SalesCloser.ai
AI voice and video sales agent (Agent Frank) that runs live discovery calls and demos.
SalesCloser AI is a no-code platform for building autonomous AI sales agents that handle live sales conversations over the phone, on Zoom, and on browser-based calls. Where many AI sales tools focus on email or LinkedIn sequences, SalesCloser is built for the spoken, real-time part of selling: qualifying a lead, running a discovery call, and walking a prospect through a demo without a human rep on the line.
Agents are role-based. A team prompts the kind of agent it needs, such as discovery, demo, onboarding, or support, then trains it by uploading sales scripts, objection handling, product specs, and FAQs, and sets the brand voice. From that material the agent conducts tailored discovery calls to surface a prospect's needs, delivers interactive screen-share product demonstrations, qualifies leads against defined criteria, and books meetings, sending calendar invites, confirmations, reminders, and reschedules, then follows up automatically.
A notable strength is global reach. Through a multilingual capability spanning dozens of languages, an agent can greet an inbound prospect in their own language and run the entire discovery and demo conversation in it, operating around the clock so leads in any time zone are engaged immediately rather than waiting for a rep to wake up. Because the platform works over voice and video rather than moderated email or LinkedIn outreach, teams avoid the account-flagging and deliverability risks that come with those channels.
SalesCloser is designed to sit alongside an existing stack rather than replace it, connecting to CRM and sales systems and a broad set of integrations so call data and booked meetings flow into the tools a team already uses. It is offered by Wishpond Technologies, a publicly traded company, and targets sales teams that depend on live conversations, including SaaS and service businesses handling inbound or outbound calls at scale.
Vendor details
Canonical URL
https://salescloser.ai
Category
GTM / revenue agent
Subcategory
Sales — voice/video sales agent
Company status
unclear
Use cases & customers
Primary use cases
Target customers
Deployment options
In practice
An inbound lead lands at 2am while your reps are asleep. SalesCloser's AI agent answers the call, qualifies the prospect, and runs a screen-share demo right then, instead of letting them wait or book with a competitor.
You need different conversations for different stages. SalesCloser uses role-based agents, so you spin up a discovery agent or a demo agent, train each on your scripts and product specs, and set your brand voice.
A prospect speaks French and your team doesn't. SalesCloser greets them in their own language and runs the entire discovery and demo call in it, across dozens of languages, around the clock.
Agentic Index coverage score
7.5 / 14 capabilities · 54%
| Integrations & Tool CallingBreadth comes from the action surface a live conversational agent necessarily spans, not from a connector catalogue. To do its job the agent operates telephony for inbound and outbound calls, video conferencing (it joins Zoom), calendar (it negotiates times and books directly), and CRM (it writes call outcomes back the moment a call ends, or acts as the system of record itself). On top of that Wishpond launched a SalesCloser INTEGRATIONS MARKETPLACE in August 2024 spanning three further software classes, CRM, email marketing and task management, with Zendesk Chat and Asana named, and users able to build automation flows across integrations. ENUMERATION GAP RECORDED: the 200 plus platforms figure appears on the demo booking page with no published catalogue behind it, and only a handful of integrations are named anywhere first party. Graded Full on the documented multi class action surface, medium confidence on the catalogue. SalesCloser product pages and Wishpond marketplace announcement 2026-07-30 | Full |
|---|---|
| Workflow OrchestrationUPGRADED FROM THE UNSOURCED GRID, which carried Partial, and this is a harder orchestration problem than anything else in the GTM cluster. The agent runs a live synchronous sales conversation end to end: it places or answers the call, qualifies, runs discovery, joins the video meeting, shares its screen, delivers the demo, handles objections as they arrive, negotiates a calendar slot, books the meeting and writes the outcome to CRM. Crucially it REPLANS IN REAL TIME under adversarial input, with the vendor describing a buyer interrupting mid demo to ask about pricing, the agent jumping to pricing, answering, and returning. That is dynamic branching against an unscripted counterparty, not a sequencer executing configured steps, which is what separates this from closely in the same pass. VERIFICATION GAP STATED PLAINLY: this capability is documented only first party, the vendor itself says it would rather buyers test it than take its word, its G2 profile sits at 2.7 of 5 across 3 reviews and has been unclaimed for over a year, and no independent confirmation of the live demo capability was found in three passes. SalesCloser homepage, demo and pricing pages 2026-07-30 | Full |
