Pepsales AI

Almost nothing survives removal of the model layer. Strip it and the buyer loses demo personalisation and generated demonstration data, which is the product's founding purpose, along with prepared discovery questions, real time in call prompting, objection handling, methodology scoring updated live, discovery and demonstration scores, extracted action items, buyer pain identification, deal scoring, the aggregated dashboard of what buyers said about pricing and competitors, and the natural language search across every opportunity.

What remains is a meeting recorder with transcription and storage and a synchronisation to the system of record. The vendor's own product naming agrees: the entry product is a meeting assistant, and every tier above it is priced on intelligence applied to what that assistant captured.

The product acts continuously and no oversight machinery was retrievable. It prompts a representative live during a customer conversation, writes notes into the system of record automatically, and assigns scores to each meeting, to the discovery, to the demonstration and to the representative who ran them. No approval step, role model, permission structure, audit trail or guardrail was located.

The unresolved question is the one this category keeps producing and it is sharper here than most, because scoring is the product rather than a byproduct: machine generated scores are attached to named employees, aggregated into coaching recommendations and rep performance views for leadership, and nothing published states whether a representative can see or contest a score of their own conversation, whether human review precedes its use in a performance discussion, or whether scoring has been tested for variation by accent or language. Note that the vendor's site refuses automated retrieval, so this row rests on indexed pages rather than direct reading.

No base model, provider, family, version or hosting arrangement was located on any retrievable page. The architecture is described by function in reasonable detail, covering what the system reads before a meeting, what it does during one and what it produces afterwards, and the vendor states that more than twenty sales methodologies and qualification frameworks are supported, which is a real specification of the reasoning scaffolding. None of it identifies what performs the reasoning. Transcription, summarisation, live prompting and scoring all imply provider dependencies that are named nowhere in the retrievable material.

Real named customers and no measured result behind any number. Four companies are named directly as customers, all of them recognisable business software firms, and one carries a published case study covering conversation intelligence, deal prioritisation and forecasting.

Several testimonials are attributed to roles including a startup advisor and former senior director at a named intent data company, and one records a competitive displacement, describing a switch away from a major conversation intelligence incumbent.

Against that, every quantitative claim is a vendor aggregate with no sample, period, denominator or method attached: an average fifty percent increase in demonstration to win conversion, a claim to improve win rates by half and double sales productivity, and an assertion that eighty percent of demonstrations fail to impress buyers. Not one of the four named customers has a published figure attached to it.

Structurally a weak fit for this axis, graded on what is knowable, with the context carried here. This platform runs no campaigns, connects no mailboxes for bulk sending and dials nobody; it works inside meetings the customer has already booked. The one surface that reaches a buyer directly is personalised follow up automation after a call, which is correspondence with a person already in conversation rather than cold solicitation. No statute, consent position or suppression mechanism was located, and none of the apparatus this axis normally measures would apply to what the vendor does.

No privacy policy, cookie policy or processing addendum was retrievable, and the reason matters for how this row should be read: the vendor's site refuses automated access, so direct verification was declined rather than attempted and failed. This is recorded as not retrieved rather than as established absence, and the row is held at the middle band accordingly.

The processing in scope is substantial and would need covering: the platform records, stores and transcribes customer meetings across three conferencing platforms with unlimited retention stated as a feature, captures video for replay and sharing, and synchronises extracted content into four named systems of record. Verifying the legal surface directly is the highest value check available on this vendor.

The baseline is clean and one layer sits outside it. The material the platform works is the customer's own: their recorded conversations, their pipeline records and their product knowledge, which removes the licensing question that dominates this axis elsewhere, and the vendor sells no contact database.

What is unsourced is the research layer, since the platform is stated to produce buyer research and an account recap before a meeting takes place, and nothing published names where that external context comes from, what it may collect about a named individual or their employer, or on what basis. The generated demonstration data the product creates is synthetic by design, which is worth noting as a point where a competitor might instead expose real customer records.

The exposure is the recorder and it carries no stated position. Meetings are captured and analysed across three named conferencing platforms, which requires either a bot participant or a recording integration governed by each platform's own rules, and no conformance statement for any of them was located. Four systems of record connect by ordinary supported integration.

Nothing scrapes, rotates an identity, rents an account or ships a browser extension that lifts data from a platform it has no arrangement with, and the posture toward the surrounding stack is cooperative throughout, which is why this sits at the middle band rather than lower despite the silence.

Nothing retrievable states whether customer content trains or improves any model, whether prompts are retained by any provider, or whether material crosses between accounts, and no document in which such a commitment could sit was reachable because the site refuses automated access.

The exposure is large and specific rather than theoretical: the platform holds recorded video and full transcripts of every sales conversation a customer runs, retained without stated limit since unlimited transcription and storage is sold as a headline feature, plus each customer's own product knowledge and demonstration environment, plus the aggregated record of what buyers said about pricing and competitors. A vendor holding the recorded negotiations of multiple competing software companies has the cross client question in its sharpest form, and no answer to it was located.

This product records real customers and prospects on live calls and nothing retrievable describes how those people are told. No statement was located on whether a recorder announces itself, joins as a visible participant, follows the customer's consent policy, or handles the jurisdictions requiring consent from every party, and no position on article 50 of the European artificial intelligence regulation appears in any retrievable material.

Video recording with replay and sharing is sold as an entry tier feature, so the captured party's likeness and words circulate inside the customer's organisation afterwards. Follow up correspondence is generated automatically after the call with nothing stated about authorship disclosure. Held at the middle band rather than lower because the site refuses automated retrieval, so absence of a published position could not be established directly.

A modest, well chosen and entirely conventional set. Seven integrations are named across the retrievable material: three conferencing platforms covering the meetings the product exists to capture, and four systems of record covering where the output needs to land, including two mid market platforms that several competitors reserve for higher tiers. Synchronisation is described as automatic for summaries, insights and next steps rather than as a manual push.

Against that, no published interface, software development kit, webhooks, protocol server, marketplace, developer documentation or partner directory was located anywhere, so the platform connects to what the vendor has built connectors for and nothing else. For a product whose whole value is producing structured intelligence, the absence of any programmatic route out of it is the notable gap.

Nothing on hosting provider, country, region, data centre or residency option was retrievable, and no corporate address or registered entity was located either. The available signals are circumstantial and point in one direction: the founding investor is an Indian venture firm, the named customers are predominantly Indian headquartered business software companies, and the founder is named in the funding announcement.

That establishes a probable centre of gravity and answers none of the questions this axis asks about where recorded customer conversations physically sit, which matters for a product storing video and transcripts of commercial negotiations without a stated retention limit.

No certification, audit, attestation, penetration test, control set, trust centre, status page or vulnerability route was retrievable, and this row is explicitly recorded as not retrieved rather than as established absence. That distinction is doing real work here: elsewhere in this index the bottom band has been applied where a complete site footer enumerated a vendor's whole document set and contained no security surface, and no footer could be reached in this case because the vendor's site declines automated access.

What is certain is the sensitivity of what the platform holds, namely recorded video and transcripts of every customer conversation across an entire sales organisation, retained without a stated limit. Confirming whether a security or trust page exists is the second priority verification on this vendor after the legal surface.

Real per seat prices are published for the products a buyer starts with, and two independent pricing directories agree on the figures. The meeting assistant carries a list rate of nineteen dollars per user per month with an annual figure of one hundred and eighty two dollars, and the conversation intelligence product a list rate of fifty nine dollars per user per month with an annual figure of five hundred and sixty seven, each accompanied by a published feature list detailing what the tier includes.

The vendor states plainly that all plans are billed annually and that the displayed rate is per user per month, which removes the most common ambiguity on a page of this kind. A free trial requires no card. Off the top band because the demonstration automation product that is the vendor's founding differentiator does not appear in the retrievable price list at all, so the capability the whole platform is sold on has no published rate, and because no seat minimum, usage ceiling, storage limit or overage behaviour was located against any tier.

One structural strength and no published position. Extracted summaries, insights, action items and next steps synchronise automatically into the customer's own system of record, so the durable output of every meeting lands in a database the buyer already governs rather than only in the vendor's store, which materially lowers the practical cost of leaving and is the same reasoning credited elsewhere in this index.

Against it, no export function, file format, bulk retrieval route, post termination window, deletion timeline or retention period was located. What would stay behind is substantial: recorded video and full transcripts of every customer conversation, scoring histories against named representatives, and the configured demonstration environments. Unlimited storage is sold as a feature, which tells a customer their material is kept and nothing about their right to take it.

Structurally a weak fit for this axis, graded on what is knowable, with the context carried here. The platform operates no sending infrastructure, runs no campaigns and connects no mailboxes for bulk outbound. Its only outbound surface is personalised follow up correspondence generated after a meeting and sent to somebody the representative has just spoken to, which raises no sender reputation question of the kind this axis measures. No warmup, authentication, pacing, bounce handling or complaint threshold is published because none of that apparatus applies here, and nothing in the retrievable material suggests poor practice.

The buyer is drawn precisely and the market sizing is argued rather than asserted. Four separate audience pages address presales and solutions leaders, individual solutions consultants, sales teams and sales leaders, which correctly recognises that in technical selling the person who buys the tool, the person who uses it daily and the person who reads its output are three different roles with three different problems.

The target is stated as business software companies selling to other businesses, and the vendor sizes its market by naming the annual sales and marketing spend of the enterprise application software sector rather than by claiming a share of it. Four named customers corroborate the segment. Off the top band because no region, country, language, industry breakdown, seat band or customer count is published anywhere retrievable, and there is no statement of who this is wrong for.