Forethought Review

What Forethought actually does

Forethought is a layered AI platform for customer support. The product breaks into four pieces: Solve (a customer-facing AI agent that resolves tickets across channels), Triage (an internal AI that classifies and routes incoming tickets), Assist (a side-panel copilot that drafts responses for human agents), and Discover (analytics on which topics drive volume and which articles are missing from the knowledge base).

The pitch is that AI in support isn't one product — it's an end-to-end pipeline from "ticket arrives" to "ticket resolved" with AI at every hop. That framing is more honest than competitors who claim a single chatbot solves the problem, but it also means evaluation is more complex.

What works well

Triage is the standout for teams not ready to put AI in front of customers. Routing every inbound ticket to the right team, with the right priority, with the right tags pre-applied, is mechanically simple but enormously time-consuming when done by humans. Forethought's classifier quality on well-trained categories is consistently high, and the integration into Zendesk, Salesforce Service Cloud, and other major helpdesks is mature.

Assist (the agent copilot) is the feature most underrated by buyers. Human agents see a side panel with suggested responses, relevant macros, and links to similar resolved tickets. Time-to-first-response drops measurably (usually 20–30%) without changing customer-facing behavior.

The analytics layer (Discover) reveals patterns most teams don't see. Common pattern: "23% of your tickets this month are about refund eligibility, and your refund policy article hasn't been updated in two years." Pointing teams at the highest-leverage knowledge base improvements is worth the subscription on its own.

Where it falls short

Pricing is opaque and skewed toward enterprise budgets. There is no public pricing — every deal is custom — and the entry point sits well above what SMB support teams can absorb. Below 5,000 tickets per month, the math rarely works.

Implementation requires real work. Unlike Fin, which inherits everything from your Intercom setup, Forethought needs a dedicated implementation phase: training the classifiers, setting up the routing rules, validating the agent responses against historical tickets. Plan for 4–8 weeks before the AI is reliably contributing.

The customer-facing Solve agent is competent but not best-in-class. For pure deflection use cases against a well-organized help center, Intercom Fin's setup is faster and the resolution quality is comparable.

Who should use it

Forethought is the right pick for mid-market and enterprise support operations running multi-channel volume on Zendesk or Salesforce, where the value comes from triage and copilot use cases as much as customer-facing deflection. Teams that already invested in clean ticket taxonomy and a maintained knowledge base get the fastest value.

Smaller teams (under 1,000 tickets per month) overpay and underuse it. The implementation overhead alone consumes more cycles than the time savings deliver.

Pricing notes

Custom enterprise pricing. Expect a six-figure annual contract minimum for any meaningful deployment. Negotiate carefully on which products are included — bundles vary deal to deal.

Where automation tops out

Forethought handles the support-internal workflow well: triage, routing, drafting, deflection. What it doesn't do is reach across systems to actually solve the customer's problem when the answer requires action — checking inventory, processing a return, modifying a subscription, escalating to a specialist team with the right context. Those cross-system transactions are the next frontier, and they require an agent layer with access to your billing, fulfillment, and product systems — not just your helpdesk.