| Knowledge Grounding & RAGUPGRADED FROM THE UNSOURCED GRID, which carried Partial. Setup is explicitly knowledge base first: the customer uploads company sales materials, product specifications and FAQ documents, plus a structured OBJECTION LIBRARY and a brand voice definition, and the agent is trained on all of it before deployment. Retrieval is described concretely rather than abstractly, with the vendor stating that when a prospect asks about technical specifications the agent pulls from that exact product documentation rather than improvising. THE HARDER CASE FOR THIS AXIS, and the reason it earns Full: retrieval happens under live conversational latency against an unscripted question, which is a materially harder grounding problem than the batch content generation that earned tofu and jasper their Full grades on the same axis. The objection library in particular is a grounding artefact no other GTM vendor in this pass ships. SalesCloser homepage, demo page and product documentation 2026-07-30 | Full |
| Human Oversight & GuardrailsDOWNGRADED FROM THE UNSOURCED GRID, which carried Full, but the posture is genuinely distinctive and worth naming: AUTONOMY BOUNDED BY DEAL STAGE. The agent runs qualification, discovery and the first demo alone, then hands to a human before the close, and the vendor states this as a hard product boundary rather than a limitation, answering the question of whether it replaces reps with a flat no and calling the opposite claim a fantasy. Humans keep strategy, negotiation and closing. Generation is constrained by the uploaded knowledge base, objection library and brand voice, which the vendor describes as guardrails the agent improvises inside. WHAT KEEPS THIS AT PARTIAL: the boundary is a design decision, not a control the buyer configures. No approval gate, no live human takeover or barge in on a call, no per agent risk calibration and no policy engine was documented. Once the agent is live it converses with prospects unsupervised. SalesCloser homepage, pricing FAQ and demo page 2026-07-30 | Partial |
| Security, Identity & GovernanceTHE BEST DISCLOSURE QUALITY OF ANY WEAK POSTURE IN THIS PASS, which is worth separating from the posture itself. SalesCloser states on its own pricing page: SOC 2 TYPE I COMPLIANT, WITH SOC 2 TYPE II CURRENTLY IN PROGRESS, and actively working toward full GDPR alignment with third party security assessors. NAMING THE TYPE UNPROMPTED IS THE RIGHT BEHAVIOUR and is better than regie-ai and autobound (type unspecified), tofu (either has or is in the process of) and closely (audits aligned with ISO 27001 standards). SUBSTANTIVELY THE POSTURE IS STILL WEAK: Type I attests design at a point in time, not operating effectiveness, and the vendor's own wording confirms GDPR alignment is NOT YET ACHIEVED. No trust centre, no subprocessor list, no penetration test evidence, no encryption specifics and no DPA were found in three passes. TWO LIVE EXPOSURES THE BUYER OWNS: an agent placing outbound calls in an artificial voice raises TCPA questions, the same issue flagged on alta and in the health index RCM lane, and recording calls across 32 languages engages two party consent regimes. The vendor says it supports consent based call handling but names no mechanism. Corporate parent is Wishpond Technologies Ltd, publicly traded in Canada. SalesCloser pricing page and Wishpond corporate filings 2026-07-30 | Partial |
| Observability & AuditabilityAN UNUSUALLY STRONG RAW SUBSTRATE THAT IS NOT EXPOSED AS A GOVERNANCE SURFACE, which is exactly the gap this axis is meant to catch. Because every interaction is a recorded call, the buyer has a complete verbatim record of what the agent said, in a way no text based vendor in this cluster can match; the vendor cites its own customer call logs as evidence in marketing. Call outcomes and data flow into CRM immediately, so campaign level results are visible in the system of record. WHAT IS MISSING is everything that turns recordings into auditability: no documented audit log, no decision trail explaining why the agent said what it said or which knowledge base passage it drew on, no retention policy on recordings, no access controls over who can replay a call, and no per user attribution. A buyer can hear what happened and cannot reconstruct why. Recording is not the same as auditability. SalesCloser homepage and product documentation 2026-07-30 | Partial |
