Sales Revenue and Performance

Implement AI-assisted forecasting of your revenue

How do you cut the error in your revenue forecast as your business grows?

Works sets up a predictive forecast on your existing sales data. A model calculates the forecast from data on each deal, such as stage and activity. The Revenue Forecaster agent tunes the model and tests it against actual results. Your managers adjust the forecast before they approve it. The agent measures forecast error, the gap between forecast and actual revenue.

Published Updated

What you get
Predictive forecast · October to December 2026
In use in forecast calls
7.5%
average model error on the 2 test quarters
5 of 5
deals at risk reviewed in the forecast call
Forecasting platform Uses HubSpot and Gong data
Deal signals Deal stage plus engagement, activity and deal risk from Gong calls
April to June test, actual $240,000 Model error 10%, current forecast error 25%
July to September test, actual $260,000 Model error 5%, current forecast error 20%
October to December forecast $250,000 in new-contract revenue
Manager overrides Each with a logged reason
Error target 10% or less on the quarter

You plan October to December against the approved forecast.

See it run

A predictive forecast in its first quarter

Quarter in forecast
October to December 2026 · 10% error target
Deals at risk reviewed 2 / 5
Schedule
Every quarter
Forecast call every Monday
Run log
  1. You
    “Set up a predictive forecast for new contracts on our HubSpot and Gong data.”
    Your instruction
  2. Revenue forecast model used Upstream
    Your revenue forecast model comes from Model Revenue Forecasts.
  3. Call data used Upstream
    The data from your recorded sales calls in Gong comes from Set Up Conversation Intelligence.
  4. Sales analytics used Upstream
    Your sales analytics come from Set Up Sales Analytics.
  5. Platform set up on your data Revenue Forecaster
    You chose a forecasting platform that uses your HubSpot and Gong data. The Revenue Forecaster agent chose three deal signals beyond deal stage. They are engagement, activity and deal risk from your recorded sales calls in Gong. You named one owner for changes to the model.
  6. Model tested on two past quarters Revenue Forecaster
    The Revenue Forecaster agent tuned the model on 12 months of your deal history. The agent tested the model against actual revenue from new contracts. Actual revenue was $240,000 in April to June and $260,000 in July to September. Today you add up the rep forecasts by hand. In April to June, the model error was 10% and the current forecast error was 25%. In July to September, the model error was 5% and the current forecast error was 20%. You adopted the model.
  7. You
    Your review of deals at risk needed You
    The Revenue Forecaster agent shows the model's forecast of $250,000 in new-contract revenue for October to December. The agent shows five deals at risk. A forecast call is the meeting where your team reviews the forecast. In Monday's forecast call, you reviewed 2 of 5. You overrode the model on one deal and logged the reason.
How it works

A forecast you can plan against, with people in control

The Revenue Forecaster agent chooses deal signals beyond deal stage, such as engagement, activity and deal risk from your recorded sales calls. You choose the error target. The agent tunes the model on your own deal history. You run the model next to your current forecast before you rely on it. In each forecast call, your managers adjust the model's forecast. A manager who overrides the model logs a reason. The model learns from each logged reason.

What it needs
Your revenue forecast model and sales analytics
Data from conversation intelligence (software that records and analyzes sales calls), and deal history
Access to your CRM
A CRM with 12 months of clean opportunity data
A Clari or Bowtie account
A revenue operations team to own the platform and the forecasting process
A budget of about $30,000 to $100,000 a year
What you get

Forecast error below your current level and earlier warning of deal risk. Less manual roll-up (adding up rep forecasts by hand), and a model forecast that managers trust and control.

Run sequence

Start to finish

You start from your revenue forecast model, your call data and your sales analytics. Then you and the Revenue Forecaster agent get your predictive forecast in three stages.

  1. 1
    Live Forecasting Platform

    Set up the forecasting platform

    You choose a platform for your data
    The Revenue Forecaster agent chooses the deal signals
    You decide who can change the model
    You + Revenue Forecaster
  2. 2
    Tuned Predictive Model

    Tune and test the model

    The Revenue Forecaster agent tunes the model
    The agent tests the model against past results
    You compare the model with your current forecast
    Revenue Forecaster + You
  3. 3
    Forecast Calls and Error Trend

    Use the model in forecast calls

    Your managers see the model's forecast next to theirs
    A manager logs a reason for each override
    The agent measures forecast error each quarter
    You + Revenue Forecaster
Before it starts

What Works asks you

A few questions in plain words. The answers set up the run for your business.

