Revenue analyst agent for a mid-size education and training company on HubSpot that forecasts each intake against cohort capacity and replaces the Monday pipeline spreadsheet

What The Intake Forecaster does

Every Monday, leadership at this 51 to 200 person training company asks the same questions: how much pipeline do we have, what closes before the next term starts, and are we going to fill the cohorts we have already staffed. Today someone on the ops or sales team exports HubSpot deals, pulls registration counts, checks the cohort calendar and rebuilds a spreadsheet by hand, and the numbers shift depending on who built it that week. The Intake Forecaster owns that job for the founder and leadership team. It answers ad hoc pipeline and enrollment questions in plain English, ties each open deal to the intake it would land in (fall, spring, summer or a rolling cohort start), compares weighted pipeline with available seats and instructor capacity, and publishes a reviewed Monday digest. It works only with aggregated, commercial-level data, keeps student records out of every answer, and is maintained by the build team so definitions, connectors and term calendars stay current as the business changes.

Built on HubSpot

HubSpot holds the pipeline, so the plan starts there: Breeze Assistant is enough for HubSpot-only questions like deal counts by stage, and HubSpot reports can keep powering rep-level views. The Monday question leadership actually asks spans HubSpot, the registration system, the cohort calendar, the LMS and billing, and needs FERPA-safe aggregation and suppression, which Breeze alone cannot enforce across systems. A custom AI build reading read-only MCP connectors (with n8n scheduling the Sunday run and handling the approval step) joins those sources under one governed metric dictionary. Because the team wants it built and maintained for them, the connectors, definitions and term calendars are owned and kept current by the build team rather than an internal analyst.

What starts it

• Scheduled Sunday evening run that assembles the Monday executive digest for review before the leadership meeting
• A leader asks a question in the team chat channel or the HubSpot sidebar, such as 'what institutional pipeline closes before the spring start'
• A new term or cohort is added to the cohort schedule, which creates a new intake bucket the agent forecasts against
• An intake crosses a risk threshold leadership has set (for example, weighted pipeline below the seats still open with fewer than a set number of weeks to start)
• Month end or fiscal year end for institutional buyers, which triggers a budget-cycle view of institutional deals at risk of slipping a year

What it reads

• HubSpot deals across separate pipelines for institutional contracts (schools, districts, employers) and individual enrollments (parent-paid and self-paid learners), plus deal properties for target intake, seats and program
• Registration or student information system, exposed through an MCP connector to an aggregated view only (confirmed enrollments per program and intake, no names or student IDs)
• Cohort schedule and capacity tracker in Airtable (term start dates, seats per cohort, assigned instructors, add/drop deadlines)
• LearnWorlds enrollment and activation counts at the program and cohort level for online and self-paced offerings
• Billing or accounting system for invoiced, paid and deposit-received amounts, used to separate signed from paid
• A governed metric dictionary (pipeline, weighted pipeline, booked seats, confirmed enrollment, fill rate) stored as a versioned document the agent reads before answering

When a person takes over

• The RevOps or ops owner reviews and approves the Monday digest before it goes to leadership, because a wrong fill forecast can trigger hiring or cancelling a cohort.
• Any request that would require learner-level data (a named student, a specific family's payment status, a roster) is refused and routed to the registrar or compliance owner, who handles it under the institution's FERPA process outside the agent.
• When HubSpot seat counts conflict with signed contract terms or billing (for example, a contract for 120 seats invoiced for 90), the agent flags the deal to the sales leader and finance to resolve rather than picking a number.
• New or changed metric definitions (what counts as a booked seat, how deposits are treated) go to the founder and RevOps owner for sign-off before the agent uses them.

Rules it follows

• FERPA: the agent only reads aggregated views of the registration system and LMS, never names, student IDs, grades or contact details of enrolled learners, and suppresses any count below the agreed small-cell threshold in answers, the digest and logs.
• Audiences are never blended: institutional, parent-paid and self-paid revenue and pipeline are always reported separately, and parent and student contacts are never used for ad hoc analysis beyond aggregated pipeline counts.
• Every number cites its definition from the metric dictionary and the source systems; if the dictionary does not cover a question, the agent says it cannot answer reliably and escalates.
• All connectors are read-only. The agent cannot change deals, enrollments, cohort capacity or invoices, and cannot email anyone outside the approved leadership channel.
• Term-aware sanity checks run before publishing: deals mapped to past intakes, intakes with zero capacity, or a weekly pipeline swing beyond a set percentage hold the digest for human review.

How you know it works

• Hours per week spent preparing the Monday pipeline numbers (target: review time only)
• Share of Monday digests published on time without manual rework
• Percentage of open deals with a valid intake and seat count mapped
• Forecast accuracy: projected versus actual confirmed enrollment per intake at the add/drop deadline
• Number of leadership pipeline questions answered by the agent without an analyst rebuilding data

From pilot to live

1. Pilot (one intake, HubSpot plus capacity)Agree the metric dictionary, add or clean the intake and seat properties in HubSpot, connect Airtable, and run the digest for the next term start in parallel with the existing spreadsheet for several weeks. Differences are reviewed line by line until the RevOps owner trusts the numbers.
2. Expand (registration, LMS, billing, ad hoc Q&A)Add the aggregated registration view, LearnWorlds counts and billing through MCP connectors after the compliance owner approves the fields and suppression threshold. Open the question channel to the leadership team and retire the manual spreadsheet.
3. Operate and maintainCover every intake and both audiences, add fiscal-year risk views for institutional buyers, and run ongoing maintenance: connector monitoring, schema change fixes, new term calendars loaded each cycle, and a quarterly review of definitions and forecast accuracy with leadership.