| Memory & State PersistenceWithin conversation context is real and is table stakes for a voice agent: the agent tracks what has been said, what objections have been raised and where it is in the demo, and returns to its place after a digression. Call outcomes persist by being written to CRM the moment the call ends, and call recordings retain the full exchange. WHAT IS CLAIMED BUT NOT DOCUMENTED is the interesting part. Third party summaries describe CONTINUOUS LEARNING, with the system evolving its strategies based on interactions to improve future performance. No mechanism, scope, boundary or control for that was found on any first party surface, and for a vendor whose agent speaks under the customer's brand, a learning loop with undefined scope is a question rather than a feature. No per prospect memory across separate calls, no retention or forgetting control at the agent level. SalesCloser product pages and third party reviews 2026-07-30 | Partial |
| Deployment & Data ResidencyDOWNGRADED FROM THE UNSOURCED GRID, which carried Partial on no evidence. No deployment or residency control of any kind was found: no region selection, no on premises, no private VPC, no single tenant option and no data residency commitment. THE ABSENCE IS REASONABLY POSITIVE RATHER THAN A SEARCH MISS, because the pricing page carries a compliance FAQ that addresses SOC 2 status and call handling directly and names no region or hosting location while doing so; secure data storage is asserted with no geography attached. That matters more here than for a text product, because the vendor processes voice recordings in 32 languages, which implies EU and other non US data subjects, while stating GDPR alignment is still being worked toward. JURISDICTION NOTED, NOT CREDITED: the corporate parent is Wishpond Technologies Ltd of Vancouver, so Canadian rather than US jurisdiction, which has different implications for EU buyers, but the vendor makes no residency claim and no DPA was located. FLAGGED FOR RE VERIFICATION. SalesCloser pricing page, demo page and Wishpond corporate filings 2026-07-30 | Unable to verify |
| Prebuilt Agents, Templates & PacksAgent creation is templated and genuinely no code: the customer defines the agent's role, uploads a knowledge base, picks a voice from the library and ships, with the vendor describing setup measured in hours and publishing a walkthrough from first login to a live agent making its first call. Documented agent roles span lead qualification, discovery, product demo, onboarding and customer support, so several distinct jobs are supported out of the box, and 32 language voices function as a packaging layer. WHAT IS ABSENT is a catalogue: no named prebuilt agents comparable to demandbase's five or tofu's three, no template marketplace, no importable play library and no community exchange. This is a good agent BUILDER rather than a library of agents, which is the distinction this axis draws. The Wishpond Integrations Marketplace is a connector catalogue, not an agent catalogue, and was not credited here. SalesCloser homepage, demo page and third party reviews 2026-07-30 | Partial |
| Triggers & Channel CoverageDOWNGRADED FROM THE UNSOURCED GRID, which carried Full. Depth in one channel class rather than breadth across classes, which is the distinction this axis turns on. Live synchronous conversation is covered three ways, phone, Zoom and browser based calls, in 32 languages, running 24 hours so an inbound lead at 9pm gets answered, plus async follow up including confirmations, reminders and rescheduling. THAT IS ONE CHANNEL CLASS DELIVERED WELL, PLUS ITS FOLLOW UP TAIL. No email sequencing, no social, no advertising, no website or in product surface. Triggers are inbound lead arrival and scheduled outbound calling rather than a signal framework; there is no intent data, no account identification and no event driven trigger library. Compare the Full grades in this cluster, where tofu spans six output formats across five channel classes. SalesCloser product pages and third party reviews 2026-07-30 | Partial |
