Case Studies / Next Decision
How Next Decision Automates Lead Qualification with Artificial Intelligence
The story of Noé BONNE, an AI consultant who transformed a time-consuming manual process into an intelligent automated system
%201.png&w=750&q=75)
A consulting firm facing a growth challenge
Next Decision is a consulting and expertise firm that has been supporting companies in all things data for over 20 years. With around 250 employees across 10 offices throughout France, the company is based in Nantes and now operates nationwide.
"We support all types of companies—small, medium, and large enterprises—across all sectors," explains Noé BONNE, AI consultant at Next Decision, responsible for implementation and training. "Historically, we have been focused on data, and we have expanded into several other areas: artificial intelligence, governance, CSR, master plans... Overall, if it involves data, we know how to handle it."
"Qualifying these leads, these incoming messages, was something that took us time and, above all, we lacked responsiveness. At Next, we don’t really have dedicated salespeople; all the consultants handle this. As a result, sometimes processing all these incoming messages and responding quickly—whether it’s a client requesting a service or a CV submission—we weren’t always fast enough."
But even for a data expert like Next Decision, an operational problem persisted: qualifying incoming leads.
Each week, around ten messages arrive through the website contact form. Requests of all kinds: large companies seeking strategic support, SMEs looking to structure their data, as well as students submitting their resumes or less qualified inquiries.
The manual process was time-consuming: for each message, a consultant had to research the company online, check if it already existed in the CRM, identify the contact’s position and experience, assess the relevance of the request, and then create or update the information in the system.
The Trigger: Leveraging Data to Unlock AI
For Next Decision, adopting artificial intelligence didn’t come from a trend, but from a natural business logic.
"The trigger for us is that we have strong integration with companies and their data foundation," explains Noé.
"Our clients—usually, we are the ones who built their entire data foundation. And what's interesting with AI is that you can generally only use it when you have quality data."
"The trigger for us is that we have strong integration with companies and their data foundation. These are our clients, and usually, we are the ones who built their entire data foundation. What’s interesting with AI is that you can generally only use it when you have quality data."
But even for a data expert like Next Decision, an operational problem persisted: qualifying incoming leads.
Each week, around ten messages arrive through the website contact form. Requests vary widely: large companies seeking strategic support, SMEs looking to structure their data, as well as students submitting their resumes or less qualified inquiries.
The manual process was time-consuming: for each message, a consultant had to research the company online, check if it already existed in the CRM, identify the contact’s position and experience, assess the relevance of the request, and then create or update the information in the system.
Why Swiftask? A carefully considered strategic decision
Faced with the numerous solutions on the market, Next Decision tested several platforms: ChatGPT, Claude, and even internal proofs of concept with open-source models like Llama.
Geographical and human proximity
"It’s a company from Nantes. We like working with nearby companies," explains Noé. This proximity isn’t just about distance: "Responsiveness is… we immediately saw that with the Swiftask team, if we request a new feature, they’re on it within a week."
An unbeatable value for money
"The value for money is extremely competitive, and that’s perhaps the most important factor," emphasizes Noé. "Starting on a platform, the entry cost can be quite high if you need many licenses. With Swiftask, you can really start very small at a very low cost, which is great for scaling up."
The versatility of the integrated tools
"What’s interesting with Swiftask is that sometimes we enter a company, map out the business ecosystem with their CRM, ERP, and the various software we might need to connect to, and we realize that all of this is already integrated into Swiftask," explains Noé. "All this initial tool setup is already done, so everything moves really fast afterward."
Governance and security
"Usually, we needed a platform for governance and to have a shared platform for all employees," explains Noé. Swiftask meets this critical need: eliminating Shadow IT, where everyone uses ChatGPT on their personal PC without oversight.
"When I discovered Swiftask, I thought, 'let’s go for it.'"
The solution: an AI agent that does the work for you
Next Decision’s idea is simple but powerful: place an AI agent between incoming messages from the website and the consultants to do the heavy lifting for them.
"The idea is to place an AI agent between the incoming website message and us, simply to do the heavy lifting for us."
