AI on the Automize platform

AI reads the documents, drafts the flows and diagnoses the failures. Automation does the work. A person approves what matters. Here is what the AI actually does – and the controls it runs under.

What the AI actually does

Not a chatbot bolted on the side – AI woven into building, running and supervising automation.

Document Intelligence

Bank statements, invoices, receipts, CVs and timesheets read into structured data – totals re-added, balances walked, exceptions flagged before anything posts. The engine behind the document tools and every Digital Worker’s reading.

Smart Action Search

Describe what you want to automate in plain English and the canvas finds the right actions among 600+ – so building starts from intent, not from a catalogue.

AI Process Chat

Ask questions about a process in natural language, get summaries, and draft changes conversationally – every suggestion yours to accept or reject.

AI Error Diagnosis

When a run fails, AI reads the error, names the root cause and proposes the fix – reducing downtime and manual debugging.

AI Data Transformation

Clean, restructure and convert data between systems automatically – format conversions, deduplication and quality checks written for you.

AI Inside Documents

Generated documents and wiki templates can carry AI instructions that resolve per document – a covering summary, a tailored paragraph – written at generation time, from the submission’s own data.

Agentic outcomes, without the leap of faith

The automation market now calls this category agentic automation: AI that reads, decides and acts across your systems. Automize delivers that outcome differently – and the difference is the point. AI does the reading and the drafting. The work itself runs as a deterministic, auditable process: a cloud flow where your system has an API, a software bot driving the screens where it does not. And wherever judgement is required, the process stops and asks a named person.

That is why an Automize automation can touch a ledger, a payroll or a trust account: when it is not certain, it stops rather than guesses, and every decision is logged with the document and the reasoning attached. You get what the agentic platforms promise – whole jobs done – with an audit trail your auditor can actually read.

It also reaches where AI-only tools cannot: systems with no API at all – desktop Sage Pastel, SYSPRO, supplier portals – because a software bot can drive their screens the way a person would.

Under your control

Enterprise AI is a governance question before it is a capability question. These are the controls.

Your Provider & Model

Powered by OpenAI and Anthropic – you choose the provider and model per workspace, and different jobs can run on different models.

Your Own Key

Bring your own OpenAI or Anthropic API key and your AI spend runs on your account, under your limits and your data agreements.

Metered Per Purpose

Every AI call is metered by what it was for – document reading, diagnosis, chat – so usage is visible, attributable and budgetable.

Only What You Submit

The AI sees the prompts and documents you submit for a task – never background data – and nothing trains models. Bot-side documents stay on your machines.

Read the Trust Center Platform architecture

Where you meet it

In a Digital Worker

AI reads every supplier invoice and bank statement the role handles; the role posts the entries and brings you the exceptions.

On the canvas

Search actions in plain English, chat with a process, transform data, and let AI diagnose a failed run while you build your own automations.

In the document tools

The same document intelligence, self-serve: drop in a statement or an invoice and get checked, structured data back – free to try.

Questions about the AI

Automize is powered by OpenAI and Anthropic. You choose the provider and the model per workspace, and different jobs can run on different models – document reading on one, error diagnosis on another. You can also bring your own API key, so your AI spend sits on your own account with your own limits.
No. The AI sees only the prompts and documents you submit for a specific task – never background data – and nothing is used to train models. Where a bot runs on your own machine, documents, credentials and screen contents stay on that machine. See the Trust Center for the full data posture.
The market calls this category agentic automation, and Automize does the same work those platforms describe: AI reads and drafts, automation executes, people stay in the loop. We deliberately do not call our roles “agents”, because they are stronger than that word implies: a Digital Worker is a deterministic, auditable process with AI where reading is needed and a human approval gate where judgement is – it stops rather than guesses. You get the outcome agentic platforms promise, with an audit trail a regulator can read.
It helps you build one. Describe the step in plain English and smart action search finds the right actions; conversational process chat answers questions about a process and drafts changes; AI data transformation writes the cleanup steps between systems. Every suggestion is yours to accept or reject – the AI never publishes a process on its own.
It stops and asks a person. Any flow can suspend, put the document and the proposal side by side in front of a named approver, and resume exactly where it stopped once they answer. Filings, payment runs, payroll and trust-account transfers always stop for a person – those gates are permanent.
AI features are included in the platform plans, with usage metered per purpose so you can see what each capability spends. On your own API key, you pay your provider directly at their rates and Automize meters the calls for you.

See it read one of your documents

The fastest way to judge the AI is to hand it something real – a statement, an invoice – and check the numbers yourself.

Try the document tools free Start a 7-Day Proof-of-Value