| Model Flexibility & RoutingDOWNGRADED FROM THE UNSOURCED GRID, which carried Partial on no evidence. No model provider is named anywhere across three passes covering the homepage, pricing page, demo page, blog and corporate announcements. The only description offered is functional, that language capabilities are powered by AI models trained for real conversational use. No model selection, no routing, no bring your own model, no bring your own key. Voice selection exists but that is a persona control, not model flexibility, and was credited under packaging instead. THIRD CONSECUTIVE BUILD WITH THIS EXACT SHAPE after tofu and closely: single provider dependency with the provider undisclosed, which is weaker than gorgias (dependency, provider known) and categorically unlike omilia, pokee-ai or leena-ai, where a 0.0 reflects deliberate vertical integration enabling a certified perimeter. For an agent speaking live under the customer's brand to their prospects, an undisclosed model provider is a live procurement question. SalesCloser product pages and Wishpond announcements 2026-07-30 | Unable to verify |
| APIs, SDKs & MCP ExtensibilityReal no code extensibility, no developer surface. The Wishpond SalesCloser INTEGRATIONS MARKETPLACE, launched August 2024, is the substantive artefact: it connects SalesCloser to CRM, email marketing and task management tools, and the corporate announcement states users can build comprehensive automation flows ACROSS multiple integrations, which is composition rather than point to point sync. Named endpoints include Zendesk Chat and Asana, with lead sync from CRM into SalesCloser and outcomes written back. WHAT WAS NOT FOUND across three passes: no public API documentation, no SDK, no MCP server, no developer portal, no partner or certification programme and no docs subdomain. Extensibility is therefore bounded by what the marketplace already supports, with no path for a customer or third party to extend the platform programmatically. Consistent with a product sold to sales teams rather than to developers. FLAGGED FOR RE VERIFICATION, since Ext is the axis this pass most often understates. Wishpond marketplace announcement and SalesCloser product pages 2026-07-30 | Partial |
| Testing, Debugging & OptimizationA different shape from the other Partials on this axis in the pass, and arguably a better one. Because every interaction is a recorded call, the buyer can evaluate THE AGENT ITSELF rather than only the outcomes of its work, which is what optimizely, demandbase and tofu could not offer. Listening to a sample of calls is a real, direct assessment of agent quality, and the vendor leans on this by inviting prospects to test the agent rather than take its word. WHAT KEEPS IT AT PARTIAL is the absence of any structured tooling around that substrate: no sandbox or dry run mode to rehearse an agent before it speaks to a real prospect, no evaluation harness, no automated conversation scoring, no regression testing when the knowledge base changes, and no A/B framework for comparing agent configurations. Review is manual, unsystematic and after the fact. For a product where a bad output is a live conversation with a real buyer, the missing dry run is the consequential gap. SalesCloser homepage and product documentation 2026-07-30 | Partial |
| Browser & Computer UseUPGRADED FROM THE UNSOURCED GRID, which carried None. SECOND NON ZERO Comp GRADE IN THE GTM CLUSTER, after closely in the same pass, and a stronger case than that one. The agent joins a video call, TAKES CONTROL OF A SCREEN SHARE, and navigates the product or deck live while narrating, adapting its path in real time to whatever the buyer asks. The vendor's own example is a buyer interrupting on slide two to ask about pricing, the agent jumping to pricing, answering, and returning. That is operation of a software interface on the customer's behalf, broadcast live, under real time external input, which is what this axis exists to measure and is categorically more than the API only architecture that earned None across the rest of this cluster. PARTIAL NOT FULL FOR TWO REASONS: the mechanism is undocumented, and the vendor's own wording is ambiguous between navigating a live product interface and navigating a structured demo or deck environment, which are materially different capabilities; and it is confined to the demo surface rather than general browser or computer control. FLAGGED AS A TAXONOMY PRECEDENT QUESTION alongside closely. SalesCloser homepage and third party reviews 2026-07-30 | Partial |
The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify 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
Pricing
From ~$990-$1,000/mo flat (entry ~2,000 min) · higher tiers to ~$2,500/mo · exact tiers demo-gated
flat
Included quota
Entry ~$990-$1,000/mo tier reportedly includes ~2,000 call minutes; higher tiers add more minutes/agents (exact allotments not officially published).
