AI sales forecasting agent Status: Live

Sales forecasting software: how Oliv AI's Forecaster builds the number before reps log in

How Oliv AI's Forecaster agent drafts each rep's forecast, suggests category moves with a reason on each deal, rolls the number up through every manager, and how it compares with other sales forecasting software.

| For sales leaders, sales managers, revenue operations and quota-carrying reps | Verified 22 Oct 2026 | 24 min read

If you own the forecast number, sales forecasting software is only useful when reps can submit quickly and the number holds up, with the reasons behind every call. Oliv AI's Forecaster agent automates the forecasting process, from collecting deal data to analyzing each deal to the manager's sign-off. It reads your pipeline and builds the forecast before anyone logs in, so reps review and submit instead of typing. It gets more accurate over time by learning from how deals actually close at your company.

At a glance

What the Forecaster agent is

Oliv AI's Forecaster agent analyzes every deal in the pipeline against historical performance, company patterns and deal progression signals. It doesn't move deals or change stages itself. It recommends and nudges, and people make the call.

Three views of the same forecast

• Submissions. Each person's number per measurement and period, with Deals Sum, Last Week and Roll-Up beside it. My Team shows how many rows each person has submitted, and the header shows the Due Date.
• Boards. The rolled-up forecast by Rep/Team: Target Attainment, Pipeline Coverage, Commit Forecast, Most Likely Forecast and Best Case. Pick the board, the granularity (Quarterly, Monthly or Weekly), the fiscal year, the period and the snapshot.
• The submission drawer. One person's commit, built as Commit plus Adjustment, with comments and what moved since the last snapshot. Its deal tabs are Committed Deals, Closed Won Deals, Most Likely Deals, Best Case Deals and Pipeline Deals.

A company can keep several boards, such as Sales Forecast, Renewals Forecast and a regional new business board.

The signals it raises

Suggested category moves appear in the drawer's deal tabs, grouped by direction, such as Move down to Most Likely. Each group shows a deal count and a total, and each deal carries a red flag with its risk reason. One click on the target button moves the deal, and X dismisses the suggestion. That feedback trains the agent.

Why the Forecaster agent matters

What it's for

Oliv AI's Forecaster agent is sales forecasting software for teams that want a number they can stand behind on Monday. It takes over the building of the forecast, so people spend their time judging it.

How it helps

• Reps stop typing updates. They open a forecast that is already filled in, check it and submit.
• Managers review changes, not everything. Risky commits, sandbagged upside and dead deals are flagged before the call.
• Leaders see decisions. The rollup shows what slipped, what moved and why.
• Accuracy compounds. Every closed deal makes the next forecast better.

What happens without it

Without it, the forecast is only as good as each rep's mood that week. Leadership sees a number with no trail behind it: no signals, no reason for the change since last week. A slipping deal shows up when the quarter is already lost, not while there is still time to act.

Why buyers care

VPs of Sales, CROs and RevOps leads buy it. The VP of Sales champions it because the weekly rollup burden goes away. RevOps backs it because clean data feeds the model. It pays off most for teams with five or more reps and at least one manager layer.

The Oliv AI approach to the Forecaster agent

It's the last week of the quarter, and a VP of Sales has her commit call with the CRO on Friday. Five managers report to her, each carrying a team number. The CRO wants one figure she can defend, plus the short list of deals it depends on.

The usual run-up is a round of rep updates, manager check-ins and a rollup sheet that comes together late on Thursday. Across a team, that routine can swallow a dozen hours in a week. Even then, the number tends to lean on how hopeful each rep felt.

Yet most of what the commit needs already sits in the pipeline and the calls. Close dates slip, stages stall and deals go quiet, and each of those leaves a trail. The hard part is reading that trail across every deal, every week, without waiting on people.

Oliv AI does that reading with its Forecaster agent. It drafts each rep's forecast before anyone logs in and checks it before the call starts. It flags risky commits and dead deals, and it surfaces sandbagged upside. Managers review only what changed, and the VP gets one rolled-up view with the risk already called out.

