How an AI agent prepares a first conversation with a prospect
My AI agent Jarvis researches prospective clients, while Robert handles the conversations and decisions. Here is how that split of work looks step by step.
The problem: we prepare for conversations too long or too little
Before a conversation with a prospective client, there is a lot of work that nobody sees, but whose absence everyone feels. You need to find out what the company is, what it does, how it communicates and which processes might cause it difficulty. If too little time is given to this work, the conversation starts with general questions, and both sides spend half an hour getting to the point. If a lot of time is given, whole evenings go on reading websites and looking for contacts, and there is no energy left for the conversation itself.
What is more, searching and research are repeatable work. This is exactly the kind of work that can be handed to an assistant, as long as you clearly separate gathering information from the human decision.
What this looks like in a company
A typical example: a business owner or salesperson receives an enquiry or decides to contact a company themselves. Then several manual stages follow:
- finding the company's website, social media accounts and public information;
- understanding what the company does and who it sells to;
- reading how it describes its services;
- recording notes so they do not get lost across different files and emails;
- working out what to offer in the first place.
Each step seems insignificant on its own, but together they take up a considerable part of the working day. And if there are several conversations, it repeats every time.
The solution: dividing the work between the AI agent and Robert
My AI agent Jarvis handles client search and research. Conversations and decisions stay with Robert. This division is deliberate: the agent prepares the groundwork, and the human takes responsibility.
The process looks like this.
Step 1. Search. Jarvis looks for prospective clients that match what AI-Lab Space offers: companies that may have repeatable processes, document flows or scattered data.
Step 2. Research. For each company it finds, the agent gathers publicly available information: what the company does, how it works with clients, and where there are opportunities to organise or automate something.
Step 3. Summary. The result is not a long pile of information, but a short overview that can be read before the conversation.
Step 4. Robert's check. Robert reads the overview and checks it. He looks at whether the conclusions are credible, whether anything has been misunderstood, and whether this company is suitable for a conversation at all. If something seems doubtful, it is corrected or discarded.
Step 5. The conversation. Robert leads the conversation himself. He talks to the person, listens, asks questions and understands what the client really wants.
Step 6. The decision. Whether to continue, what to offer, how to do it and whether cooperation makes sense at all is decided by Robert. The agent does not take part at this stage.
Why this particular split
Research is work that calls for patience and a systematic approach. Here the agent is useful, because it does not get tired of repetition. In a conversation, on the other hand, understanding, trust and the ability to read a situation matter. That is human work.
The checking part is also important. The information the agent finds is prepared material, not ready-made truth. That is why Robert checks it before the conversation rather than trusting it blindly. This principle suits any company introducing AI agents: automate the preparation, but leave responsibility with a person.
What this gives you
- Less time on preparation. Gathering information no longer takes up a person's working time.
- A better-considered conversation. The context is already clear at the start, and you can talk about important matters straight away.
- A uniform approach. The overview prepared for each prospective client follows a similar logic, so they are easy to compare.
- Clear responsibility. It is known who finds the information and who makes the decisions.
- Less chaos in notes. Research results are in one place, not scattered across emails and files.
Where to start
If you want to introduce the same approach in your company, do not start with a complicated system. The steps can be simple:
- write down what work you do before each conversation with a client;
- mark which of it is only searching for and compiling information;
- decide which parts need your experience and judgement, and keep those for yourself;
- determine at which point a person checks the material prepared by the agent;
- start with one type of work and check whether it really makes the work easier.
The main thing is not to hand over to the assistant what a person must decide. Search and research can be handed over, but conversations and decisions stay with you.
Which pre-conversation task would you like to hand over to an assistant?