Skip to main content
InApp-Agent

InApp-Agent by STARTUP-UP

Bring agentic AI to your application.

Your users ask. The agent gets it done.

Once connected to your application, the agent uses your data to guide users and carry out authorised tasks. Right inside your software.

  • Hosted in France
  • Backup site in Germany
  • Encrypted at every step
  • Kept separate per customer

Case management

Case

Demo case

History
12 Sept · inbound call — asked to be called back Tuesday morning
Quote
DEMO-014 sent on 9 Sept
Follow-up scheduled

Call-back · Tuesday 10:00 – 10:15 · scheduled

Hello Camille, as agreed I will call you back on Tuesday morning to go over quote DEMO-014.

Draft prepared by InApp-Agent · review before sending

InApp-Agent

by STARTUP-UP
Added to your application

Context: Demo case

Prepare the follow-up for this case.

Done, in your application

  1. Call-back Tuesday 10:00, as requested on the phone
  2. Follow-up message, as a draft
  3. Status → Follow-up scheduled

The message is not sent: you review it.

3 changes recorded in the history · approved by you

Today, six steps. With InApp-Agent, one sentence.

Today

Too many menus, too many settings, too many selections.

TodayScreen of a fictional business application: the six-step path to schedule a call-back is highlighted and a seven-field form is open.

Illustration — fictional application

  • Find: a menu, a sub-menu, filters, a tab.
  • Enter: a form, seven fields.
  • Finish: change the status, write the follow-up somewhere else.

With InApp-Agent

One sentence, one click.

With InApp-AgentInApp-Agent panel: the request, three actions ready, the limit “the message is not sent”, the Apply button.

Illustration — fictional data

  • One sentence: “Prepare the follow-up for this case.”
  • Three actions ready, nothing more.
  • One click: Apply, and it is saved in your software. Nothing is saved without your approval.

Same case, same application, no menu to learn.

Your application, data and rules. An agent that takes action.

  1. Connect your application.

    We define the information the agent can access and the actions it can use.

  2. Your users ask.

    The agent understands the request and finds the relevant information.

  3. Work gets done in the software.

    The agent carries out authorised tasks, asks for approval when required and reports the outcome.

Try the agent for free. See what it actually does.

Discover a conversation that leads to a result in a real application. Try it yourself, then imagine what the agent could do in your software.

Sign up with your email, 14 days free, no credit card required.

The trial application is currently available in French.

Three successive block arrangements: a scattered pile, an organised proposal, then a finished, clearer structure.

The agent remembers your cases and draws on your documents.

Memory

Users pick up their work without explaining everything again. They see what the agent has kept, and can correct or erase it.

Explore memory

Sources

Your documents, your industry news, your competitors: you choose what feeds the answers. If the information is missing, the agent says so — it never makes things up.

Explore knowledge and monitoring

The agent spots what gets in your users’ way and suggests better.

Recurring questions, tasks dropped halfway, misunderstandings: the agent spots them. Each frequent difficulty becomes an improvement, tested before it is switched on. That is Vigie: the agent improves, under your control.

Explore Vigie

Illustrative cycle

  1. Difficulty detected

    Several users abandon the same task at the same point.

  2. Improvement selected

    The missing question is asked earlier in the exchange.

  3. Result verified

    Tested and measured before being switched on. No change without your approval.

Hosting and compliance

Your data stays in France, protected and kept separate.

What is in place today.

  • Hosted in France

    With a backup site in Germany should the first one stop. Your data and its copy stay in Europe.

  • Encrypted at every step

    Unreadable to anyone not authorised, in transit as in storage. The encryption keys are kept by us, separately from the data.

  • Kept separate per customer

    Your data never mixes with another customer’s: the separation is enforced inside the database itself.

  • Traced

    Every agent action is recorded: you always know who did what, and when.

Continuity and recovery are set out in the contract. Technical details on request.

A calm glass-walled server room in a light stone building; conceptual illustration, not a real site.

Hosting, GDPR, AI Act: the Trust page

Already in real use

BANANA-CONTENT, a STARTUP-UP application, before and now.

BeforeFormer BANANA-CONTENT welcome screen: a six-option role questionnaire.
Before. A questionnaire on arrival, dashboards to read on your own.
NowBANANA-CONTENT dashboard with the agent panel open and its analysis.
Now. The agent welcomes, reads the results, suggests.

« Teste un nouveau contenu consacré à l’équipe sur LinkedIn »

Exact excerpt from the agent’s answer (in French: “Try a new post about the team on LinkedIn”)

In your application: the same mechanism, on your cases and your indicators.

Read the use case

BANANA-CONTENT screenshots · demo data · analysis produced after confirmation. BANANA-CONTENT is a STARTUP-UP product, not a third-party customer testimonial.

Frequently asked questions

Do we need to replace our existing software?

No. InApp-Agent is added to the application you already run, which keeps its reference data, its rules and its execution responsibilities.

What can the agent do?

It answers, advises, prepares work and carries out the tasks you have authorised. Exactly what it can do depends on the features and permissions connected in your application.

Where is the data hosted?

In data centres in France (Paris region), with a backup site in Germany; database in Paris, up-to-date copy in Frankfurt. The encryption keys are kept by STARTUP-UP, separately from the data. Where AI models process your data is specified for the chosen configuration.

Can we choose private AI?

Yes. An AI reserved for you, or the agent installed on your own servers: both options are assessed and sized with you.

How do we get started?

Try the agent for free in BANANA-CONTENT to see what it does, then let's discuss your application and the tasks to hand over.

Give your application the ability to act.