Designing a sales copilot
How can you design and test an AI copilot suited to the work of your salespeople?
3500 € per session, excl. VAT
In short
Two days to design a sales copilot tailored to your team: task selection, instructions, an authorised knowledge base, a prototype configured in the available tool, tests using fictional cases and a pilot plan, with no promise of deployment at scale.
Why this training?
Many sales teams use AI individually and in a fragmented way. An assistant configured for a few specific tasks (preparation, summarising, drafting) enables good practices to be shared, provided it is well designed and tested. These two days take you from identifying tasks to a tested prototype, with its instructions, authorised sources, limitations and a pilot plan. Actual deployment will be subject to a separate decision and support. Running case study: The fictional sales team at « Clario », a planning software publisher, designs a copilot that prepares meetings, summarises exchanges and suggests responses to objections based on its documents. Transfer to the workplace: The organisation has a tested prototype and a pilot plan awaiting a decision; any move into production will require separate IT and security approval.
Objectives of this training
- Choose three to five sales tasks suited to a copilot
- Write the instructions and expected behaviour of the assistant
- Select the authorised documents that will form the knowledge base
- Configure a prototype in the tool available within the organisation
- Test the prototype using cases and an evaluation rubric
- Write a pilot plan with its rules of use
Who is it for and what are the prerequisites?
Roles and tasks concerned
- Sales managers, sales ops and sales enablement
- AI leads within sales teams
Prior skills
- Be proficient in sales applications of generative AI (level 3 or validated equivalent prior learning).
- Have responsibility for sales organisation or tools (manager, sales ops, AI lead).
- Have access to a tool that allows a customised assistant to be created and is authorised by the organisation (checked during scoping).
What you will work on
Which tasks should be assigned to a copilot?
Objective: Choose useful and manageable tasks. Criteria: frequency, value, risk, data required; tasks to exclude.
What the available tool allows
Objective: Start from actual capabilities. Features of the authorised tool (customised assistant, documents, sharing) checked in its official documentation; limitations and permissions.
Writing the instructions
Objective: Define the copilot's behaviour. Role, audience, tasks, response format, refusals, mandatory statements, tone.
Knowledge base
Objective: Provide reliable and authorised sources. Document selection, cleaning, updating and ownership; what must never be included.
Building the prototype
Objective: Obtain a first working version. Configuration in the tool; initial trials; corrections to the instructions.
Test suite and evaluation rubric
Objective: Evaluate the prototype objectively. Normal test cases, edge cases and traps; criteria: accuracy, relevance, compliance with instructions, expected refusals.
Testing and improving
Objective: Correct the defects observed. Running tests, analysing failures, iterating on instructions and documents.
Rules of use and responsibilities
Objective: Set a framework for using the copilot. Authorised data, checking responses, reporting errors, owner and updates.
Pilot plan
Objective: Prepare a controlled trial. Pilot group, duration, usage and quality indicators, decision criteria, risks.
What you take away
- List of prioritised tasks for the copilot
- Version-controlled copilot instructions
- Documented knowledge base
- Configured prototype and test report
- Rules of use and pilot plan
Method and assessment
- Short teaching inputs followed by immediate practice on the running case study.
- Individual work followed by work in pairs, with peer review of outputs.
- Guided group discussion: what works, what needs correcting and what remains a human decision.
- Preparation: List the sales team's repetitive tasks. Gather authorised public or internal documents: product sheets, sales pitches, FAQs. Check the available tool and permissions with IT.
Observable criteria: Each selected task is justified against explicit criteria; The prototype passes at least the normal test cases defined by the group; Failed cases are documented along with their corrections; The pilot plan specifies the group, duration, indicators and decision criterion.
Frequently asked questions
Will the whole team be able to use the copilot by the end?
No. The two days produce a tested prototype and a pilot plan; deployment then requires internal approval.
What tool is needed?
An AI tool that allows a customised assistant to be created and is authorised by your organisation; its features are checked in the official documentation during scoping.
Can client documents be included?
Not during the session. In a real-world setting, only documents authorised under your confidentiality rules may be used.
Are technical skills required?
Not for configuring an assistant; someone from IT can join the section on the tool and permissions.
Last updated: