Map case study · Prestant
Five agents defined before a single one was built.
Prestant has 45 real-estate professionals across several French markets. The first engagement was an audit: observe the work, quantify recurring tasks and define where an agent could help without taking away the final decision.
The first job was choosing before building
Prestant had identified broad potential for AI, but several roles and tasks could qualify. Building one general assistant would have hidden the differences between market analysis, listing copy, client follow-up and transaction work.
Map therefore followed a mandate end to end, observing tasks, sources, duration and the moments when a professional must keep the final decision.
The audit found that research, listing content, follow-ups and transaction preparation occupied a large part of the week. A bilingual description took two to three hours; a market analysis took three to four.
Before
The team had a broad ambition around AI, but no agreed order, scope or validation rules.
What Basalt built
Basalt mapped a complete mandate cycle, measured repeated work and turned it into five bounded agent specifications.
After
Prestant received a deployment plan: what to build first, which data each agent needs and where a professional must review the output.
Key figures
45
people included in the audit
Organisation size included in the map.
≈60%
of the week mapped to recurring non-selling work
Recurring workload identified during the audit.
5
agent workflows prioritised
A proposed roadmap, not five deployed agents.
02
Five possible agents, five distinct responsibilities
The audit separated market analysis, listing content, client follow-up, transaction preparation and prospecting. Each workflow received its own scope, required data and human review point.
That separation is the result of the engagement. It orders a future deployment without implying that an agent is already in production.
Select a step
How the workflow works
Observe
Follow the work from a new mandate through pricing, launch, follow-up and transaction.
Measure
Record the repeated time, including two to three hours for a bilingual description and three to four for market analysis.
Plan
Give each proposed agent a scope, data sources, human review point and place in the rollout.
03
Audit figures, not realised gains
Two to three hours for a bilingual description and three to four for market analysis are observed baselines. The roughly 60% figure describes recurring non-selling work mapped in the week.
The five workflows are a prioritised roadmap. No future saving is counted as achieved until Deploy and Measure have taken place.
Workflows identified
- Market analysis
- Listing content
- Client follow-up
- Transaction preparation
- Prospecting
Bring us one workflow that takes too much time.
In 30 minutes, we will see whether it is worth mapping.
Book 30 minutes- Start with a real task
- Look at the modules involved
- Identify the control points
- Choose a clear next step