Ross Keating

The problem is the handoff between tools, not simply the number of tabs.

Keating asks readers to count the tabs open during a live deal: CRM, call recorder, email, Slack or Teams, a deal room, ChatGPT and private notes.

• Reps carry information from call to CRM to follow-up and next week’s forecast by hand.
• They call this the execution gap.

Read original

In the Overpath launch post, Molly is described as helping with deal execution, CRM hygiene and coaching, while leaders get signals about deal attention and sales behaviours.

Keating says Molly can bring a brief before a call, surface a MEDDPICC gap afterward and prepare a CRM update for approval. Reps stay in an existing channel while the underlying systems continue to run.

Read original · Original 2

The distinction is whether agents have enough organisational context to act and provide useful insight—not whether staff have access to an AI tool.

In a post about AI maturity, Keating says a Claude licence is not enough. The post contrasts leaders treating CRM as a data hub and considering proactive agents with teams where missed follow-ups lose deals and MEDDICC comes up once a week in pipeline meetings.

Read original

The idea connects the execution-gap argument to the claim that CRM records alone do not make salespeople more effective.

Molly is described as preparing a pre-call brief, identifying a MEDDPICC gap after the call and drafting a CRM update. Keating repeats the line “No CRM ever made a salesperson sell better” in both the Overpath launch post and the post on AI maturity.

Read original · Original 2 · Original 3

Start here for the sharpest before-and-after picture of the execution gap and how an AI teammate is meant to close it.

Ross Keating asks readers to open a rep’s screen during a live deal and count the tabs: CRM, call recorder, Gmail, Slack or Teams, a deal room, ChatGPT and a private notes doc.

• Each tool works on its own, but the rep becomes the integration layer, carrying context by hand from call to CRM, into follow-up and into next week’s forecast.
• They call that manual handoff the execution gap and say switching tabs does not improve execution.
• The alternative is a revenue execution layer where the rep stays in the channel already open while Molly brings a brief before the call, surfaces the MEDDPICC gap after it, and prepares the CRM update for one-click approval.

For brands planning a campaign

For a sales-enablement company building an AI sales coach, this is the clearest pain statement to build a talk around: buyers do not need another dashboard; they need automatic context transfer across the tools reps already use.

Start here for the product thesis, the positioning of Molly and the claim that traditional software interfaces are being replaced.

Ross Keating announces Overpath and introduces Molly, an AI teammate built from what they describe as decades of pain around keeping CRMs clean, making training stick and knowing where to improve a sales motion.

• Molly is positioned as another member of the sales organisation: she works proactively on deal execution and CRM hygiene, coaches when a deal needs it, and shows leaders which behaviours actually drive revenue.
• Keating says traditional software interfaces are dead and that Molly is accessible through Slack, Teams, WhatsApp, web and voice.
• The recurring line is blunt: “No CRM ever made a salesperson sell better.
• Molly does.”

For brands planning a campaign

This post shows Ross Keating is not only commenting on the market but building and launching an AI sales agent product. A campaign talk should move past theory and address practical architecture, integration and adoption choices.

Start here for the adoption spectrum, the warning about falling behind, and the argument that context is the missing layer.

Ross Keating contrasts two kinds of sales leaders.

• Some have AI-native solutions in place, treat the CRM as a data hub and think about agents doing proactive work and supplying insight.
• Others still centre their world on HubSpot or Salesforce, update manually, lose deals when a rep fails to follow up, and talk about MEDDICC once a week in pipeline meetings.
• They say “context” is still an engineering buzzword many have not grasped, that a Claude licence alone is not enough, and that every month without building context and agentic capability means losing ground that will be exposed fast.

For brands planning a campaign

For an AI sales coach brief, this post supplies the change-management argument: the buyer’s problem is not access to AI tools but the gap between a licence and a working context layer with agentic follow-through.

Each tool works on its own, but the rep becomes the integration layer, carrying context by hand from the call into the CRM, into the follow-up and into next week’s forecast. They call that manual handoff the execution gap and argue that switching tabs does not improve execution—it just creates more places for methodology, context and next steps to get lost.

A concrete snapshot argument: it turns a familiar scene—the tabs on a rep’s screen—into a named problem rather than asserting a trend.

For brands planning a campaign

A sales-enablement or AI-coaching vendor can build a talk around removing handoffs between existing systems, not adding another dashboard. The “count the tabs” framing is a ready-made opening for a live demo.

Examples: Source 1 · Source 2

Her jobs are proactive deal execution, CRM hygiene and coaching that arrives exactly when a deal needs it.

• For leaders, they say, the same agent shows which behaviours are driving revenue, where deals need attention and where coaching should focus.
• The product is framed around work that previously fell to the rep between calls and follow-ups.

For brands planning a campaign

Positioning material should treat the agent as a colleague with a job description, not a feature list. A B2B SaaS buyer brief can ask which specific behaviours the agent measures and which coaching triggers it uses.

Examples: Source 1 · Source 2

Keating’s earlier post makes the same point through a rep’s day-to-day work: the rep stays in the channel already open while Molly brings pre-call preparation, surfaces the MEDDPICC gap after the call and prepares the CRM update for approval in one click, as the underlying systems keep running.