What is public
SalesCloser.ai (salescloser.ai - AI sales agent that runs live phone/Zoom/browser sales conversations: qualifies, runs demos, schedules, closes, in 32 languages, no-code) advertises flat-rate pricing 'from $1,000/mo' (no per-minute charges) on its pricing page, but the page routes to a demo to buy. Third-party directories document tiers ~$990-$2,500/mo flat (e.g., a $990/mo tier with ~2,000 included minutes; Enterprise from ~$2,500/mo).
Billing mechanics
Flat-rate monthly subscription (advertised 'from $1,000/mo,' no per-minute charges; bundled call minutes per tier); exact tiers are demo-gated and documented by third parties (~$990-$2,500/mo). 200+ integrations; CRM/calendar connect.
Cost watchouts
Flat-rate only pencils out at very high call volume (break-even ~12,500 min/mo vs ~$0.08/min usage models); reviews flag billing issues (post-trial charges, unclear cancellation, auto-renewal) - confirm trial/cancellation terms in writing.
Variable cost rationale
Flat monthly fee with bundled minutes - exposure is mainly tier selection + any minute overages; not per-minute at base.
Additional watchouts
Flat $1,000+/mo entry is steep for SMBs; trust signals immature (few reviews, billing complaints, voice-quality concerns); validate with a trial before annualizing.
Sales call required
Mixed (some tiers require a call)
Commercial notes
Vancouver; founded 2024 (by Wishpond); 150+ customers, ~$1.8M ARR; targets SMB/services verticals (HVAC, solar, roofing, etc.); mixed third-party ratings (G2 ~2.7/5 on few reviews; Trustpilot billing/quality complaints); competes with other AI voice-sales agents
Key ambiguities
Official page advertises 'from $1,000/mo' but routes to a demo for exact tiers; third-party tiers ($990-$2,500/mo) may vary by package/negotiation; call-quality/latency not published.
Cancellation / refund
Cancel-anytime advertised; free trial offered, but multiple reviews report unexpected post-trial charges/auto-renewal - verify terms.
Support SLA / resale
No-code agent builder; 200+ integrations; SOC 2 Type I (Type II in progress); hands-on onboarding (per reviews)
Missing data
Exact tiers are demo-gated; third-party directories document ~$990-$2,500/mo flat (entry ~$990-$1,000/mo with ~2,000 included minutes). Call-quality/latency not published. Multiple reviews report billing friction (post-trial charges, auto-renewal, unclear cancellation) - confirm trial and cancellation terms in writing.
Related vendors
- 11x — AI sales workers (Alice for outbound, Jordan for inbound) that…
- 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 SalesCloser.ai
The closest documented capability profiles to SalesCloser.ai 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.
- AutoRFP.ai7.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs
- Inventive AI7.0 / 14Fuller documented coverage on Prebuilt Agents, Templates & Packs
- Swiftly8.0 / 14Fuller documented coverage on Memory & State Persistence and Triggers & Channel Coverage
- TextYess8.0 / 14Adds documented Deployment & Data Residency
- Actively AI8.5 / 14Fuller documented coverage on Observability & Auditability and Memory & State Persistence
- AutogenAI6.5 / 14Adds documented Model Flexibility & Routing
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded