Forethought and LayBuild solve the same first problem, answering customer questions with AI, from opposite ends. Forethought is an AI layer for support teams that already run a large helpdesk: it plugs into tools like Zendesk, Salesforce and ServiceNow and adds agents for answering, triage, QA and knowledge-gap analysis, priced by quote. LayBuild is a standalone AI agent for a website widget, WhatsApp and a REST API, with published flat prices in INR and far fewer integrations.
If you have a big helpdesk and want AI inside it across email, voice and Slack, Forethought is built for that and LayBuild is not. If you have a small team and want a grounded bot at a known price, the comparison tilts the other way. Forethought facts below come from Forethought's pricing page and integrations page, checked September 2026.
A note on ownership
Zendesk's pricing page (checked September 2026) now describes "Forethought AI agents by Zendesk" and says they "work on any platform". So Forethought's agents are part of Zendesk's AI offering while still marketed for other helpdesks. If you are evaluating Forethought on a non-Zendesk stack, ask how that affects their roadmap for your platform.
What Forethought offers
Forethought sells a set of agents. From its pricing and navigation pages: Solve (an omnichannel agent that resolves issues), Triage (ticket classification and routing), Discover (surfaces knowledge and workflow gaps), a QA agent that scores human agents' tickets, and an agent-assist copilot.
Plans are Team, Professional and Enterprise, and none has a published price: every plan says "Get a Quote". Forethought describes the model as "a blend of platform access fees and an outcome-based pricing cost", with optional add-ons, and says overage charges may apply beyond purchased usage. Instead of a free trial it runs a proof of value with your data.
The plan tiers differ mainly by channel and depth. Team includes AI agents for chat and mobile, Autoflows, custom actions, CSAT collection and ready-made triage models. Professional adds agents for email, voice and Slack, AI-surfaced insights, multi-brand management and custom handoff models. Enterprise adds the Solve API, knowledge gap detection with AI article creation, and an Analytics API.
The integrations page is long. Helpdesks and CRMs listed include Zendesk, Salesforce, ServiceNow, Freshworks, Freshchat, Intercom, Kustomer, Gladly, Gorgias, Help Scout, HubSpot, Front, Talkdesk, Jira and more. Knowledge sources include Confluence, Notion, Guru, Contentful, Document360, Glean, ReadMe and others.
What LayBuild offers
LayBuild is one product: an AI agent that answers from the files, URLs and Q&A pairs you give it, on a web widget, WhatsApp and our REST API. Human agents take over in our dashboard. It publishes its prices: Starter ₹1,499/month, Pro ₹3,999/month and Premium ₹9,999/month (₹1,199, ₹3,199 and ₹7,999 billed annually), capped at 1,500, 8,000 and 20,000 conversations a month. There is no per-resolution fee. Model token costs depend on the LLM provider you configure.
Side by side
| Forethought | LayBuild | |
|---|---|---|
| Role | AI layer on top of an existing helpdesk | Standalone AI agent with its own dashboard |
| Pricing | By quote: platform fee plus outcome-based cost | Published flat plans in INR with conversation caps |
| Channels | Chat and mobile (Team); email, voice, Slack (Professional); API (Enterprise) | Web widget, WhatsApp, REST API |
| Helpdesk integrations | Zendesk, Salesforce, ServiceNow, Freshworks, Intercom, Kustomer and more | None native; four outbound webhooks and a REST API |
| Knowledge sources | Integrations with Confluence, Notion, Guru and others | File upload, listed URLs, manual Q&A pairs, pinned docs |
| Triage and QA | Triage agent, QA agent | Not offered |
| Trial | Proof of value | No free plan or trial |
| Self-hosting | Not listed on the pricing page | Yes, docker-compose from the open-source repo |
Where Forethought is stronger
It works inside the helpdesk you already have. For a team with a large Zendesk or Salesforce deployment, an AI layer built to integrate with that system avoids running a second support tool. LayBuild cannot read or update tickets in any helpdesk.
It covers more channels. Email, voice and Slack are available on Professional. LayBuild has none of those.
It does more than answer. Ticket triage, QA scoring of human agents, and knowledge-gap detection with AI-drafted articles are separate products at Forethought. LayBuild has none of them. Our learning loop is the closest thing, and it is cruder: after a qualifying reply, we save the exchange as a knowledge doc without human approval, and retract it only if the customer rates the conversation 2 stars or lower.
Its integrations list covers the tools where docs often live, such as Confluence, Notion and Guru. LayBuild has no Notion, Confluence, Google Docs, GitHub or Zendesk sync; you upload files or list URLs, and we fetch only those URLs, with no scheduled re-sync.
Where LayBuild is a better fit
You can see the price before talking to sales. For a small team, "Get a Quote" on every tier plus an outcome-based component is a lot of process for a first bot.
You can run it yourself. The whole stack (Postgres, Redis, Qdrant, API, web, nginx) runs from a docker-compose file, with one LLM key and a database URL.
You pick the model. Hugging Face is the default (meta-llama/Llama-3.1-8B-Instruct), with OpenAI, Gemini, DeepSeek, Qwen, Cloudflare Workers AI, OpenRouter, NVIDIA and any OpenAI-compatible endpoint available, and failover between providers.
The grounding rules are simple and documented. Strict knowledge-base mode is always on. No retrieved sources means no LLM call and a fixed reply. An answer whose stemmed words overlap less than 25% with the retrieved sources is replaced by the same fixed reply. That lexical check can block a correct paraphrase and pass a wrong answer that reuses source words, so treat it as a filter, not proof.
What LayBuild does not do
- No ticketing system, no helpdesk integrations, no email, Slack, SMS or voice channel.
- No triage, QA scoring or sentiment detection, and no AI summary on handoff.
- API tools are not LLM function calling. After the answer is generated, attached HTTP tools run with their configured defaults and the first successful result is returned.
- Retrieval is English-first: the default embeddings and full-text search are English.
- Webhooks are logged but not retried.
- No SSO or SAML, and no SOC 2, HIPAA, ISO 27001 or GDPR certification.
How to choose
Choose Forethought if you run a large support organisation on Zendesk, Salesforce or ServiceNow, need AI across email, voice and Slack, and want triage and QA as well as answers. Budget time for a sales process and a proof of value.
Choose LayBuild if you want a bot on your website or WhatsApp, answering from your own docs, with a published INR price and the option to self-host, and you can live without helpdesk integration.
Next steps
See the plan limits on pricing. If you are weighing outcome-based pricing, deflection vs resolution pricing shows how to model it, and self-hosted AI support covers what running LayBuild yourself involves.