Each board keeps weekly snapshots and an end-of-quarter snapshot, so any commit can be reopened as it stood and checked against what actually closed. The comparison further down shows how that learning loop lines up against other forecasting tools.

Set up Forecaster

1. Connect your CRM. Salesforce, HubSpot or Zoho. Forecaster reads stage, amount, close date and owner for every deal. See the supported integrations.
2. Load your roster. Add reps and managers with the team hierarchy, so forecasts roll up to the right person at every level.
3. Set targets. In Settings › Targets, Create Target adds a target with its type, such as Revenue, and its users. Active Targets and Completed Targets show each target's overall progress.
4. Check your measurements. Settings › Measurements holds the numbers the forecast is built from, such as Commit, Most Likely, Best Case and Pipeline Coverage. Each is Basic (a sum or count) or Formula.
5. Organize your boards. Under Settings › Organisational Settings › Monitoring › Boards, boards sit in collections such as Forecasting and Regional. Each board has a description, a target and its users.
6. Add a cadence. Organisational Settings › Agents › Forecaster lists your cadences by Name, Board and Interval. Add opens the New Cadence wizard, described below.
7. Plan the weeks. Monitoring › Cadence shows a grid of weeks against measurements, on a Quarterly or Monthly period, so you can tick which numbers are due in which week.

Create a cadence

The New Cadence wizard has three steps: Overview, Boards & Measurements and People. On Overview, give the cadence a Title and a Description, then choose its Period: Weekly, Monthly or Quarterly. With a monthly period, set the Rep Submission Day and the Manager Submission Day by week and day.

Most teams are live within two weeks. Reps need no training: they see a pre-drafted view where they'd normally submit by hand.

How a sales forecast gets built

Before you defend a number on the commit call, work out which reading of the pipeline it rests on. Most forecasts lean on one of three, and each one misses something different. The guide to AI sales forecasting software covers what to weigh across tools.

Reps' own calls, rolled up: what each person believes

Each rep sorts deals into categories and calls a number. The manager adjusts it, and the total moves up to the next level. HubSpot's forecast category property, for example, offers Not forecasted, Pipeline, Best case, Commit and Closed won.

A rep's call catches what only the seller knows, such as a buyer's verbal yes or a budget that moves next month. It misses optimism and sandbagging, and it goes stale as soon as a rep stops updating. Gartner's June 2026 abstract on pipeline metrics advises against doubling down on seller inputs and points to pipeline measures instead.

Stage-weighted pipeline: what the stages imply

Here, each stage carries a probability of closing. HubSpot's weighted amount is the deal amount multiplied by that probability, and you add the results up.

The result is a total with nobody's opinion in it, so it makes a useful check on a rep's call. It misses whatever stages can't see. A late-stage deal that went quiet three weeks ago keeps its full weight. Gartner argued in 2019 that setting close probability from stage completion no longer applies, because buyers revisit steps instead of moving in a line.

Deal activity: what is actually happening

This method reads the trail each deal leaves: calls held, emails answered, close dates moved and weeks spent in one stage. It catches a commit that looks fine on paper but has stalled, and it can find upside that a cautious rep left out.

It misses anything that never reached a call or an email, such as a decision made in the buyer's budget meeting. It also needs a history of closed deals before you can say what a pattern means at your company.

Why sales forecasts miss

After a miss, work out which of these four happened before you change the process:

• The call was hopeful. A deal stayed in Commit with no recent evidence from the buyer.
• The data was stale. Close dates and amounts were weeks old when the number was built.
• The stage said more than the deal did. The weighted number counted a deal that had stopped moving.
• The change came late. A deal slipped after the last check, and nobody saw it until the next call.

Which of these Forecaster does

Oliv AI's Forecaster agent combines rep calls with deal activity. The rep's call stays: reps review a pre-drafted forecast, adjust it and submit, then managers add their own adjustments and the number rolls up. The agent also checks every deal against your history and its progress so far. It generates its recommendations from what's happening in calls and emails, and its value adjustment signal draws on historical close rates and deal trajectory.

Run the weekly forecast

For reps

Open Apps › Forecast › Submissions. See your target for the period, then review the forecast computed from your pipeline in the Submission column, next to Deals Sum and Last Week. Check Forecaster's recommended signals, adjust anything you disagree with, and submit with a comment explaining any changes. It takes about 5 to 10 minutes.

For managers

In Submissions, My Team shows each person's progress, such as 9/9 when every row is in, with the due date at the top. Open a person's submission drawer to review their commit: Commit plus Adjustment equals Commit Forecast. Add your own adjustment and a comment, then work through the suggested moves in the deal tabs. For a team of 5 to 10, that's about 15 to 30 minutes in total.

The drawer's comments include Forecaster's own note on how it built the number, so you can see what it changed before you add yours. Deals in this Snapshot then reconciles the start and end commit: Total Decrease (deals that left the category or pushed out) and Total Increase (new deals and deals moved in).

On the Boards tab, expand a manager's row to see each rep, or click the manager's name to drill into that team. Teams that run evidence-based forecast commits every Monday start from this rolled-up view. The full meeting routine is in Run a pipeline review.

Several cadences at once

Monthly, quarterly and pipeline forecasts can run in parallel, each with its own targets and rollup views. One Submissions table can hold a quarter, the next quarter, two months and the current week at once. If you forecast quarterly but want monthly visibility, Forecaster can show a monthly equivalent of the quarterly number. Weekly manager calls, monthly reviews and quarterly board forecasts all run from the same signals, targets, rollups and snapshots in Oliv Forecast.

Snapshots and how it learns

Every submission creates a timestamped record. Each board keeps the Current snapshot, weekly snapshots with the week they cover, and end-of-quarter snapshots such as End of Q3. Pick one from Snapshot to see the board as it stood that day.

In the drawer, a Timeline lists each weekly submission, marked as scheduled, submitted by Oliv or submitted by the manager. View as of reopens a past week, and a Commit Forecast chart plots the submission against the actual.

When deals close or push, Forecaster compares the forecast value to the actual outcome. That reconciliation trains the agent. Over time it learns company-specific patterns, such as a deal type that always needs an RFP or a rep who runs two weeks optimistic on timelines. Feedback on its signals trains it too. Read more on improving forecast accuracy with AI.

Operating modes (Beta)

Forecaster's operating modes (Beta) let it take over more of the process as it learns your patterns.

• AI-Assisted (Beta). Forecaster recommends and your team decides. Managers see where the agent and the rep disagree.
• AI-Managed (Beta). Forecaster generates the full forecast, and your team reviews and adjusts. Reps are largely hands-free.
• AI-Owned (Beta). Forecaster produces the forecast, and managers review only flagged areas of uncertainty. No rep involvement is needed.

Teams typically move through all three operating modes (Beta) within 6 to 18 months. Managers stay in the loop in every mode.

Where results go

Every Monday, Forecaster's Weekly Forecast report lands in Reports › My Reports. Its Current Standing compares your team's forecast with the AI forecast and shows the gap. Then come Deals likely to slip, Deals to pull in and Deals closed last week, each deal with a Review Deal button. Present shows the report in the meeting.

Forecaster also posts to the Agent Feed on Home, for example when deals shifted forecast categories overnight. Forecast submissions, snapshots, signals and adjustments are kept in Oliv AI. If configured, Forecaster syncs the forecast category and adjusted values back to your CRM. Managers get a weekly team forecast summary with flagged deals in Slack or email.

How to measure sales forecast accuracy

Say your team committed 600,000 USD for the month and 550,000 USD closed. Before you call that close enough, you need to know how far off the call was and which way it leaned. Percentage error and bias answer those two questions, and Oliv AI's Forecaster agent records what you need for both. Score the number as it stood before the period ended, never a version someone edited later.

