Overview

Every customer on one record.

14 features across 4 domains. AI agents on top of all of them.

See all

Sales & CRM

Marketing

Campaign Management

Finance & Accounting

IT plumbing

AI Agents

Start free

Forecast Agent

Flags Commit numbers your pipeline can't support.

Watches the rep-by-rep forecast submission against historical close rates, deal-velocity, and current pipeline shape. Flags Commit numbers that diverge from what the underlying pipeline can support, surfaces gaps to managers before quarter-end, not after.

Start free

No credit card Live in 60 seconds Off-switch per agent

Saves

7 days of forecast wrangling

Touches

ForecastPipelineCommit

01What it actually does

No demo magic. Real actions, real audit trail.

Each action runs through the same engine your humans use: same routing rules, same approval policy, same audit log. The agent's name is on every record it touches.

Computes a baseline per rep

Per Rep × Period: historical close rate per Stage, average deal-cycle, pipeline-coverage ratio. The baseline is what the agent thinks the rep should commit.

Compares against the rep's submission

On rep submit, computes the gap: Submitted Commit − Baseline. Flags above-baseline (sandbagging risk: pipeline supports more) and below-baseline (over-call risk: pipeline can't support).

Drafts the manager view

Manager sees the team roll-up with the agent's per-rep flags inline. Drill any rep to see which opps make up the Commit and where the agent disagrees.

Warns the CRO on roll-up

Before final consolidated submission, agent surfaces: 'Org-wide Commit is $X. Historical accuracy at this stage suggests landing $X − Y to X + Z. The N reps below contribute most of the variance.'

02How it decides

The logic behind every action.

Not a black box. The decision rules are explicit, configurable, and surfaced on every action so the manager (and you) can trace the why.

Coverage ratio per stage

Stage 1 (Discovery) typically needs 3-5x coverage to support a closed dollar. Stage 4 (Negotiation) needs 1.2-1.5x. The agent uses your historical ratios, not industry defaults.

Velocity + remaining time

If a Commit-bucket opp's average days-in-stage exceeds days-to-quarter-end, the agent flags it as unrealistic regardless of rep confidence.

Trust the rep on small samples

If a rep has fewer than 8 historical closes, the baseline isn't statistically meaningful. Agent flags 'low-confidence baseline' and defers to the rep's submission with a soft note.

03What you control

Off-switch per agent. Always.

Every agent runs on a leash you set. Approval thresholds, dry-run mode, per-rep / per-account disable, audit trail, your call.

Block vs warn

Default: warn only, never blocks a submission. Toggle to block-above-policy for periods when forecast accuracy matters more (board quarters, financing rounds).

Per-rep override

Mark a rep's forecast as 'manager-trusted' to suppress agent flags (e.g. CRO override for senior strategic rep). Audit-logged.

Read-only mode

Computes flags, surfaces nothing. Use to evaluate the agent against a closed period before turning on for next.

04This morning

A real activity log.

Excerpted from a production tenant. Every line is signed, timestamped, and traceable to the action it represents.

Forecast Agent · activity log · today

MON 09:00

Forecast submission opened for FY26 Q2. Computed baselines for 47 reps.

TUE 16:14

23 reps submitted. 4 flagged: 2 above baseline (sandbagging risk), 2 below baseline (over-call). Manager queue updated.

WED 11:02

Sales-Manager-East rolled up team. Agent surfaced: 'Team Commit $4.2M. Pipeline coverage 1.8x; historical avg at this stage 2.4x. Suggests landing $3.6M-$4.0M.'

THU 18:45

CRO consolidated submission: $18.4M. Agent flag: 'High-confidence range $16.8M-$18.1M. Variance from 3 of 6 regions (named).'

05Built on

The skills this agent calls.

Every agent is composed from the underlying CRM skills, the same ones your humans use. No private data path, no shadow record.

Pipeline & forecasting Reports & dashboards

06FAQ

Common questions.

Asked by the people whose name goes on the renewal forecast.

Does the agent change the rep's number?

No. The agent annotates and flags. Reps and managers always own the submitted number. The agent's job is to make the disagreement visible and citable, not to overrule.

What if our forecast accuracy data is sparse?

The agent reports its confidence inline. With < 4 historical periods, it defers to manager judgement and surfaces only the most extreme flags (>3σ from coverage norms). Confidence improves quarter over quarter as the dataset grows.

Can we customize which categories the agent watches?

Yes. By default it watches Commit (the high-stakes number). Toggle BestCase or Pipeline coverage warnings on for periods where you want broader scrutiny.

07The other 7 agents