Enterpret

What it is

Enterpret reads customer feedback that already exists somewhere else and turns the free text into one queryable structure. It pulls from 60-plus sources: Zendesk, Intercom, Front, Pylon, Kustomer and Salesforce Service Cloud on support; Gong, Chorus, Fathom and Grain on calls; G2, Trustpilot and both app stores on reviews; Qualtrics, Sprig, SurveyMonkey and Typeform on surveys; Discord, Discourse and the social channels on community. Amplitude, Mixpanel, Segment and Snowflake come in on the quantitative side so a theme can be sized against behavior.

Two pieces do the work. Adaptive Taxonomy builds a model of your company’s own language instead of asking you to maintain a tag list — the vendor’s framing is “no manual tagging, no drift.” The Customer Context Graph joins each piece of feedback to the account, the product area and the revenue attached to it, which is what turns a theme into a number a QBR can act on.

The company raised a $20.8M Series A led by Canaan Partners on 4 December 2024, with Kleiner Perkins, Peak XV Partners and Wing Ventures participating, bringing total funding to $25.1M. Canva, Notion, Atlassian, Descript, ElevenLabs, Western Union, Fanatics, Bitvavo and Feeld appear on the public customer list. On 27 October 2025 it shipped Agent OS, which is the agentic layer described below.

Why it shows up in Customer Success stacks

• The MCP server is the reason this rates above the category. It exposes seven named tools — get_organization_details, get_graph_schema, get_query_examples, search_graph_fields, search_graph_values, run_graph_query and find_user_quote — so an assistant can chain queries, filter by segment and pull verbatim quotes with citation links back to the source record. Auth is OAuth per member, or a bearer token that expires after six months. Supported clients are Claude on web, desktop and Code, Cursor, Codex, Notion Custom Agents and Glean. Per-member authentication means queries are attributable to a person, which is what lets this survive a data-access review. The practical difference: the VoC corpus is reachable from the session where the analysis is already happening, rather than from a dashboard someone has to remember to open.

• Adaptive Taxonomy removes the tagging job, not just the tagging tool. Every hand-maintained feedback taxonomy decays the moment the product ships a new surface, and the decay is invisible — old labels keep absorbing new problems. A model that re-learns the label set against current text is the difference between a taxonomy that describes last year’s product and one that describes this quarter’s.

• Agent Automations run on a schedule or on a signal. Scheduled templates ship as Help Center Gaps, Account Health Monitor, Launch Monitor, VoC Report, Support Team Scorecard, Customer Story Tracker and Wall of Love. Signal-driven ones are Quality Monitor, which detects a feedback spike and goes after the root cause, and Escalation Shield, which fires on urgent feedback from high-value accounts. Each posts to Slack, opens a ticket or writes a document. The Enterpret Agent itself holds memory across sessions and packages repeated analyses as reusable Skills.

Pricing reality

Nothing is published and there is no free trial. Vendr’s marketplace data puts the median Enterpret contract at $36,302/yr, ranging $23,040 to $98,000. The variable is data volume and integration count, not seats — aggregator listings describe the top tier as carrying unlimited user licenses and customer-specific models. That inverts the arithmetic most Customer Success buyers are used to: adding readers costs nothing, adding sources costs real money. Size the business case on channels you will actually query, and treat $30–50K/yr as the honest mid-market band.

Best for

Product, CX and Customer Success leaders past roughly 5,000 inbound feedback items a month across four or more channels, where someone is hand-maintaining a tag taxonomy today and the quarterly VoC readout costs an analyst most of a week. The fit sharpens when an AI assistant is already in the team’s daily workflow — the MCP server is the part competitors do not match, and it is worth little to a team that will only ever open the web app.

Do not buy it if

Feedback lives in one system. If Zendesk is your entire surface, the native intelligence features plus a saved view get you most of the way, and the integration work you avoid outweighs the theme quality you give up. The same holds under about 1,000 pieces of feedback a month, where a monthly export into a Claude project answers the same questions for the cost of an afternoon; at the $36K median, this tool is roughly a quarter of an analyst’s fully loaded cost, so the corpus has to be bigger than one person can read. And it is not a survey platform — Enterpret reads Qualtrics, Sprig and SurveyMonkey, it does not run the survey.

Versus the alternatives

• Qualtrics — the largest install base in experience management, and it closed its $6.75B acquisition of Press Ganey Forsta on 18 May 2026. Pick it when you need survey collection, case management and program governance on one contract. Pick Enterpret when the feedback already exists and the problem is reading it.
• Medallia — the other incumbent by install base, and the one to price risk against. On 22 April 2026 a creditor group including Blackstone, KKR, Apollo Global and Antares Capital took control from Thoma Bravo in a debt-for-equity swap that wiped out roughly $5B of sponsor equity and injected $150M of new capital. The product runs and the accounts are being served; what changed is who sets the roadmap. Ask for a written product-investment commitment before a multi-year term.
• Chattermill — the closest AI-native peer, stronger where consumer and retail CX benchmarking is the job. Pick it when you need cross-brand comparison more than you need the corpus addressable by an agent.
• Unwrap.ai — the fastest-growing entrant in the segment: a $12M Series A led by Scale Venture Partners in January 2025 with Atlassian Ventures participating, about $15M raised in total, and Microsoft, Lyft, jetBlue, Oura and Perplexity named as customers. Pick it when the buyer is a product team and the budget sits below Enterpret’s floor.
• Gainsight — not a substitute, a neighbor. It owns the account health and CSM motion; Enterpret feeds it the verbatim evidence. Teams that buy both wire Enterpret’s escalation signal into the health score rather than running two opinions about the same account.

Watch-outs

• Taxonomy setup and upkeep is the recurring complaint, not price. Enterpret rates 4.5 across 111 reviews on G2, and what reviewers raise is the setup investment and the maintenance the taxonomy still asks for. Guard: name the 10 to 15 themes your QBR actually runs on before kickoff, then validate the model’s labels against a hand-labeled sample of 200 records in month one. Disagreement above roughly 10% is a scoping conversation, not a launch.
• Metering on volume means “connect everything” is how the renewal doubles. Guard: start with the four sources that already drive decisions, and add a channel only when a specific unanswered question required it. Get the volume tier and the per-unit overage rate written into the first order form, not discovered at renewal.
• The MCP bearer tokens expire after six months, and each member authenticates individually. Guard: prefer OAuth for humans and reserve bearer tokens for machine automations, then calendar the rotation at signature. Behind a corporate firewall, HTTPS access to mcp.enterpret.com has to be allowed — confirm that with IT before the pilot rather than during it.
• Agent Automations write into other systems. Escalation Shield and Quality Monitor post to Slack and can open tickets, which means a false positive becomes someone’s Jira backlog. Guard: run both into a private channel for two weeks and count the false-positive rate before granting ticket-creation rights.
• The headline customer metrics are vendor-published. Enterpret’s site cites 360% time savings at Notion and a 40% support-ticket reduction at Apollo.io. Guard: treat them as directional and ask for the reference call with a company at your feedback volume, not at Canva’s.