UserGems
Also known as: Gem-E, ShelfFlip, Inc., ShelfFlip, User Gems
Signal-driven outbound and ABM platform whose Gem-E agents score accounts against 600+ ICP criteria derived from a company's own closed deals, then research, draft and orchestrate outreach through the sales tools it already runs on. Also connects to Claude or ChatGPT as an MCP server that builds and runs campaigns by prompt.
UserGems is a signal-driven outbound and ABM platform for B2B revenue teams, built on the idea that the moment to reach an account matters more than the size of the list. It watches 21+ native signals, including job changes and champion movements, contact-level buyer intent, website visits, hiring, funding, M&A, tech-stack shifts and 10-K filings, and takes in the company's own first-party material alongside them, including call transcripts, product usage, event attendance, marketing engagement and NPS.
From a company's own opportunities, customers and deal outcomes it derives more than 600 ICP criteria and works out the personas and value proposition that actually convert, then maps accounts into buying stages with weights the team can inspect and edit and a decay that keeps prioritization current. A set of specialized Gem-E agents carries the work from there.
A research agent reads call transcripts and past email exchanges to recover why a deal was lost and what the buyer objected to, a writing agent drafts personalized multi-step sequences that reference those past interactions and refer back to their own earlier emails, discovery agents find and enrich the missing contacts at an account, and an orchestration agent coordinates the result across systems, creating records in Salesforce or HubSpot, enrolling contacts into Outreach or Salesloft, syncing contact-level audiences to LinkedIn, Google and Meta, and alerting the team in Slack.
A Chrome extension puts the same signals and context in front of reps inside whichever tool they are already working in. UserGems does not send email itself, so a connected sales engagement platform is part of the setup. The platform is also reachable as an MCP server, letting a team run outbound and ABM from Claude or ChatGPT by prompt, and it calls other MCP-connected tools mid-campaign to pull transcripts, fetch the right case study or trigger a gift.
It is delivered as multi-tenant cloud software hosted in the United States with no region or self-hosted option, holds a SOC 2 Type II attestation published through its trust portal, and supports SAML single sign-on. Plans are published at $40,000, $75,000 and $150,000 a year with credit pools and unlimited users, plus a quoted Enterprise tier.
Vendor details
Canonical URL
https://usergems.com
Category
GTM / revenue agent
Subcategory
Sales — GTM signals + AI outbound
Funding status
Independent. The vendor of record is ShelfFlip, Inc., trading as UserGems, of 2443 Fillmore St, San Francisco. The company's own news index carries a $2.4M seed announcement and a $20M raise announced in October 2021; no later round appears on that index. Customers named on its own property include Outreach, Mimecast, Docebo, Quickbase, CaptivateIQ, Sendoso, Lattice, Segment, Mixpanel and Cato Networks.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Writes into Salesforce and HubSpot, enrolls contacts into sequences in Outreach and Salesloft, syncs contact-level audiences to LinkedIn, Google and Meta, posts alerts to Slack and triggers gifts through Sendoso. Pulls call transcripts from Gong and content from a Webflow CMS mid-campaign, and ingests 21+ native signal types plus the customer's own first-party data. A UserGems MCP server lets any MCP-compatible assistant drive the platform and lets the agents call other MCP tools during a run, which is the route to extend beyond the named connectors. A product-level API page is published, but no SDK or public developer documentation site is.
In practice
A buyer who championed you changes jobs and your pipeline doesn't notice. UserGems watches for champion moves and surfaces the warm opening while it's still warm.
Your reps chase accounts that don't fit while good-fit ones sit idle. UserGems scores accounts against criteria drawn from your own closed deals, and shows you which signals moved each one, so you can argue with the ranking instead of taking it on faith.
A deal you lost a year ago quietly comes back into market. UserGems notices the signals, reads the call transcripts and emails from the original cycle to recover what the buyer actually objected to, and drafts the re-approach around it.
