lead ──▶ AI agent ──▶ record in CRM
│
knowledge base
(it answers the usual ones)
At peak, 100+ leads a day were landing on three reps: qualifying by hand, the same questions over and over, a queue for the first reply. We built an AI layer between the ad channels and the sales team. The agent runs the conversation, answers from a knowledge base, gathers the details and hands the rep a structured lead record. Reps pick up the people who are ready to buy.
sorted by hand
notion + crm + chats ──▶ AI manager
│
plan for today ◀──┴──▶ reports
+ weekly summary
The plan for the day was assembled by hand out of Notion, a CRM, kanban boards and chat threads; reports came late or went missing. The AI project manager collects tasks from every tool, builds a plan and sets priorities for each person, chases status through the day, sends a report in the evening and a summary at the end of the week. Missed reports get logged and reach the lead.
summaries build themselves
creator ◀──▶ AI agent ──▶ creator database
│
AI shortlist ◀── site · bot
[ in progress ]
Bloger agency runs one of the largest creator databases on the market. It used to be maintained by 3+ people, and it went stale faster than they could update it: follower counts, reach and rates all drifted out of date. We are building an AI agent that contacts creators itself and refreshes their records in the background, plus an AI shortlist against a brief on the site and in the bot. The project is in active development.
the database by hand