01 the business
Property management and lettings agency
50 staff, twelve agents
Agents are good at relationships and bad at logging them. The CRM is full of duplicate landlords, blank fields and deals stuck in old stages. Management has stopped using its reports and asks for updates in meetings instead.
02 before
The problem
Salespeople put off CRM updates until Friday, then fill them in from memory or skip them. Records end up with duplicate contacts, blank stages and next steps that are months old. Managers stop trusting the reports and ask for updates in meetings instead.
- Updates happen on Friday afternoons from memory, or not at all.
- The same landlord appears three times under slightly different names.
- Pipeline reports are rebuilt by hand in a spreadsheet before every board meeting.
03 the build
What the AI does
- Step 1: Watches sent and received email in Gmail or Outlook and attaches each thread to the matching contact and deal.
- Step 2: Turns Zoom or Teams call recordings into short notes with agreed next steps and logs them on the deal.
- Step 3: Suggests a deal stage change when the notes show progress, such as a proposal sent or a verbal yes.
- Step 4: Fills in missing fields like job title, company size and phone number from email signatures and the company website.
- Step 5: Finds likely duplicate contacts and companies and queues them for merging with the evidence side by side.
paused: waiting for a person to approve
How it was put together
- Matched every sent and received email to the right contact and deal using addresses and thread history.
- Turned call recordings into short notes with agreed next steps and logged them on the deal.
- Filled blank fields from email signatures and public company pages, with the source stored on the record.
- Built a review list for suggested merges and stage changes, with the evidence side by side.
04 human in the loop
Where a person stays in charge
Logging notes and filling blank fields happens automatically. Stage changes and record merges wait in a review list until the deal owner or sales ops person accepts them.
- Logging and field filling are automatic; merges and stage changes wait for a person.
- Personal email threads can be excluded by label or sender.
- Every automated change is recorded, so it can be traced and undone.
05 rollout
From kickoff to live in 4–5 weeks
- Week 1Audit the CRM, agree field rules and the merge policy with sales operations.
- Weeks 2–3Run in suggestion mode only and measure how often suggestions are accepted.
- Weeks 4–5Turn on automatic logging and filling; keep merges and stages in review.
06 after
What changes
The CRM stays current without asking salespeople to do data entry. Managers can run reports and forecasts from the system again, and pipeline meetings spend their time on decisions rather than status updates.
Assume twelve agents each spending 3 hours a week on CRM updates, and automatic logging removing two of those hours.
12 × 2 hours = 24 hours a week handed back to selling and letting, or roughly 100 hours a month.
What to measure
- Share of deals with activity logged in the last 14 days.
- Blank required fields across open deals.
- Duplicate records remaining.
- Suggestion acceptance rate for stage changes.
07 evidence
What the research says
A survey of 4,050 sales professionals found the average seller spends about 40% of their time actually selling, and 51% of sales leaders using AI say disconnected systems are slowing their AI work.
Salesforce, State of Sales report, 2026 edition (opens in a new tab) · 2026
Gartner research put the average cost of poor data quality at $12.9 million a year per organisation.
Gartner, Data quality: why it matters and how to achieve it (opens in a new tab) · 2020
08 lessons
Mistakes to avoid
- Merging automatically. One bad merge can join two different landlords' histories.
- Logging everything, including personal email. Agree exclusions first.
- Cleaning once and stopping. The value is in keeping it clean every day.
09 questions
Common questions
- Which CRMs does this work with?
- HubSpot, Salesforce and Pipedrive are common. Most CRMs with an API can be connected.
- Will agents notice anything different?
- Mainly that their records are already up to date. They see a short review list for the few changes that need a decision.