Signals are useless if acting on them means leaving your sequencer. UserGems creates the records in Salesforce or HubSpot and enrolls contacts into Outreach or Salesloft, so the play runs in the tools your team already uses.
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
8.0 / 14 capabilities · 57%
| Integrations & Tool Calling | Full |
|---|---|
|
The agents take authenticated action across many different kinds of system. Outward they create and enrich records in Salesforce and HubSpot, enroll contacts into sequences in Outreach and Salesloft, sync contact level audiences to LinkedIn, Google and Meta, post alerts to Slack, and trigger physical gifts through Sendoso. Inward they ingest 21+ native signal types plus the customer's own first party material, including call transcripts, product usage, events and content engagement, and they pull call transcripts from Gong and matching customer stories from a Webflow CMS mid run. Any MCP compatible tool can also be wired into a workflow and called by the agents during a campaign, so the connector list stays open. Together that spans CRM, sales engagement, advertising, messaging, conversation intelligence, content management and gifting. Sourceusergems.com/product/mcp and usergems.com/product/overviewread 2026-09-02 |
|
| Workflow Orchestration | Full |
|
Specialized agents run a named chain end to end, on a schedule, across systems UserGems does not own. In UserGems' own worked campaign, the agents detect past champion movements at ICP accounts, enrich or create the records in Salesforce, write a personalized five email sequence, enroll the contacts into a named Outreach sequence, sync them to a LinkedIn ad audience and trigger a gift through Sendoso for anyone scoring above 85. The campaign then repeats every Monday without being asked again. A second example runs on closed lost accounts that reach Tier A, pulling Gong transcripts and the right Webflow case study before drafting. Output passes between agents, not between separate tools. Data agents capture and enrich, intelligence agents score and prioritize, the writing agent drafts from that context, and the orchestration agent coordinates outbound, ads, alerts and CRM writes. UserGems sums this up as set once, runs forever, and the steps are named. Sourceusergems.com/product/mcp and usergems.com/product/overviewread 2026-09-02 |
|
| Knowledge Grounding & RAG | Full |
|
A model built from the customer's own history is read by the agents on every run. UserGems analyzes the customer's opportunities, customers and deal outcomes to derive 600+ ICP criteria and to identify personas and value proposition. That model updates continuously as outcomes accumulate, instead of being assembled per request. The customer's own material is ingested directly, including call transcripts, email exchanges, product usage, event attendance and content engagement. The research agent reads that corpus instead of relying on what reps wrote down. It analyzes call transcripts and email exchanges to recover a buyer's goals, objections and pain points and carries them into the messaging. The writing agent is also trained on uploaded brand guidelines, best performing emails and customer stories. So the agents draw on the customer's own knowledge of both who it sells to and how it sells. Sourceusergems.com/product/overview and usergems.com/product/ai-emailsread 2026-09-02 |
|
| Human Oversight & Guardrails | Partial |
|
The customer can edit the scoring weights directly, so the team, not the model, sets the targeting decisions that determine who gets worked. Gem-E is trained on uploaded brand guidelines, best performing emails and approved customer stories, which bounds what it may produce. UserGems ships no send capability. Drafts are enrolled into Outreach or Salesloft, so the review and send step belongs to the sequencer the customer already runs. UserGems names no approval step, review queue, hold or confidence threshold of its own. The always on campaigns detect movements, create CRM records, write outreach, enroll contacts and trigger gifts on a weekly schedule with no human step. Sourceusergems.com/product/mcp and usergems.com/product/ai-emailsread 2026-09-02 |
|
| Security, Identity & Governance | Full |
|
