ENUA

Book a demoJoin the waitlist

Sell & talk

Deliver

Run the business

Intelligence

Platform

Every module works on the same records, in your own workspace.

AI Studio · part of the AI module

AI agents that work like colleagues.

Give an agent a job, the skills it needs and a level of autonomy. Assign it tasks, put it on a schedule, and approve anything it shouldn’t do alone. Everything it does is in the history.

Join the waitlistBook a demo
Capabilities · 25

Agents for the work that keeps coming back.

Agents, the skills they use, the knowledge they share and the approvals that keep them in line.

  • 25capabilities
  • 6areas

Agents

6
  • Assist
  • Semi
  • Autonomous
  1. 12 preset agents.APROS comes with 12 preset agents, including the AI Brain and two that work in the background. Specialists such as a Data Analyst, Researcher, Document Manager, Image Creator, Call Summarizer, Sales Call Analyst, Action Planner and Deal Assessor are ready to use, or create your own.
  2. Your own agents.Give an agent a name, an emoji and a color, and set its personality, model, temperature and extended thinking. Decide whether people can chat with it.
  3. A level for each agent.Assist runs only reads. Semi-autonomous also makes internal changes. Autonomous asks only before deleting.
  4. Agents that delegate.Let an agent hand parts of a job to the agents you choose, with a cap on how many work at once.
  5. One overview of the work.See which agents are active, today’s runs and tokens, every schedule, and each agent’s open and blocked work.
  6. Talk to any agent.Open a chat with an agent directly from its page.
Show all 6Show fewer

Agents at work

6
  • Work
  • Schedules
  • Pulse
  1. Assign a task to an agent.Pick an agent as the assignee, just like a person. Watch its runs on the task, or press “Run now”.
  2. Schedules.An agent runs a prompt on a cadence, with a checklist and a repeat limit, in your company’s time zone. The result goes to a task, a notification, a chat or Telegram.
  3. Pulse.An agent wakes every 15 minutes or more, or once a day, and checks its own tasks.
  4. Working hours and limits.Set when an agent works, how many steps a run may take, and whether it pauses after an error.
  5. Memory you can read.Agents remember facts, preferences, rules and lessons across conversations. You can view and edit every memory.
  6. History and quality.See each agent’s runs, success rate, average tokens and tools used, plus your team’s thumbs-up / thumbs-down ratings over time, by skill.
Show all 6Show fewer

Skills

6
  • SKILL.md
  • references/
  • scripts/
  1. Skills give agents their tools.Attach skills to an agent. What it can do follows from those skills, the MCP tools you connect to it, and a small system kernel.
  2. 18 ready skills.Word documents, presentations, Excel and CSV, print-ready PDF, interactive pages, diagrams, deep research, reports, CRM work, mail work, call follow-up, tasks with checklists, reminders, recurring work, warehouse and catalogue, the finance ledger, file work and checking results.
  3. A skill catalogue.Create, edit, search and filter skills, or duplicate a ready one and make it yours.
  4. Skills as folders.Edit a skill as a folder (SKILL.md, references, scripts) right in the browser. Bundled scripts run in an isolated container.
  5. Bring your Claude skills.Import a skill from a .zip or a pasted SKILL.md, and download any skill to use elsewhere.
  6. A manifest for every skill.Each skill says what it does, where it stops, and which tools it needs or may use.
Show all 6Show fewer

Approvals and safety

2
  • Approve
  • Deny
  1. One Approvals inbox.Every paused agent action lands in one inbox with a count badge. Approve or deny it there, from a notification link, or in the Telegram Mini App.
  2. Outside text is flagged.When a run acts on customer messages, email or web pages, it’s marked and logged.

Knowledge

3
  • Company Brain
  • Entity Index
  • Knowledge Graph
  1. Company Brain.Keep what every agent should know in one place: your legal and brand profile, brand voice, company facts and reference documents such as brand books, price lists and regulations. Agents suggest new facts, and you approve or reject them.
  2. The knowledge index.AdminChoose which of 10 data types are indexed for search by meaning, reindex or purge, and pick the embedding model.
  3. Knowledge graph.AdminExplore what your company knows as an interactive graph.

