.svg)
The Agent-a-thon: the corporate competition for building AI agents
Everyone is talking about AI agents. Few really know what they are, and even fewer have built one. The Agent-a-thon changes that in half a day: your teams leave with a working agent, built on their real use cases.
What is a corporate Agent-a-thon?
The Agent-a-thon is a corporate AI competition format. Your teams compete over half a day to design and build real AI agents, connected to your tools and designed for your roles.
The result: you don't leave with yet another PowerPoint. You leave with a catalogue of working agents, built by your own employees and ready to deploy the next day.
They train their teams in AI with AI Sisters
They train their teams in AI with AI Sisters
AI Sisters is an agency specialising in generative artificial intelligence for businesses. We support organisations through AI training, strategic consulting and the AI Sisters Lab, which designs and deploys practical AI solutions.
AI Agent-a-thons in pictures
95%
of corporate generative AI projects
never make it into
production.
Source: MIT NANDA, State of AI in Business 2025
Connected to your data
Your teams connect their agents directly to your business sources (SharePoint, CRM, internal databases, processes).
A competition that engages
Teams compete, challenge each other and share what they discover in real time.
Up and running the next day
Every team has produced a working AI agent, grounded in your business reality.
How does an Agent-a-thon work?
Half a day, 3 main phases, working AI agents.
Strategic scoping
Analysis of your use cases and their potential for agentic AI. Design of a tailor-made programme suited to your profiles.
Introductory AI Show
Expert speakers demystify agentic AI in business: how it works, which tools to choose, with agents running live.
The Agent-a-thon
5 rounds on the agentic platform of your choice. Teams scope, connect, test and deploy their business agent.
Pitch & prize-giving
Pitches of the top 5 agents built, announcement of the winners and group debrief.
What the participants say
Decision-makers who organised an Agent-a-thon for their teams share their experience.
I've taken part in a lot of AI formats in recent months, and this is one of the most practical I've seen. HR teams aren't there to listen to a presentation: they actually build their own agents. Engagement was very high, and several employees kept developing their agents after the event.
What struck me was how much the teams bought in. In a few hours, employees who had never built an AI agent left with concrete results. The feedback was excellent, so much so that we decided to schedule several new editions in other divisions.
The format fitted perfectly into our programme. It's lively, very well paced and above all extremely practical. Participants quickly get into it and leave with agents they can use straight away in their work.
Your teams build their first
AI agents
4 immediate results for your
company
A catalogue of working AI agents
By the end of the session, your company has working agents, connected to your data and ready to be deployed at scale.
Your managers become self-sufficient with agentic AI
No more waiting for IT: your directors and managers now know how to scope, connect and improve an agent themselves.
An agentic AI roadmap for your company
In half a day, you identify which agents to deploy first, which to drop, and the next steps in your AI strategy.
Accelerate your AI transformation
Organising an Agent-a-thon shows in concrete terms (internally and externally) that your company is not on the receiving end of agentic AI: it is in the driving seat.
Your dedicated platform
AI Sisters has built its own platform to collect, assess and compare the agents built, in real time. It keeps the level consistent across teams and gives organisers precise tracking.
Accelerate your AI transformation
Secure login with a password dedicated to the event.
Choice of team and appointment of a Tech Leader as facilitator.
Agents submitted by the Tech Leader at every round.
Direct integration with your agentic platform to build and test agents.
A smooth experience for non-technical profiles.
For the organiser
Automatic collection of all agents built, in real time.
All agent configurations, instructions and tests in one place.
IA Rating assessment of agent quality by team.
Automatic ranking against set criteria (consistency, reliability, autonomy).
Live dashboard that can be projected to follow each team's progress round by round.
5 steps to build a working AI agent
A step-by-step method that turns any business idea into a working AI agent, guaranteed by the end of the session.
From a business pain point to a deployable agent
Each team starts from a real business challenge and turns it, round after round, into a structured AI agent, connected to your data and able to carry out tasks autonomously.
Define the agent's role
In half a day, you identify which agents to deploy first, which to drop, and the next steps in your AI strategy.
Connect the agent to knowledge
Feed the agent with your document sources (SharePoint, files, links) to ground its knowledge in your reality. Deliverable: an agent that answers from your real data.
Make the agent act
Configure the agent to carry out a task autonomously (write an email, create a ticket, update a file). Deliverable: an agent that carries out a concrete task.
Improve answer quality
Optimise the instructions and give the agent reference examples. Deliverable: an agent whose answers are accurate and suited to the role.
Build a workflow
Build a logical sequence of automated steps. Deliverable: an agent that runs a complete process end to end.
✦ Each round is supervised by an AI Sisters trainer who specialises in agentic AI.
Your future business agents
A few examples of agents built by our clients during an Agent-a-thon. Each one adapted to their operational reality.
The "Deeds & Contracts Drafter" agent
Drafts, rewords and harmonises your contracts, amendments and sensitive letters, making sure your legal guidelines are respected.
The "Express Summary" agent
Analyses your company agreements or collective agreements to pull out the key points and red flags in seconds.
The "HR Pen" agent
Produces your internal memos and announcements, instantly adapting the tone to the audience (works council, managers, employees).
The "Methods & Procedures" agent
Turns your working notes into clear operating procedures, practical guides or standardised onboarding checklists.
The "Learning Designer" agent
Builds your internal training modules, assessment quizzes and memo sheets from a simple business objective.
The "Data Reporting" agent
Turns your raw data into visual presentations and annotated dashboards, ready to share with the management committee (CODIR).
Who is the Agent-a-thon for?
All your functions
Sales, marketing, operations, HR, finance, data, tech, communications, legal. The Agent-a-thon is for every function in the company, with no technical or IT prerequisites.
All skill levels
From people discovering agentic platforms to those already using them. The rounds adapt to challenge everyone from their own starting point.
All your key moments
Management convention, extended leadership committee, transformation seminar, AI programme launch, agentic kick-off.
They took part in their own Agentathon
Teams that went from curiosity to mastery. Read their feedback and see the concrete deliverables produced during the competition.
What is an AI agent?
An AI agent is a program based on a language model that can understand a request, fetch information from your data and then carry out a task autonomously (writing, analysing, triggering an action). Unlike a simple chatbot, an agent acts rather than just answering. The best way to understand what an AI agent is, is to build one. That is exactly what the AI Sisters Agent-a-thon offers: in half a day, your teams build a working one on their own use cases.
What is the difference between an AI agent and ChatGPT?
ChatGPT is a general-purpose assistant that answers your questions from its own knowledge. An AI agent specialises in a precise task, is connected to your company data (SharePoint, CRM, internal databases) and can carry out concrete actions. One converses, the other works. During the Agent-a-thon, your teams use both and immediately grasp the difference by building their own business agent.
How do you build an AI agent in a company?
+Building an AI agent follows 5 steps: define its role, connect it to your data, have it carry out tasks, improve the quality of its answers, build a complete workflow. Most companies get stuck at step 1 for lack of method and hands-on experience. The AI Sisters Agent-a-thon was designed around these 5 steps: in half a day, your employees build a working agent from end to end, supported by 5 AI Sisters experts.
Why do 95% of generative AI projects never reach production?
According to MIT NANDA (State of AI in Business 2025), 95% of corporate generative AI projects never reach production. Not for technical reasons, but because teams do not take ownership of them. The project is driven by IT or an outside consultancy, never by the business teams themselves. The Agent-a-thon solves this by having your employees build the agents, on their real use cases, which secures adoption from day one.
Which platform should I use to build an AI agent?
The main agentic platforms in 2026 are Microsoft Copilot Studio, OpenAI GPTs, Claude Projects and Google Gemini. The choice depends on your existing stack (Microsoft 365, Google Workspace, etc.) and your security constraints. The Agent-a-thon is platform-agnostic: we adapt the event to the platform of your choice, or run a comparative format if you are still undecided.
How can I train my teams to design AI agents?
Theory courses on agentic AI are not enough: you have to build an agent to understand how it works. The most effective format combines quick AI upskilling (a 1.5-hour AI Show), supervised practice (3 hours of team building on real cases) and a wrap-up (pitches of the best agents). That is exactly the format of the AI Sisters Agent-a-thon, designed for employees with no technical prerequisites.
How long does it take to build an AI agent?
+A simple AI agent can be built in a few hours by a trained person. An agent connected to your data with a complete workflow takes a supported team about 3 hours. That is exactly the length of the Agent-a-thon: 3 hours of supervised competition, after which each team leaves with a working, tested agent ready to be extended internally.
How much does a corporate Agent-a-thon cost?
The price of an Agent-a-thon depends on the number of participants (up to 80), how far the use cases are personalised, the format (in person or remote) and any follow-on options (strategic debrief, post-event support, scaling up the agents). Contact AI Sisters for a quote tailored to your scope.
What happens to the agents built after the event?
+A simple AI agent can be built in a few hours by a trained person. An agent connected to your data with a complete workflow takes a supported team about 3 hours. That is exactly the length of the Agent-a-thon: 3 hours of supervised competition, after which each team leaves with a working, tested agent ready to be extended internally.
.avif)