SAML single sign on runs through the customer's own identity provider, so an organization controls authentication to UserGems and enforces its own password policy, account recovery and multi factor requirements. Sign-in with Outreach via OpenID, least privilege access to customer data and MFA on administrative access sit alongside it. Data is encrypted with AES-256 at rest and with TLS and HSTS in transit, and secrets are protected with KMS. UserGems runs annual third party penetration testing and has passed a Salesforce AppExchange security review that included a further independent test. A SafeBase trust portal publishes a SOC 2 Type II report alongside a pentest report, subprocessor list, data processing agreement, SIG Core and CAIQ questionnaires, data flow diagram and certificate of insurance, with GDPR, CCPA and an EU AI Act limited risk classification. The vendor of record is ShelfFlip, Inc., trading as UserGems, and the attestation and the trading name sit with the same entity. Sourceusergems.com/legal-security/security and trust.usergems.comread 2026-09-02 |
|
| Observability & Auditability | Partial |
|
Scoring is transparent. Accounts are mapped into buying stages with customizable weights and automatic decay, and teams can see exactly why an account sits in a given stage, which signals moved it there and how close it is to sales ready. UserGems positions this against opaque intent only scoring. Reporting covers TAM progression, pipeline generated and return on dollar spent. That transparency covers prioritization only. UserGems names no trace of the agents' actions, no record of what Gem-E wrote for a given contact and on what basis, what the research agent read, which campaign step fired, or what was enrolled and when, and no log surface, export, retention setting or configuration audit trail. A customer can see why an account was chosen, but nothing published shows why it was approached the way it was. UserGems also runs central logging of API and operating system calls through detection rules, which is its own internal monitoring, not an audit trail for customers. Sourceusergems.com/product/overview and usergems.com/ai-agents/gem-e-for-abmread 2026-09-02 |
|
| Memory & State Persistence | Partial |
|
When Gem-E composes a multi step sequence, it knows which signals it has already used and refers to its own previous emails so the thread reads as continuous. It remembers previous steps and adjusts later messaging based on the reply received or on signals that arrived since. Always on campaigns keep their configuration and rerun on schedule indefinitely once set. The account, contact and signal store stays synchronized with the CRM. What Gem-E keeps lasts for one sequence or one campaign. UserGems names no recall across sessions, no per agent scratchpad carried between campaigns, no learned preferences that build up across runs, and no retention, inspection or forgetting control for the customer. Sourceusergems.com/product/ai-emails and usergems.com/product/mcpread 2026-09-02 |
|
| Deployment & Data Residency | Not documented |
|
The service runs as a multi tenant cloud in one fixed location. It uses a cloud service provider's data centers in the United States, and customer data is not authorized to leave the production service environment. Hosting is on AWS in the US, with EU data transfer documentation that covers moving data out of Europe, not an option to keep it there. Customers cannot choose a region. No EU or other non US option, private VPC, single tenant tier, on premises route or install inside the customer's own tenant is published. The VPC and firewall configuration UserGems describes belongs to its own production network, not to a deployment the customer controls. Sourceusergems.com/legal-security/security and trust.usergems.comread 2026-09-02 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
The platform ships a fixed roster of agents the customer does not author. Signals, Research, Account and Contact Discovery and Data Hygiene are data agents, and Scoring, List Building, Personalization, Orchestration and Advertising are intelligence agents, all on top of 21+ native signal types. The Playbooks library is browsable and filterable by team, with each play tagged to the signal it runs on. It sits under Resources beside the blog, guides and webinars, and its entries are authored articles carrying bylines and collaborations with named people. A customer reads a play and then builds the campaign itself, instead of selecting a packaged configuration into the product. UserGems also offers three worked setup prompts a customer can paste to stand up an always on campaign. Sourceusergems.com/playbooks and usergems.com/product/overviewread 2026-09-02 |
|
| Triggers & Channel Coverage | Full |
|
