Skip to content
evolus

AI automations

Processes that run themselves.

An email, a document or a WhatsApp message arrives, or a set time comes round: an Evolus automation gets going and carries the work through to the end. It reads, understands, decides, updates your software and lets the right people know.

Tell Evo about your processSee the plans

From the Growth plan, with 2 automations

Example: an email with a quote request starts the automation. It reads the email, picks out products and quantities while looking up the customer in the management system, works out the price, replies with the quote and records the offer. Had the discount been higher, it would have handed the case to a salesperson.

What an automation does

An automation doesn't do one task, it carries a whole process forward, across email, documents, your software and people.

It doesn't stop at the first step: it reads, decides, repeats, delivers and brings in whoever is needed. And you can see everything it did, step by step.

A whole process

Not just invoices: any work that follows rules.

Seven very different processes that an Evolus automation carries through on its own, from start to finish. Each one is built with the blocks you find in the editor.

  1. 01Sales

    From request to quote

    It starts when an email arrives with a request for a quote.

    1. Reads the text and the attachments, spreadsheets included.
    2. Picks out products, quantities and customer.
    3. Takes the prices from your management system and works out the total with a script, with no guesswork.
    4. If the discount goes over the threshold, it passes the case to a salesperson.

    In the end: The customer gets the quote and the offer is recorded in the management system.

    Blocks used

    • Incoming email
    • Read the document
    • Process with AI
    • Call an endpoint (API)
    • Run a script
    • Choice (if… else)
    • Send an email

  2. 02Customer service

    A complaint, from arrival to reply

    It starts when a customer writes in to complain.

    1. Works out what it is about and how urgent it is.
    2. Sorts the case: delivery, faulty product or invoice.
    3. An agent that knows your procedures drafts the reply.
    4. Opens the case in the management system and alerts the person in charge of that kind of case.

    In the end: The customer gets a careful reply and the case is already in the right hands.

    Blocks used

    • Incoming email
    • Process with AI
    • Routing
    • Ask an agent
    • Call an endpoint (API)
    • Send an email

  3. 03People

    Screening CVs

    It starts when CVs land in the applications mailbox.

    1. Works through each CV on its own, one per document.
    2. Picks out experience, skills and contact details.
    3. Compares them with the requirements of the role.
    4. Writes kindly to those who aren't a fit and passes suitable profiles to the digital employee, who schedules the interviews.

    In the end: The right candidates have an interview in the calendar, the others a reply.

    Blocks used

    • Incoming email
    • Repeat for each item
    • Call a data extractor
    • Process with AI
    • Choice (if… else)
    • Send an email
    • Task for the digital employee

  4. 04Legal

    A contract, analysed

    It starts when a contract comes in for review.

    1. Reads the whole document, even if it is a scan.
    2. Finds the parties, term, automatic renewals and notice periods.
    3. Flags penalties and risky clauses, and explains why.
    4. If the risk is high, sends it straight to whoever has to decide.

    In the end: The decision-maker gets a clear summary, with the deadlines and the clauses to look at.

    Blocks used

    • Incoming email
    • Read the document
    • Process with AI
    • Choice (if… else)
    • Send an email

  5. 05Management

    The Monday report

    It starts when it's Monday morning at 8.

    1. Asks the management system, the CRM and the ticketing system for their figures, all at once.
    2. Merges the results into a single table.
    3. Works out the changes against the previous week.
    4. Writes the commentary: what's going well, what isn't, what to watch.

    In the end: The report is in the team's inbox before the meeting starts.

    Blocks used

    • Schedule
    • Call an endpoint (API)
    • Merge results
    • Run a script
    • Process with AI
    • Send an email

  6. 06Purchasing

    Orders followed through to delivery

    It starts when it's morning, every day.

    1. Takes the open supplier orders from the management system.
    2. Checks the promised delivery date for each one.
    3. Writes to late suppliers to ask for a new date.
    4. Marks the chased orders in the management system.

    In the end: Purchasing knows what's arriving, what's late and who has been chased.

    Blocks used

    • Schedule
    • Call an endpoint (API)
    • Repeat for each item
    • Choice (if… else)
    • Send an email

  7. 07Personal data

    The AI works, the real data stays yours

    It starts when your management system sends a file with names, tax codes and IBANs.

    1. Replaces names, codes and contact details with fictitious values.
    2. The AI model summarises and classifies the file seeing only those.
    3. At the end it puts the real values back in place.
    4. The mapping between real and fictitious values is never stored: it only lives for the duration of the run.

    In the end: The summary goes back to the management system with the real data. The AI and the external services in the flow only saw the fictitious values.

    Blocks used

    • Call from your software
    • Read the document
    • Anonymize data
    • Process with AI
    • Restore data
    • Call an endpoint (API)

The building blocks

Simple blocks. Together, they do it all.

You design an automation by connecting blocks on a canvas. Each one does a single thing and does it well; the ones that don't depend on each other work at the same time.

Example of an automation on the canvas: from the start, the document is read; then two blocks work together, an extractor and an AI step; the results are merged, a choice decides the branch and the result is sent to the management system. The branch not chosen is skipped.

Read

Read the document
PDFs, scans, photos, Word, Excel and even emails forwarded as attachments. Long documents are worked in batches of pages, in parallel.
Call a data extractor
Pulls the data out of a document in the format you decided, always the same.

