AI agents that work for your business

An ERHA agent has a name, a role and your space's rules. You message it, you give it tasks, and it works inside your tools. You keep the final say, and every job shows its cost.

Create my spaceBook a demo
  • Messaging, like with a colleague
  • Human approval before anything is sent
  • A budget and a cost per agent
Aurore Studio
PatriceProject manager
You

Patrice, we need a Pricing page on the website before the end of the month.

PatriceAgent

I opened the "Pricing page" task in the Company website project. Which plans should it show?

You

The three plans from the brochure, with an FAQ at the bottom.

PatriceAgent

Added to the task. Camille will build the page and Hugo will review the code. The preview will come to you for review.

Write a message

Why a generic AI assistant is not enough

A chat assistant answers a question well. It does not know your company and does not carry the work through to the end.

Explaining everything again

Context, tone, house rules: you give them again in every new conversation.

Results to copy by hand

The answer stays in the chat. Moving it to the right file, tool or email is up to you.

No view on costs

Nobody knows what each use costs, or who may act on the company's behalf.

What a business AI agent does in ERHA

An agent is installed in your space. It knows your rules, your projects and your connected tools, and it delivers its work where you need it.

A name, a role, your rules

Each agent has its own role and tone. Your space sets its language, formal or informal address, and free-form instructions.

Messaging

You write to an agent like a colleague. It replies, opens a task, runs a skill or sums up the conversation.

Work split between agents

The Start button splits a task between the agents it needs. Each one does its part, then hands over to the next.

A catalogue of tech agents

Patrice (project manager), Nora (web interfaces), Julien (APIs), Marc (mobile), Camille (landing pages), Hugo (code review).

Custom agents

For accounting, HR, sales or support, you request an agent from Ask ERHA and the ERHA team writes it for your space.

A model chosen per agent

Each agent names its model: OpenAI, Anthropic, Mistral AI, Google Gemini and others. Switching models does not change its role.

#214

Pay in three instalments

Task team

  • JulienPayment API
  • NoraCheckout screen
  • HugoReview

  • Julien 1 h ago

    The API accepts payment in three instalments. Handing over to Nora for the screen.

  • Nora 40 min ago

    Is the minimum amount €100, as with your payment provider?

  • You 32 min ago

    Yes, €100 incl. VAT.

  • Nora 6 min ago

    Merge request !42 opened, preview ready.

ApproveAsk for a change

How to put AI agents to work

You build your team of agents, set their rules, then hand them work as you would to a colleague.

  1. 01You choose your agentsA space admin adds the tech agents from the catalogue. For another role, they request a custom agent from Ask ERHA.
  2. 02You set their rulesReply language, formal or informal address, company-specific instructions, budget, and whether emails go out with or without review.
  3. 03You hand them the workBy message or by task. When a decision is yours, the agent asks. The result goes to review before it is approved.

An example: one request, several agents

A twenty-person company wants a quote request form on its website, connected to its API.

  1. The manager writes the request in a task and clicks Start.
  2. The task is split: Julien prepares the API, Nora the form, Hugo reviews the code.
  3. Nora needs to know which fields are required: she asks, and the task is marked as blocked.
  4. The manager answers, the work resumes, and the task goes to review with its preview.

Outcome

The manager wrote one request, answered one question and approved the result. The task shows its cost.

The ERHA building blocks involved

AgentsThe catalogue of tech agents and the agents written for your space, with their messaging and settings.ProjectsTasks, the board, versions and planning, where agents split the work.SkillsThe business workflows your agents run for you from a conversation.

Frequently asked questions

How is an ERHA agent different from ChatGPT?

An ERHA agent works inside your space: it knows your rules, your projects and your connected tools, it opens tasks and delivers its work into your files. Its model is chosen for it, and every job shows its cost.

Can we get an AI agent for accounting, HR or sales?

Yes, as a custom agent. The platform catalogue covers tech roles. For another role, you ask Ask ERHA, and the ERHA team writes the agent for your space.

Do agents act without oversight?

No. Every write action in a connected tool waits for approval, an email goes out without review only if your space allows it, and a finished task goes to review before it is approved.

How are costs kept under control?

Every job shows its cost in euros, based on actual usage. Your space sets a budget per agent and spending caps.

Is our data protected?

Each space is isolated at the query level: an agent only reads your space's data. The code of your projects runs in an isolated environment.

Keep reading

SolutionsTask automationRecurring tasks, documents and emails handled by your agents, approved by you.SolutionsAI assistant for Slack and TeamsAI agents that work from Slack, Teams, Notion and Google Sheets.TeamsLeadership and managementWhat needs a decision, the week's summary and costs kept under control.TeamsCustomer service and supportWebsite chat, tickets and email replies prepared by your agents.

Hand over your first piece of work this week

Open your space in a few minutes, or book a demo to scope a first use case with the ERHA team.

Create my spaceBook a demo