When a message is sent through Next Decision’s website form, the AI agent automatically retrieves it and starts a series of searches:
Company research
Identification of the company, its industry, location, and revenue via Papers
Contact research
Position in the company, experience, professional background
CRM review
Check if the company is already a client, if there are ongoing processes, and identify the account manager
Score assignment
Evaluation of the lead’s relevance on a 1-to-5 star scale
Automatic CRM update
Creation of the contact and company if necessary, or updating existing information
"These are searches we used to do manually, but now they’re automated," explains Noé. "We have all this information right in front of us when we handle the qualification."
During the webinar, Noé conducted a live demonstration using a fictional example: Tim Cook from Apple US contacting Next Decision for support on process optimization.
In a few seconds, the agent has:
Identified Apple US, its industry, and its revenue
Found Tim Cook’s position (CEO) and his experience
Assigned a 5/5 score ("strategic AI project aimed at process automation, high potential")
Automatically created the opportunity and contact in the Zoho CRM
"No human intervention in this entire process," emphasizes Noé. "Human involvement comes when we are in our software with the list of all incoming messages, and that’s when we can handle them."
An accessible setup: no code, just business logic
Not a single line of code
"There really isn’t any code," emphasizes Noé. "We take the building blocks, the connectors, tell it to look in this Drive folder that’s automatically updated. We describe what it should do in natural language."
Three simple steps:
Choose the brain (the LLM model)
"We chose GPT; it’s the same model as ChatGPT," explains Noé. "We tried several, and this one offers the best performance at the lowest cost." The cost is transparent: one Swiftask credit per word generated.
Select the tools
This is where the magic happens. The agent receives a "toolbox" with:
The ability to perform internet searches
Searching in Papers for reliable data
Connection to the Zoho CRM via an MCP server
"When we add tools, I click 'add a tool,' search for 'web search,' and voilà, I have all the tools. I can just click and add it," demonstrates Noé. "It’s really like Legos. It’s your little Sims that you build."
For non-native tools like Zoho, Swiftask offers MCP servers: "Any software, no matter how niche, a business application… the fact that Swiftask manages MCP servers means you can build your own tool, and it’s relatively simple."
And the most interesting part: "Once we’ve built it for one company, any other agent I create can use this tool that’s already built. It’s shared, and once it’s created, it works for everyone."
Give the instruction
"We describe its role," explains Noé, showing the agent’s instruction. "It’s an agent tasked with analyzing incoming leads at Next Decision. We explain our business sectors so it knows where we can respond or not."
The instruction outlines the processing logic, step by step: analyze the lead, research the company, check the CRM, identify the contact’s role, assign a score, and create the opportunity.
"It’s really chronological, a business-oriented approach," confirms Noé. "It’s what we could give to an intern if we wanted to explain how to do this task. We explain: you start by doing this research, then that
Immediate and measurable results
The impact of this agent on Next Decision is direct and measurable.
A significant time saving
"The clearest ROI is simply how much time we’ve saved," explains Noé. "We have around ten messages per week. For each message, we would spend about ten minutes doing all the research in our CRM, on the internet, on the client, and assessing the interest."
Calculated ROI
"We are a relatively small group where the volume isn’t huge. We have many clients for whom we handle similar use cases, except that instead of 10 messages per week, it would be 10,000. And there, the ROI is different."
And that’s just for 10 messages per week!
Imagine the impact with a higher volume. Every minute saved translates into more time for your high-value activities.
Increased responsiveness
No need to wait for a consultant to be available to qualify a lead. The agent works 24/7, instantly. "We weren’t always responsive enough. Now, as soon as a message arrives, it’s processed immediately."
Immediate scalability
"Once we have the tool, it scales easily," emphasizes Noé. "We spend more time automating and optimizing than building the agent."
A CRM that’s always up to date
"Sometimes, the sales team doesn’t always have the time to create or update the CRM, add the contact, or input the information," notes Noé. "We don’t always do it automatically and quickly. Now, it’s done systematically."
Enhanced lead qualification quality
The agent doesn’t get tired, misses no steps, and consistently applies the same rigor. "The agent is smart. It doesn’t just create something new every time; it checks if the contact already exists, and if it does, it links it. Same for the client. It’s truly equivalent to what we would do manually."
Beyond lead scoring: a comprehensive AI strategy
The lead qualification agent is just one part of a broader AI strategy at Next Decision.
Noé’s recommendation: "It’s useful to have guidance at the start to get the right direction, know which use cases to tackle, and which tools to build. But once you have that, the goal is for you to be completely autonomous with this platform."
