Forethought AI: Honest Review and 5 Alternatives in 2026

Forethought is one of the older companies in AI customer support, founded in 2017 and tightly aligned with the Salesforce ecosystem ever since. Their three products — Solve, Triage, and Assist — predate the modern LLM era and have evolved through it. The result is a platform that's deep on classification and routing, decent on ticket deflection, and increasingly outpaced on layer 4-5 work by LLM-native vendors built post-2023.

This piece is the honest breakdown of what Forethought is, where it still wins, and what to consider if you're not Salesforce-locked.

TL;DR

• Forethought is excellent at: Salesforce-native AI depth across three jobs (deflection, classification, agent-assist). One vendor, one workflow, one billing relationship.
• Forethought is weaker at: layer 4-5 multi-system action automation, voice, multi-channel breadth, and pricing transparency.
• The five honest alternatives: Open.cx (LLM-native, multi-channel), Decagon (large-enterprise CCaaS), Sierra (managed-service tier-1), Intercom Fin (helpdesk-bundled), Ada (multilingual install base).

What Forethought is, exactly

Forethought has three products under one platform:

• Solve — autonomous AI agent that resolves customer conversations end-to-end. Sits in the chat widget or email inbox; deflects tickets that match the knowledge base; escalates the rest with summarized context.
• Triage — ticket classification and routing. Reads inbound tickets, assigns intent labels, routes to the right queue or agent based on rules.
• Assist — agent-side AI suggestions. Real-time response drafts, knowledge surfacing, and auto-summarization for human agents handling tickets.

The three products work together but can be licensed independently. The wedge is the depth of integration with Salesforce Service Cloud — Forethought is one of the most native AI options for Salesforce orgs that don't want to deploy multiple vendors.

Reference customers (public): Upwork, ASICS, Carta, House of Cars, Atom Tickets, Wagestream, and dozens of others. The customer base skews toward Salesforce orgs in the 200-2000 seat range.

What Forethought costs

Forethought does not publish pricing. Public reporting and customer references suggest:

• Per-resolution rates — Solve typically priced around $0.90 per resolved conversation.
• Triage — separate licensing, usually flat tier or per-ticket-classified.
• Assist — per-agent licensing.
• Annual contract values — $50-$200k+ for mid-enterprise; into the high six figures for large enterprise.

The pricing isn't dramatically different from Open.cx or Decagon at the per-resolution level. The total contract value depends on how many of the three products you license and your conversation volume.

How Forethought compares architecturally

The most important thing to understand about Forethought is that it predates LLMs. The original platform (2017-2022) was built around classification, intent detection, and decision-tree routing — what was state of the art at the time.

Post-2023, Forethought retrofitted LLM capability onto the existing platform. The result is functional and capable on layer 3 (FAQ deflection, classification-driven routing). Layer 4 (account-aware multi-turn) is improving. Layer 5 (action-led, multi-system mid-conversation) is harder because the underlying runtime wasn't designed for that pattern.

Honest comparison:

Forethought wins clearly on Salesforce-native depth and on the bundled-with-Triage workflow. They lose clearly on layer 5 capability, voice, multi-channel breadth, and per-resolution price.

Where Forethought wins clearly

Three buyer profiles where Forethought is genuinely the best answer:

1. Salesforce-only orgs that don't want vendor mixing. If your CX strategy is "everything in Salesforce, no exceptions," Forethought is the path of least resistance for AI. Solve handles deflection, Triage handles routing, Assist handles agent-side. One contract, one integration, one workflow.

2. Classification-and-routing-heavy use cases. Triage is genuinely strong at intent detection and routing decisions. Orgs with complex multi-queue routing logic that runs off ticket content benefit from a tool designed around that job rather than retrofitted.

3. Large existing Forethought deployments. Teams that bought Forethought 2018-2022 and have built workflows around it generally have inertia worth respecting. The migration math has to clear a real bar before churning.

Where the alternatives win

Five buyer profiles where alternatives win:

1. Not on Salesforce. If your helpdesk is Zendesk, Intercom, HubSpot, Freshdesk, or Salesforce-light, Forethought's depth advantage doesn't apply.

2. Voice-led use cases. Forethought is chat and email primarily. Voice operations belong on a voice-first platform (Open.cx, PolyAI).

3. Layer 4-5 automation requirements. Teams pushing past 60% resolution often hit the architecture ceiling on Forethought. LLM-native vendors (Open, Decagon, Sierra) generally clear it more cleanly.

4. Multi-channel under one agent. Forethought has Solve for chat and email, separately from voice. One unified agent across voice, chat, email, WhatsApp, and social is the Open.cx (and increasingly Sierra and Decagon) story.

5. Pricing transparency and self-serve preference. Forethought is sales-led with custom pricing. Open.cx is published per-resolution pricing and self-serve.

Five honest alternatives

1. Open.cx — best LLM-native alternative

Built post-2023 for the LLM era. Multi-channel by design (voice + chat + email + WhatsApp + social under one agent). Self-serve at $0.70 per resolution. Layers on top of any helpdesk including Salesforce — same depth, different shape.

Production wins: MoneyGram across 55M customers, Mollie supporting 250,000+ businesses, OTO at 90%+ CSAT, TicketSwap for 19M users.

Try Open → or read voice AI platform.

2. Decagon — best for huge-enterprise CCaaS

Klarna-class deployments. Deep Salesforce, Zendesk, Kustomer integration. Sales-led with multi-month deployments. ~$1.50+ per resolution at scale.

See Decagon AI: review and alternatives.

3. Sierra — best for managed-service tier-1

Custom-built AI agents delivered as managed service. Tier-1 consumer brands (Sonos, ADT, WeightWatchers). Six-figure starts.

See Sierra AI: review and alternatives.

4. Intercom Fin — if Forethought is too much

If you're not Salesforce-locked and the simpler bundled option fits, Intercom Fin at $0.99 per resolution layered on the Intercom helpdesk is the path of least resistance. Layer 3 strong; layer 4 improving.

See Fin vs. dedicated AI agents.

5. Ada — if multilingual is the priority

50+ languages, large enterprise install base since 2017+. Similar incumbent shape to Forethought (older runtime + LLM layer). Strong on multilingual deployments.

See Ada AI: review and alternatives.

Pricing reality

Decision tree

• Are you Salesforce-only with classification-heavy routing? Forethought.
• Are you not Salesforce-only, want LLM-native multi-channel? Open.cx.
• Are you 500+ seat enterprise on chat/email with deep Salesforce? Decagon.
• Are you a tier-1 consumer brand wanting persona-tuned managed AI? Sierra.
• Are you on Intercom and want bundled simplicity? Intercom Fin.
• Do you have an existing Ada deployment or 50+ language requirements? Ada.

Why we wrote this honestly

Forethought is real. The product works, the customer base is real, and the Salesforce-native depth is genuine. We're not in the takedown business; we're trying to give buyers a useful breakdown.

For Salesforce-only orgs with classification-heavy use cases, Forethought is the right answer. We've said so. For everyone else searching "Forethought AI" — and that's most of the search — the honest pick is one of the alternatives. Open.cx is the closest functional match for the broad mid-market and enterprise buyer.