Percentage error: how far off the call was

Take the commit minus what closed, drop the sign and divide by what closed. Average that across periods when you want one figure. Because the result is a percentage, you can compare a rep carrying 100,000 USD with a team carrying 1,000,000 USD.

Watch for two weak spots. The measure turns infinite or undefined when the actual is zero, and extreme when it's close to zero, so a rep who closed almost nothing can swamp an average. It can also flatter a team, because one rep's overcall cancels another's undercall in the total. Score each rep as well as the team.

Bias: which way the call leans

Take the commit minus what closed for each period, keep the sign and average the results. A positive result means you overcall, and a negative one means you undercall. A forecasting textbook puts the idea plainly: when the leftover errors have a mean other than zero, the forecasts are biased.

Bias needs several periods before it tells you anything, which is slow on a long sales cycle. It also hides big misses that point in opposite directions, so read it beside percentage error. And it tells you the direction, not the reason. For that, go back to the deals.

Where Forecaster shows them

Open a person's submission drawer first. The Commit Forecast chart plots the submission against the actual, which gives you both inputs. The Timeline shows each weekly submission and who made it, and View as of reopens a past week, so you score the number as it stood.

On a board, pick an end-of-quarter snapshot from Snapshot to compare reps and managers as the quarter ended. RevOps can run the same comparison across reps to spot systematic bias. The Current Standing view in the Weekly Forecast report adds one more check: the gap between your team's forecast and the AI forecast.

How to build a sales forecast by hand

Run this routine once and you'll see where the hours go, and what you should still check in any tool. It works for a month or a quarter.

1. Set the period and the deadlines. Pick a day for reps and a later day for managers, so reviews land after submissions.
2. Export the open deals. Pull owner, stage, amount, close date and forecast category for every deal in the period.
3. Fix what you can. Chase stale close dates and missing amounts. Flag whatever you can't fix before the deadline.
4. Collect each rep's call. Each rep places deals in Commit, Best Case or Pipeline, then totals each category.
5. Test the commits in the 1:1. For each committed deal, ask what the buyer did last, what happens next and who signs. Move a deal down when those answers are missing.
6. Record your adjustment separately. Write down the rep's commit, your change and your reason. Never overwrite the rep's number, or you can't score either call later.
7. Roll it up. Add up each team's commit and send it to the next level, with the short list of deals it depends on.
8. Freeze a dated copy. Compare it with next week's copy to see what moved.
9. Score it at the end of the period. Compare the frozen commit with what closed, for each rep and for the team.

The routine tends to break in three places. Late updates push step 7 to the night before the call. Spreadsheet copies drift apart, so nobody is sure which one was submitted. And step 9 gets skipped when nobody kept the frozen copy.

The same job with Forecaster

Oliv AI's Forecaster agent takes over the building, and you keep the checking.

Worked example: one manager's commit against what closed

This is an illustrative calculation with sample figures, not an Oliv AI output. A manager with three reps freezes a monthly commit and then scores it once the month closes. The manager cut Rep A's call by 50,000 USD after the 1:1.

The team's 9.1% looks close, but the reps' errors average (25 + 20 + 50) ÷ 3 = 31.7%. Rep B's undercall hides part of what Rep A and Rep C overcalled. The manager's cut also shows up here. Without it, the team's error would have been 100,000 ÷ 550,000 = 18.2%.

Now add the next two months for the same team.

A 7.2% average error sounds fine on its own. The bias says more: the team came in high all three months, by 40,000 USD on average. So in this example, you'd start with Rep C's committed deals and ask what they had in common, instead of asking the whole team to forecast harder.

How Forecaster works with other agents

CRM Manager

Oliv AI's CRM Manager agent is the data foundation. It pulls from call transcripts and meeting notes to keep stage, close date and amount current, so Forecaster builds from today's deal state. Read why clean CRM data drives predictability.