Signals feed the agents from 21+ native types, including job changes and champion movements, buyer intent, website visits, role changes, hiring, funding, M&A and 10-K filings. Contact level resolution identifies which individual is researching, not only that an account is in market. The customer's own first party feeds of marketing engagement, form submissions, event attendance, product usage and NPS come in too, and a research agent handles custom signals. Campaigns also run on a clock, and UserGems' worked examples fire weekly and monthly. Results go out as email sequences through connected engagement platforms, contact level ad audiences to LinkedIn, Google and Meta, Slack alerts, CRM record creation, and insights surfaced in the rep's browser during calls. Sourceusergems.com/product/overview and usergems.com/product/mcpread 2026-09-02 |
|
| Model Flexibility & Routing | Not documented |
|
No provider is named for the models powering Gem-E, and no model selector, per agent choice, routing policy, bring your own model or bring your own key option is published. Claude, ChatGPT and GPT-4o appear only as the assistants a customer prompts from, calling UserGems as a tool. Sourceusergems.com/product/mcp and usergems.com/product/overviewread 2026-09-02 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Outside callers can drive the platform through UserGems MCP, which is added as an MCP server in any MCP compatible interface, with Claude and ChatGPT named. It executes instead of only returning data, so a connected assistant builds campaigns, creates ad audiences, enrolls contacts into sequences, drafts outreach and updates the CRM, using the same agent that runs inside the product. It also accepts inbound connections from the customer's existing GTM tools and calls other MCP servers mid workflow. UserGems supports the build case too, feeding a customer's homegrown model the signals, scored accounts and enriched contacts it needs. An API is also sold as its own product level offering. A help center runs at help.usergems.com. No public developer documentation site, SDK, client library or endpoint reference is published, and the MCP has a marketing product page but no developer docs. Sourceusergems.com/product/mcp and usergems.com/product/apiread 2026-09-02 |
|
| Testing, Debugging & Optimization | Not documented |
|
Reporting covers TAM progression, pipeline generated and return on dollar spent, and reply rates are measured against the agent's own output. UserGems names no controlled comparison against a criterion, such as an A/B test of message or sequence variants, a holdout, a scored test set, a sandbox, a dry run or a regression check when a prompt or scoring model changes. The writing agent adjusts messaging in light of a reply or a newly arrived signal, but that adaptation happens inside a single sequence, and the customer has no surface to drive it. A customer can read what happened but has no stated way to test anything before it reaches a prospect. UserGems also offers a money back ROI guarantee and claims reply rates of 6-20% against an industry average. Sourceusergems.com/product/overview and usergems.com/product/ai-emailsread 2026-09-02 |
|
| Browser & Computer Use | Not documented |
|
The AI Chrome extension brings AI assistance into reps' browsers, so they get insights while working in their CRM, sales engagement tools or during calls, and UserGems says the point is that reps have no new tools to learn. It surfaces signals and context on a page a person is already looking at, and the person acts. UserGems names no case of the agent driving an interface itself. Everything else runs through programmatic routes. CRM writes, sequence enrollment, ad audience sync, Slack alerts and gifting all go through integrations, and the MCP executes against the product's own surface. No headless browsing, virtual desktop, screen or click automation, or navigation of a third party console is published. Sourceusergems.com/product/overviewread 2026-09-02 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
UserGems launched an MCP server that exposes its AI command center to assistants such as Claude and ChatGPT, along with new writing and research capabilities.
Bears on: MCP / tool calling / API
View sourcePricing
Core $40,000 a year with 4,000,000 credits · Advanced $75,000 a year with 10,000,000 credits · Elite $150,000 a year with 25,000,000 credits · Enterprise quoted · unlimited admin and end users · money back guarantee on ROI
annual platform subscription with a credit pool spent on data actions and AI actions; unlimited admin and end users
Included quota
Core: 4,000,000 credits (40K data actions or 400K AI actions), TAM coverage of 2,000 to 4,000 accounts, $3k implementation. Advanced: 10,000,000 credits (100K data actions or 1M AI actions), 4,000 to 10,000 accounts, $5k implementation. Elite: 25,000,000 credits (250K data actions or 2.5M AI actions), 10,000+ accounts, $10k implementation. Unlimited admin and end users on every plan.