Understand and write

Process with AI
An AI model with your instructions: it summarises, classifies, writes and answers in a precise format, which is checked.
Ask an agent
Hands the work to one of your agents, with its documents and its tools.
AI Tool
Ready-made summaries, translations, minutes and comparisons, even on whole documents.

Decide and repeat

Choice (if… else)
Takes one branch or the other, with simple rules or by letting the AI decide.
Routing
Sends each case down its own path, with a fallback branch for the unexpected.
Repeat for each item
The same work for every item on a list or every document, even all at once.

Act and deliver

Call an endpoint (API)
Reads from and writes to your software: management system, CRM, any service with an API.
Send an email
Writes to whoever you want, or replies to the sender, with the flow's attachments.
Send on WhatsApp
Sends a message or an approved template from one of your numbers.
Create a file
Prepares a CSV or JSON file to attach or deliver.
Call a number
The AI phone service calls a number and reports back the outcome and the transcript.

Protect and calculate

Anonymize data
Replaces personal data with fictitious values before it reaches the AI.
Restore data
Puts the real values back in place, without calling the AI.
Run a script
Exact sums, dates and conversions, with no AI and no credits used.

Organise

Task for the digital employee
Gives a task to your digital employee, with text and attachments.
Merge results
Puts data coming from several steps back together into a single result.
Start another automation
Calls another automation and passes it text and documents.
Automation output
Declares the final result of the automation.

Build with Evo

Describe the process in words. Evo prepares the automation.

In the editor, write what you want to automate: Evo adds the blocks to the canvas, connects them, sets them up and tells you what's left to sort out. You check, correct and save.

  • Everything stays in draft until you press Save, and every step can be undone.
  • If you need a data extractor or an agent you don't have yet, Evo suggests one and only creates it once you say yes.
  • A new automation starts out paused: you test it live, watch the blocks light up along the path taken, and switch it on when you're ready.
Example: you ask Evo to handle complaints. Evo adds six connected blocks to the canvas, from the start to the email for the person in charge, and points out the only field left to complete.

People in the loop

People stay in the loop, when they're needed.

An automation knows when a case isn't its own: it passes it to whoever should deal with it, with everything they need to decide.

  • The right case to the right person

    With a choice or a routing block, cases outside the rules reach a colleague by email or on WhatsApp, with the data and the attachments.

  • A task for the digital employee

    The automation creates a task for your digital employee, with text and attachments. They carry it forward and, if in doubt, ask a colleague.

  • A phone call, with the outcome

    The AI phone service calls a number with the instructions you write, always says it is an artificial intelligence, and passes the outcome and the transcript to the next blocks.

  • A reply to the sender

    It answers whoever sent the email or WhatsApp message that set everything in motion.

  • It starts when you decide

    A new automation starts out paused. You switch it on when you're ready and pause it whenever you like; testing it from the portal always works.

Connections

It connects to what you already use.

No migrations: Evolus works on top of the software you have and starts from events that already happen every day.

Where it starts

  • An email in a Microsoft 365 mailbox or any IMAP mailbox, with its attachments.
  • A message to one of your WhatsApp numbers.
  • A time: every morning, every Monday, even every 30 seconds, in the time zone you choose.
  • A call from your software, with the documents and the values to use for that run.
  • Another automation, which passes it text and documents.
  • A click on «Try it», whenever you want.

Where it arrives

  • Any software with an API

    The «Call an endpoint» block reads from and writes to your systems: method, headers and body tailored to them, even starting from their OpenAPI description. If the response is a file, it becomes a document to read.

  • Reliable delivery

    The result reaches your systems with the key encrypted. If the destination doesn't respond, Evolus retries the delivery alone, without redoing the work.

  • MCP connectors

    Your agents use the connectors in the catalogue, such as Notion, Jira and Confluence, GitHub, Linear, Airtable and Stripe, and from the Business plan your own connectors too. In an automation, the «Ask an agent» block brings them in.

  • The Evolus MCP server

    Connect your favourite AI assistant, if it supports MCP: create, set up and monitor your automations in conversation.

  • API on every plan

    With an API key, your software can start automations on every plan. Full API access comes with the Scale plan.

Control

Every run leaves a trail.

You always know what happened, when and why. No black boxes.

  • What went in, what came out

    For each run you see the starting text, documents and values, and what each block produced.

  • The path taken

    Which branch it chose, which steps it skipped, how many items a loop worked through and with what outcome.

  • What it cost

    The credits spent by that single run, so you can see which automation really weighs.

  • Chains

    If an automation starts others, follow the thread from one run to the next, all the way back to the first.

  • Data quality

    Turn on quality checks and see how often the data read was corrected by hand, field by field.

  • No surprises when things go wrong

    Automatic retries don't redo deliveries that already succeeded. If credits run out, emails wait and everything resumes as soon as you top up.

Example of a run's detail: every block with its outcome and duration, the branch chosen, the skipped block and the credits used by that run.

Plans

From the Growth plan.

Automations are included in the business plans, with a number of automations per plan. Each run uses credits for the AI's work; scripts and data restore use none.

  • Growth

    2 automations

    €199/mo

  • Business

    10 automations

    €349/mo

  • Scale

    20 automations

    €599/mo

The Starter plan doesn't include automations. Prices exclude VAT.

Compare plans

Which process eats up most of your time?

Tell Evo about it: it will tell you how it becomes an automation, with which blocks and from which plan.

Tell Evo about your processWould you rather talk to a person? Book a demo