What is the main goal of the predictive forecast: lower forecast error, less manual work, or earlier risk signals?
YouLower forecast error and earlier risk, with less manual roll-up
Which signals should the model weigh beyond deal stage, such as engagement, activity, or conversation risk?
YouEngagement, activity, and conversation risk signals
How do you keep human judgment in the forecast instead of trusting the model without question?
YouThe model proposes a forecast, and managers adjust and commit
What forecast error target do you want to reach, measured over a time span such as a quarter?
YouA tighter error target than today, measured on the quarter
What happens when the model and a manager disagree on a deal or on the forecast number?
YouThe manager overrides with a logged reason that the model learns from
How do you run the model next to your current forecast so you can compare them before you rely on it?
YouRun both side by side and compare before relying on the model
Who runs it

The agents behind this workflow

The Revenue Forecaster agent sets up and tunes the predictive forecast. The same agent shows deals at risk and measures forecast error.

On this workflow

Revenue Forecaster

Sets up, tunes, and measures the predictive forecast, and shows its forecast and deal risks

What it can do
Sets up the predictive forecast on your existing dataTunes the model and tests it against actual resultsShows the model's forecast and deal risksMeasures forecast error and learns from manager overrides
Jeanie, your right hand Works Crew

Jeanie leads a core crew of seven that runs your company day to day. They direct the specialist and workflow agents in each playbook, including the ones on this workflow.

Connect your tools

Bring what you already run

Works connects to the apps you already use.

  • Gong The Revenue Forecaster agent connects to Gong.
  • Google Sheets This workflow uses Google Sheets.
  • HubSpot This workflow uses HubSpot.
  • Salesforce The Revenue Forecaster agent connects to Salesforce.
Other apps you can use
Your CRM

or any CRM you use

Your call recorder
  • Fathom
  • Fireflies
  • Otter.ai
  • Chorus

or any call recorder you use

Your spreadsheet app
  • Microsoft Excel
  • Airtable
  • Smartsheet
  • Zoho Sheet

or any spreadsheet app you use

These are starting points, not a limit. Connect the tools you already use.

Connect your work

The workflows before your predictive forecast

Linked cards are live pages. The others are part of the Sales playbook in Works.

Works · AI business OS

One workflow inside your whole business

In Sales, Implement AI-Assisted Forecasting is part of the Revenue and Performance work.

Prebuilt playbooks, plus your own
StrategyMarketingSalesCustomer SuccessPartnershipsProducts & ServicesOperationsFinancePeopleLegal & Compliance + Build your own
The Sales playbook
64
Workflows
16
Specialist agents
17
Workflow agents
173
Templates
You're here
Implement AI-Assisted Forecasting
One of 64 Sales workflows · run by 1 workflow agent
Templates

What comes with it

Built for this workflow
AI Forecast Signal Inventory XLSX
A sheet where you list the signals for your forecast. It opens in Works after you sign up.
AI Forecasting Rollout Plan DOCX
A document where you plan the rollout of your forecast. It opens in Works after you sign up.
Forecast Reconciliation Model XLSX
A sheet where you compare two forecasts. It opens in Works after you sign up.
Also included here
Forecast Accuracy Tracker XLSX
A sheet where you track the accuracy of your forecast. It opens in Works after you sign up.
Scenario Forecast Model XLSX
A sheet where you build a forecast for each scenario. It opens in Works after you sign up.
Sales Analytics Plan DOCX
A document where you plan your sales analytics. It opens in Works after you sign up.
Call Quality Scorecard XLSX
A sheet where you score the quality of each sales call. It opens in Works after you sign up.

Each template opens inside Works after you sign up.

Questions

What people ask

How do you use AI to cut the error in your revenue forecast?

The Revenue Forecaster agent sets up a predictive forecast on your existing data. The agent tunes the model. The agent then tests the model against actual results. The agent shows the model's forecast and the deals at risk. You choose the error target. Your managers adjust the forecast before they approve the number you plan with.

How is forecast error measured?

Forecast error is the gap between the forecast and actual revenue. You choose the error target and the time span, such as a quarter. The Revenue Forecaster agent measures the error at the end of each period. The Revenue Forecaster agent also learns from the reasons your managers log when they override the model.

Why run the model next to your current forecast?

You compare the two forecasts before you rely on the model. The Revenue Forecaster agent tests the model on recent quarters against actual results. You check whether the model's error is lower than the error of your current forecast. If the error is not lower yet, the agent tunes the model again. You can also start with one pilot team.

What happens when the model and a manager disagree?

You decide what happens when the model and a manager disagree on a deal or on the forecast number. For example, the manager overrides the model with a logged reason. The model then learns from each logged reason. You can also keep the model's forecast unchanged.

Plan against a forecast you can rely on

Sign up to set up a predictive forecast on your existing sales data.

Sign Up