What is public
A pricing page with three priced annual plans, Core $40,000, Advanced $75,000 and Elite $150,000, each with a credit pool stated in data actions and AI actions, TAM coverage, implementation fees and unlimited users, plus a quoted Enterprise tier. Every plan's call to action is Contact Us, and the page carries a money back guarantee if a customer does not see ROI in closed won revenue.
Billing mechanics
Annual subscription per plan with a pooled credit allowance; one data action costs 100 credits and one AI action 10, per the plan lines (4,000,000 credits equal 40K data actions or 400K AI actions). Users are unlimited. An implementation fee applies by plan, and add-ons include Retroactive Job Change Leads at $3k and a vendor sponsored SPIFF. Purchase runs through Contact Us.
Cost watchouts
Implementation fees of $3k on Core, $5k on Advanced and $10k on Elite sit on top of the annual price, and Retroactive Job Change Leads is $3k. UserGems drafts and enrols outreach but does not send it, so a connected sales engagement platform such as Outreach or Salesloft is a separate subscription.
Variable cost rationale
Plans are fixed annual fees with a pooled credit allowance stated in data actions and AI actions; the price of credits beyond the pool is not published.
Additional watchouts
Budget for the sequencer as well as the platform, because Gem-E writes and enrols but does not send. Implementation fees and add-ons are priced separately from the annual plan.
Overage / add-ons
Not published.
Sales call required
Yes, required for paid access
Free / trial
No trial is offered on the property, and no self-serve route exists: the only account link is a login for existing customers. The vendor publishes a money-back guarantee instead, promising at least 1x ROI in revenue generated.
Lowest paid plan
Core, $40,000 a year with 4,000,000 credits (40K data actions or 400K AI actions) and TAM coverage of 2,000 to 4,000 accounts.
Commercial notes
Signal-driven outbound and ABM with Gem-E agents, sold through a demo-led motion to B2B revenue teams. Publicly documented commercial terms are limited to a money-back guarantee at 1x revenue generated and a typical three to four week implementation. Customers named on the vendor's own property include Outreach, Mimecast, Docebo, Quickbase, CaptivateIQ and Sendoso.
Key ambiguities
Contract length, renewal terms and the cost of credits beyond the pool are not published, and the terms of the money back guarantee beyond ROI in closed won revenue are not stated.
Cancellation / refund
Not published beyond a money back guarantee if a customer does not see ROI in closed won revenue; contract length, notice and renewal terms are not stated.
Support SLA / resale
Not published. Implementation typically takes three to four weeks per the vendor's homepage. Native Salesforce and HubSpot integration, sales engagement integration with Outreach and Salesloft, and ad platform sync to LinkedIn, Google and Meta are documented on product pages; whether any is tier-gated is not published, since no tiers are published. SAML single sign-on is documented on the security page with no indication that it is an add-on.
Missing data
Published: the three plan prices, credit pools, TAM coverage, implementation fees, add-on prices and unlimited users. Not published: contract length, renewal terms, Enterprise pricing and the price of extra credits.
Related vendors
- 11x — AI digital workers (Alice for outbound SDR work, Julian for inbound…
- 1up — AI answer engine for sales teams that automates RFPs and security…
- Actively AI — GTM superintelligence platform that trains a custom reasoning model…
- Aircover.ai — AI sales assistant that delivers real time in call coaching, a…
- AiSDR — AI SDR platform with multichannel outreach automation, quota visible…
- Akkari — Autonomous customer ops agent that captures every commitment,…
Alternatives to UserGems
The closest documented capability profiles to UserGems among GTM and revenue agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Lawmatics7.5 / 14A lighter documented profile than UserGems
- Day AI9.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Memory & State Persistence
- Otter.ai8.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs
- Qualified8.0 / 14Adds documented Testing, Debugging & Optimization
- Warmly8.0 / 14Adds documented Testing, Debugging & OptimizationUserGems vs Warmly →
- Amplemarket9.5 / 14Adds documented Testing, Debugging & Optimization
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded