Zendesk Couldn't Build It, So They Bought It

Zendesk announced on March 11, 2026 that they entered a definitive agreement to acquire Forethought, the AI support company handling over one billion monthly interactions for companies like Grammarly, Airtable, and Upwork. The deal is expected to close by end of March. Terms weren't disclosed. The strategy is obvious.

Zendesk has been trying to ship credible AI for years. Their "AI Agents" feature launched late, priced high ($1.50 per resolution on committed plans, $2.00 pay-as-you-go), and never caught up to what Forethought and Intercom were doing with intent detection and automated workflows. So they bought the capability instead.

This is another major acquisition in the AI support space. The pattern is clear: standalone AI support tools are disappearing into larger platforms.

What Forethought Actually Did Well

Forethought wasn't just another chatbot company. They built a multi-agent platform called SupportGPT with separate products for different stages of the support pipeline. Their Solve product automated responses to common questions. Their Triage product classified incoming tickets by intent and priority, then routed them to the right team. Their Assist product gave agents suggested responses and relevant knowledge articles.

That triage piece is the part Zendesk wanted most. Zendesk's own routing was rule-based and brittle. You'd set up triggers like "if subject contains 'refund' then assign to billing team" and pray. Forethought's ML-based classification was genuinely better.

The numbers back this up. Forethought processed over a billion interactions per month. Their customers reported 50-70% automation rates on tier-one support. Those are real results from companies operating at scale.

What This Means If You're Choosing Tools Right Now

Four things to watch.

First, if you're a Forethought customer, expect changes. Acquisitions always mean product integration, and "integration" is a polite word for "we're going to merge this into Zendesk Suite and you'll need a Zendesk subscription." Standalone Forethought contracts will probably be honored through their term, then you'll be migrated.

Second, Zendesk's AI pricing will likely change. They were charging $1.50/resolution for AI Agents that weren't as good as Forethought's. Now they have better AI. Do you think the price goes down? It won't.

Third, the standalone AI support tool market just got smaller. Forethought is gone. Ada is enterprise-only and opaque on pricing. The options for teams who want AI classification and automation without committing to a massive platform are shrinking.

Fourth, vendor lock-in risk just went up for everyone in Zendesk's orbit. If your support stack depends on a single vendor for ticketing, live chat, knowledge base, and AI, that vendor controls your pricing forever. You can't leave without rebuilding everything.

The Buy-vs-Build Pattern in AI

Zendesk buying Forethought follows the same playbook as Salesforce buying Slack, or HubSpot acquiring Clearbit. Big platforms don't build the best AI. They buy companies who did, absorb the talent, and bundle it into their existing product at a higher price point.

The teams that stay independent during consolidation tend to be the ones with a narrow, well-defined product that doesn't make sense as an acquisition target. A classification API that costs $0.20 per request with no seat fees doesn't fit neatly into a platform bundle. It's a utility, not a suite feature.

What to Do About It

If you're evaluating support tools right now, think about dependency. How much of your workflow lives inside one vendor? If Zendesk raises prices next year (their 2023 update increased pricing by an average of 16%), can you move?

The safest architecture is one where each layer is replaceable: ticketing, AI classification, and knowledge base as separate tools from separate vendors. If any one of them gets acquired, raises prices, or degrades their product, you swap that layer without rebuilding everything else.

Forethought's customers are about to learn this lesson the hard way. The product they chose doesn't exist as an independent company anymore. Their roadmap is now Zendesk's roadmap.

Pick tools that work together but don't require each other.