Connect more tools

2
  • MCP
  • HTTP
  • SSE
  1. MCP connections.Connect external MCP servers (HTTP or SSE, an auth header, a tool allowlist, a connection test) and give their tools to your agents.
  2. All of it in one hub.Agents, Skills, Workflows, Automations, MCP, Company Brain and Approvals are tabs of one AI Studio.
Assign it like a person

Give the research task to the Researcher.

Olena needs a shortlist of furniture distributors in Poland by Thursday. She creates the task and picks the Researcher agent as the assignee. The agent searches the web, reads the sites, and saves page screenshots to the Research folder. It posts a report with sources on the task. Olena reads it over coffee and sends it for review.

Schedules

The pipeline report arrives in Telegram every Monday at 9.

Maria schedules the Data Analyst for Mondays at 09:00 with a three-point checklist: new leads, stalled deals, invoices awaiting payment. She sets delivery to Telegram. Each week the summary lands in her Telegram before stand-up.

Approvals

It drafts the email. You decide whether it goes.

The Deal Assessor runs at the Semi-autonomous level. It reviews Harbor & Pine’s deal, updates the next step and drafts a follow-up email to Emma. Updating the deal is an internal change, so it runs. Sending the email goes outside the company, so it waits. Maria gets a notification, opens the Approvals inbox, reads the draft and approves it.

Skills

Bring the skill you already wrote.

Your team already has a Claude skill for writing proposals in your house style. Import the .zip into AI Studio, open the folder in the browser, and adjust SKILL.md to point at your price list. Attach the skill to the Document Manager. From then on, that agent can use the tools the skill declares.

Guardrails

Where it stops.

Every agent has a written boundary. You set it, and you can read it at any time.

  1. Tools come from skills.An agent can use the tools its skills declare, the MCP tools you connect to it, and a small system kernel. A skill’s manifest lists its tools.
  2. The same permissions as people.Agent tools check roles, row access and modules before they act. Changes are recorded with the agent that made them.
  3. A level per agent.Assist: every change waits. Semi-autonomous: outbound, destructive and delete actions wait. Autonomous: deletions wait.
  4. Channel agents reply on their own by default.An agent attached to a messaging channel answers customers without approval in First Line (the default) and Autonomous mode. Assistant mode answers only when your team asks.
  5. Outside content is marked.Runs that act on customer, email or web text are flagged and logged.
  6. Company Brain is shared by design.Documents and facts in Company Brain are available to all employees and agents. Keep them to what everyone may read.
  7. Memory is visible.You can read and edit what each agent remembers.
  8. No training on your data.We don’t train or fine-tune models on your workspace data. Agent “learning” means the memories it stores in your own workspace, which you can read and edit.
Show all 8Show fewer
Connected

Connected to the rest of APROS.

  • AI Brain

    The Brain delegates parts of a job to your Studio agents and shows their progress in the chat.

    See AI

  • Automation

    Rules and workflows are tabs of AI Studio. Run an agent when a deal is won, or put one inside a workflow.

    See Automation

  • Projects & tasks

    A task’s assignee can be a person, an AI agent or a Claude Code machine.

    See Projects & tasks

  • APROS in Telegram

    Approve paused agent actions in the Telegram Mini App, and get scheduled results as Telegram messages.

    See APROS in Telegram
FAQ

AI Studio questions.

Is AI Studio in every plan?

AI Studio comes with the AI module, together with automations and workflows. On the launch price list, AI agents are named in the Business plan. Ask us about the others.

How many agents can we have?

APROS comes with 12 preset agents, including the AI Brain and two that work in the background, and you create your own on top of them. Each one has its own skills, level and schedule.

Can an agent do something we didn’t give it?

No. Its tools come from the skills you attach and the MCP servers you connect, and every tool checks the same permissions as a person. Its level decides what waits for approval, and deletions always wait.

Who can see what’s in Company Brain?

Everyone in the company, and every agent. Company Brain is for shared knowledge such as your brand voice, facts and reference documents, so keep private material out of it.

Can our agents use our own tools?

Yes, through MCP. Connect an external MCP server, choose which of its tools are allowed, test the connection, and give the tools to your agents.

Put your first agent to work. Join the waitlist, or book a demo and we’ll build an agent around one of your real jobs.

Join the waitlistBook a demo