Other deployed use cases:
Automated competitive intelligence
"The first cases we implemented were for competitive intelligence," says Noé. "The principle is quite simple. We want an agent to scan everything happening with competitors online every week or month, all the news, and send us a short summary email. It's easy to set up and incredibly fast in Swiftask."
The HR agent for internal processes
"Our HR team frequently answered questions and emails from employees requesting information contained in Next's documents," explains Noé. "Searching through the 30 PDFs we have and simply sending an email takes time and isn't something where we add value."
Solution: an agent integrated into the intranet, accessible to all, which instantly answers questions about HR processes based on PDFs automatically updated on the Drive.
The technical agent for documentation
"We did the same thing with a technical agent who comes to look in the internal documentation that we have in the company on all the processes, the technologies that we use, and comes to answer the employees on this," explains Noé.
A gradual and strategic adoption
"It’s something relatively simple to experiment with, and that’s the advantage of AI agent tools like Swiftask," explains Noé. "Setting up a new agent is easy, and a business user can create their agent and try things out."
"It's something that's relatively simple to experiment with, and that's what makes AI agent tools like Swiftask so advantageous. The basic setup of a new agent is easy to do, and a business user can create their own agent and try things out. It's helpful to have support at the beginning to get the right direction, know which use cases to use, and which tools to build. But once you have that, the goal is for you to be completely autonomous with this platform."
A surge in demand
Next Decision's expertise in AI does not go unnoticed.
This explosion can be explained by two converging needs:
Security and Governance
"There is a great need for data security. We want to stop Shadow IT and clients are coming in saying: we know that everyone is using ChatGPT on their PC, this is not possible. We need to have an enterprise-wide platform where we can govern all of this."
The speed of implementation
"We quickly move towards automation because once we have this platform, as you have seen, it is quick to set up," explains Noé.
"Just this month, I think we have 5 or 6 clients, 5 or 6 clients where it's really ongoing. Most with Swiftask, other clients with other platforms. These are really major accounts."
Deployment times
A few days to a maximum of ten days
10 days
"To go from nothing to 'I have an AI platform in my company and several use cases deployed', we're talking about a few days, ten days, maybe a little more. It's something that happens super fast and that's what customers like right now."
Expert advice to get started
Drawing on his experience, Noé shares his recommendations for companies wishing to get started.
Experiment without delay
"The advantage of these tools is that you can test them right now," encourages Noé. "It's not a complicated tool to learn. You can choose ChatGPT or take out a free one-week subscription to Swiftask; you have the platform, you run tests, you create agents. It's something that goes very quickly."
Get support for the strategy
"When you're trying to build an AI integration strategy within your company, it's beneficial to get support," recommends Noé. "We have experience in this area and we've seen what happens with other clients. We can guide you toward the use cases with the greatest added value and which ones we can address as quickly as possible."
Build a shared tool library
"It's interesting at the beginning to think: we're going to build the tools that allow each agent to connect to our software," explains Noé. "Once we have these tools, building the agents can go super fast because we have all the building blocks."
Aiming for autonomy
"We often work collaboratively, building with you, and that's also the goal," explains Noé. "The goal is for you to be completely autonomous with this platform."
Key takeaways
The story of Next Decision demonstrates that integrating artificial intelligence into business is not reserved for tech giants or companies with colossal budgets.
And this is just the beginning. As Noé points out: "We are a relatively small group. We have many clients for whom the volume is enormous. Instead of 10 messages per week, it will be 10,000. And there, the ROI is different."
Artificial intelligence does not replace Next Decision's consultants. It frees them from repetitive tasks so they can focus on what really matters:
strategic support for their clients, building innovative data solutions, creating value.
"AI is becoming transparent," Noé concludes. "It's the integration of AI tools into business processes. A Swiftask agent is running in the background, but you can't see it."
"The ROI is obvious. We're trying to automate a process that we used to do manually. The clearest ROI is how much time we've saved."
Identified a time-consuming process: lead qualification (15 minutes per message)
Choose the right tool: Swiftask for its flexibility, value for money, and governance
Create an intelligent agent in a few hours, without coding
Immediate results achieved: one day of work saved per month
Deployed a comprehensive AI strategy: competitive intelligence, HR support, technical documentation
Shared expertise: 5 to 6 clients supported simultaneously