from a generic model instead of from how your shop actually operates.
See how everything connectsThe questions this actually answers.
The long version, because this is where the category has real objections.
What has your assistant actually read?
Your workspace, not the open internet. Benny answers from your records, articles, notes, and recordings, scoped to what the person asking is allowed to see, and cites the paragraph an answer came from. A generic model gives you a plausible answer. An assistant that has read your workspace gives you your answer, with the receipt attached.
Cited, not asserted
Every answer keeps a note of where it came from.
Permission-scoped
Benny sees what you see. No more.
Your data stays yours
Retrieval runs over the workspace, no training on your content.
Can the AI do the setup work?
Yes. Describe the intake form, the approval workflow, the calculated field, or the import mapping in plain language and Benny drafts it, ready to refine before it goes live. The configuration work that usually eats the first month of a rollout becomes an afternoon of reviewing drafts.
How does a goal become a plan?
Hand Benny a goal and the context already in the workspace, and it drafts an action plan with sequence and owners to review. Complex asks translate into structured queries and working views, so the question about last quarter becomes a screen instead of a data request.
Who stays in control?
People do, on purpose. Everything the AI produces arrives as a draft to review, permissions gate what it can see, and credits are budgeted per organization so usage is visible and the bill is never a surprise. MCP access extends the same rules to external tools: your assistant of choice reads and acts under real permissions.
Questions people actually ask
Does the AI train on our data?
No. Retrieval runs over your workspace so answers cite the paragraph they came from, scoped to what the person asking is permitted to see.
What can Benny generate?
Drafts of forms, workflows, calculated fields, and import mappings, plus summaries, replies, action plans, and structured queries. Everything arrives for review before it goes live.
Can Benny answer questions about a specific client?
Yes. Ask about an account and the answer draws on its records, notes, meetings, and articles, with citations back to the source.
What happens to meeting recordings?
Transcription, a summary, and extracted action items, filed against the right account or project, with the follow-up work one move from becoming tasks.
What is MCP access?
An authenticated bridge that lets external AI tools, like Claude, read and act on your workspace under your real permissions. It ships on Pro and Business.
How is AI usage billed?
Through AI credits included with each plan and budgeted per organization. Usage is visible, so the bill is never a surprise.
It does not stop at one capability.
Scrambl is one workspace, so the records here are the same records the rest of the platform reads. These pair with it most often.
.svg)
Ready to crack the chaos?
One workspace for projects, clients, finance, and everything in between. Try for free.