Freshworks used its Refresh event on November 13, 2025 to push its customer service suite further into agentic AI, announcing the Freshdesk Command Center, a set of industry-specific "Vertical AI Agents," and Freddy AI Insights. The pitch is familiar from every vendor in this category right now: fewer tabs for agents, more tickets closed without a human. What makes the Freshworks version worth examining is less the feature list than the billing model underneath it, which differs from how most rivals charge for AI.
We checked the announcement against Freshworks' own product documentation and its public pricing page to separate what ships today from what is still marketing language.
What Freshworks Actually Announced
Freshdesk Command Center
The Command Center is a consolidated workspace that pulls email, chat, WhatsApp, and social messages into a single view alongside Freddy's AI tools. The stated goal is removing the app-switching tax on agents. In practice this is Freshworks catching up to the unified-inbox pattern that omnichannel desks have converged on, with the addition that AI suggestions and one-click actions live in the same pane rather than a sidebar.
Vertical AI Agents
This is the more differentiated piece. Rather than shipping a blank bot builder, Freshworks now offers prebuilt agentic workflows tuned for ecommerce, fintech, travel, and logistics. According to the company's announcement, these ship with 50+ prebuilt workflows covering the repetitive intents those industries share: order status, refund eligibility, booking changes, shipment tracking.
Freshworks says early access customers averaged a 65% deflection rate, with some reaching 80% resolution by AI agents, and cites customer iPostal1 cutting interaction time to seven seconds. Treat these as vendor-supplied figures from self-selected early adopters, not independently audited benchmarks. They are directionally useful and nothing more.
Freddy AI Copilot and Insights
Copilot is the agent-assist layer: conversation summarization, translation, reply drafting, solution-article suggestions, sentiment analysis, and auto-triage. Freshworks' support documentation for Freddy AI confirms these run on both classic Freshdesk and Freshdesk Omni, and that all Freddy components require purchasing the add-on plus explicit role assignment before agents can access them.
Freddy AI Insights targets team leads rather than frontline agents, surfacing trend detection, SLA breach warnings, anomaly flags, and root-cause analysis, with a natural-language query interface over support data.
What It Actually Costs
Pricing is where coverage of this product most often goes wrong, so here are the figures straight from the official Freshdesk pricing page, all on annual billing:
- Growth — $19 per agent/month
- Pro — $55 per agent/month
- Enterprise — $89 per agent/month
- Freddy AI Copilot — $29 per agent/month, available on Pro and Enterprise only
- Freddy AI Agent sessions — $49 per 100 sessions, available across all plans
- Day passes — $2 (Growth), $7 (Pro), $12 (Enterprise) per pass
- Connector app tasks — $80 per 5,000 tasks
The stacking matters more than any single line. A team on Pro that wants Copilot for everyone is at $84 per agent/month before a single AI session is billed, within striking distance of Enterprise list price but without Enterprise's feature set. Budget the AI add-ons as a second product, because that is effectively what they are.
The Session Billing Catch
Freshworks bills AI Agent usage per session, not per resolution. That distinction is the single most important thing to understand before signing.
An analysis of Freddy AI pricing by eesel documents how sessions are counted: for the email AI agent, a session is a 72-hour window from the customer's first email, with multiple AI responses inside that window counting once. For chat, it is a 24-hour window. Pro and Enterprise include 500 free sessions, but this is a one-time allocation rather than a monthly refresh, and unused sessions do not roll over.
The consequence is straightforward arithmetic. If your AI resolves a meaningful fraction of contacts but not all of them, you still pay for every session it touched, including the ones that escalated to a human anyway. That same analysis argues effective cost per resolved ticket lands several times above the headline per-session rate. Note that eesel sells a competing product, so weigh its conclusions accordingly; the session-counting mechanics it describes, however, match Freshworks' own documented behavior.
Several competitors price AI on resolutions instead, meaning the vendor absorbs the cost of failed attempts. Neither model is universally cheaper, since resolution pricing usually carries a higher unit rate, but session pricing shifts the risk of poor AI accuracy onto the buyer. Model both against your actual ticket volume before committing.
Strengths
- Vertical prebuilt workflows meaningfully shorten time-to-value versus building intents from scratch
- Freshdesk remains genuinely easy to deploy, and reviewers on G2 consistently cite fast onboarding as the product's defining advantage
- Copilot and AI Agent both work on classic Freshdesk and Omni, so buyers are not forced onto the newer SKU to get AI
- Entry pricing at $19 per agent/month remains among the more accessible in the category
Weaknesses
- Session-based AI billing charges for attempts rather than outcomes
- Copilot is gated to Pro and above, putting the full AI story out of reach for Growth-tier buyers
- Deflection statistics come exclusively from Freshworks and its early-access customers
- Longstanding user complaints about spam filtering and email handling predate this release and were not addressed by it
- Deeper analytics has historically required higher tiers, a pattern Insights does not fully change since it is itself a paid add-on
Who Should Consider It
Freshworks has built a credible agentic support stack, and the vertical workflow templates are a real answer to the cold-start problem that sinks most support bots. Teams in ecommerce, fintech, travel, or logistics running high volumes of predictable, repetitive intents will get the most from it. Those are exactly the conditions where per-session billing works in the buyer's favor, because most sessions end in a resolution.
Teams with complex, low-volume, high-touch support should run the numbers hard first. If the AI fails on a third of attempts, you pay for those failures. Ask Freshworks for a session-volume estimate based on a sample of your real ticket history before signing, and treat any deflection figure not derived from your own data as a hypothesis rather than a forecast.