Deal Driver

Forecaster surfaces deals that don't add up, such as closing-date conflicts. Oliv AI's Deal Driver agent fills in the context: last touchpoint, open next steps and days since last activity. A manager can review the flagged items in minutes instead of reading the whole pipeline.

Oliver

When leadership needs a cross-deal breakdown, such as why one segment is lagging, Oliv AI's Oliver agent runs it. Oliver uses the structured data layer that Forecaster and CRM Manager maintain.

Olivia

Oliv AI's Olivia agent is the front-facing master agent for reps. Forecaster stays in the background, drafting the forecast each rep reviews. The Ask button on the Forecast header opens Olivia beside the board, so you can ask a question without leaving it.

Who uses Forecaster, and how

Sales leaders

VPs of Sales and CROs get one forecast view, split by motion, with risk already called out. Leadership sees what moved, what slipped and where the gap is before the commit call. More in Oliv AI for sales leaders and a CRO's guide to forecast accuracy.

Sales managers

Managers review submissions, add their own adjustments and sign off. They spend time on the deals Forecaster flagged, not the whole board. For the view of a single deal, see the Deal Driver agent docs. Read more about Oliv AI for sales managers.

Revenue operations

RevOps configures cadences, targets and metrics during onboarding and watches forecast against actual to spot systematic bias. Learn how Oliv AI works for revenue operations teams.

Account executives

Reps open a pre-filled forecast, adjust a deal or two and submit in minutes. The rest of their week is in Run deals as an account executive.

How Oliv AI compares for sales forecasting software

Most forecasting tools give reps a board to fill in. Forecaster drafts the rep's forecast, suggests category moves with a reason on each deal, and learns from every closed deal, across monthly and quarterly forecasts at once. Oliv AI is the only platform in this table that does all four.

AI sales forecasting features, from each platform's public documentation:

Yes: documented and built in. Partly: documented with a limit, shown in brackets. Not in docs: not found in the vendor's public documentation.

Competitor details checked against each vendor's public documentation on 21 October 2026. Features and plans change, so confirm with the vendor before you decide.

Limits and things to know

• It recommends and doesn't move deals on its own. Forecaster doesn't change deal stages or categories itself. It suggests a move, and a person accepts it.
• It isn't deal scoring. Oliv AI's Deal Insights layer evaluates deal health, and Forecaster consumes those insights.
• It isn't a real-time alert tool. Deal Driver handles alerts on deals as they happen.
• It isn't a coaching tool. Coach handles rep development.
• It's only as good as the data. CRM Manager should be active so deal fields are current before Forecaster analyzes them.
• Operating modes are Beta. AI-Assisted, AI-Managed and AI-Owned are all labeled Beta.

Common questions

Further reading

• Focus on 3 Pipeline Metrics to Improve Your Sales Forecast (Gartner, 3 June 2026): why Gartner points forecasters toward pipeline measures over more seller input.
• Gartner says the traditional linear sales process no longer exists (Gartner, 13 November 2019): why stage completion is a weak guide to the odds of closing.
• HubSpot's default deal properties (HubSpot Knowledge Base): how forecast category and weighted amount are defined.
• Set up and manage object pipelines (HubSpot Knowledge Base): where each stage's probability is set.
• The forecaster's toolbox (Forecasting: Principles and Practice, an open textbook): how forecast errors, percentage errors and bias are worked out.

Related

• Forecaster on oliv.ai: the agent's own page, with its operating modes.
• The Oliv Forecast app: forecast boards, deal inspection and scenario modelling.
• The Monday rollup, retired: how agentic forecasting replaces the weekly scramble.
• AI sales forecasting software compared: what to weigh across tools.
• Agentic forecasting webinar: moving to agent-led forecasting.
• Agent tools and run modes: how approval settings work for agent actions.
• Oliv AI pricing: plans and what each includes.
• Oliv AI product documentation: every documented agent and feature.

Updated 2 November 2026: added how a sales forecast gets built, how to measure sales forecast accuracy, the manual method with a worked example, one more question and further reading.