Contrasts an old interface model with an embedded one, repeating the idea that “systems run underneath” in both posts.

For brands planning a campaign

For a sales-enablement company building an AI coach, this is a direct argument for embedding coaching in collaboration tools instead of creating a new portal. The specific channel list can structure a product comparison or interview question.

Examples: Source 1 · Source 2

They warn that a Claude licence is not enough, that every month without context and agentic capability is lost ground, and that the gap will be exposed quickly.

A binary contrast and forward-looking warning use two observed groups to make a maturity argument rather than a product pitch.

For brands planning a campaign

This is the strongest hook for a conference talk on closing the execution gap: a maturity framework brands can self-assess against, with “the context layer” as the named concept. It also prompts a useful question: what counts as context, and who owns it?

Examples: Source 1

Molly produces a brief before the call, flags the MEDDPICC gap after it and drafts the CRM update for approval.

• Keating says deals are lost when reps do not follow up manually, and that manual updates leave the CRM unclean.
• In this account, coaching is triggered by the state of the deal.

For brands planning a campaign

A vendor can assess Keating as a source on making methodology stick: training decays unless it is delivered against live deal data. This could anchor a talk segment on why enablement content alone does not change rep behaviour.

Examples: Source 1 · Source 2 · Source 3

In the launch post, it sets up Molly as the thing that does; in the later post, it closes a wider argument about context and agents. The line captures Keating’s consistent position that tooling alone does not change selling behaviour.

For brands planning a campaign

A memorable, quotable claim that a brand can build a campaign or talk title around. It also marks the limit of Keating’s argument: systems record and support, while people and agents change outcomes.

Examples: Source 1 · Source 2

They write as part of the team launching it and describe Molly as the product of "decades of pain that Dermot, the team, and I have lived through".

• The posts do not say whether they own Overpath or hold an employed role there; ask them to clarify their relationship with the company.
• They frame Overpath as a "revenue execution layer" that briefs reps before calls, surfaces MEDDPICC gaps afterwards and prepares CRM updates for approval.
• They argue that reps currently act as the integration layer between CRM, call recorders, email, chat and AI tools.

View example

Ask Keating to confirm whether they have any investment, advisory or commercial relationship with Elkstone.

View example

Ask Keating to confirm whether they have any investment, advisory or commercial relationship with Sure Valley Ventures.

View example

The posts cover routing a question to the right agent, keeping agents away from CRM data they should not touch, and leaving a readable trail of what the agent did. One post makes the same tabs-and-handoffs argument and names Ross and Dermot among the team building it.

Jain writes from the build side, explaining what the other 80% of an AI product consists of once the model choice stops mattering.

• Keating writes from the go-to-market side, arguing about process, coaching and what changes for a rep or sales leader.
• Readers who want engineering trade-offs get more from Jain; those who want the sales argument get more from Keating.

Read original · Original 2

Doyon's posts report findings from scoring thousands of calls with an AI coach, including which MEDDICC categories reps cover thinly and why Decision Process scores lowest. That evidence supports Keating's argument for briefing before the call and identifying MEDDPICC gaps afterward.

Doyon argues from large call datasets, drawing out patterns across many sellers and geographies.

• Keating argues from a product position and the friction of a rep's working day.
• Doyon also explores how customer success and sales roles have merged, while Keating stays closer to the execution layer and the tools that sit on it.

Read original · Original 2

Tveter walks through the HubSpot sales plugin for Claude and the Deal Progression beta skill by skill, flags which settings an admin has to switch on, and notes where recordings are required. The practical detail matches Keating's claim that AI belongs inside the existing channel, not behind a new interface.

Tveter writes as a hands-on user reviewing what shipped, including the setup steps and awkward parts a team hits on day one.

• Keating writes as someone launching and positioning a product, explaining the problem and intended behaviour rather than the configuration.
• Read Tveter for what a tool actually does; read Keating for why the sales process should change.

Read original · Original 2

Jain's posts argue an AI product is about 20% model, and that what users actually notice is agent routing (a planner and coordinator deciding which skill answers a question), guardrails that check before an agent touches pipeline data, and an audit trail a rep can read back after the fact.

• That is directly useful for a sales-enablement company deciding what its coach must build: which component answers a rep's question, what stops an agent reaching records it should not, and how a rep or manager can reconstruct what the coach did.
• The same posts also show awareness of the rep-facing problem — one counts the tabs on a rep's screen and describes a brief before the call, a qualification gap surfaced after it, and a CRM update ready to approve in one click.
• One post credits Ross and Dermot among the people who argue over those product decisions, so Jain is close to the same subject matter rather than an outside observer.

The tradeoff is framing and altitude.

• Jain's posts read as product-engineering argument for people building the agent, with less on the sales-leader questions at the centre: how coaching changes rep behaviour, how a mid-market team should run the workflow, and how a buyer measures whether execution improved.
• The posts don't provide audience or commercial information, or published speaking and sponsorship terms.
• Ask Jain's company to confirm whether they can speak to a buyer audience rather than a build audience, and what terms are available.

Examples: Source 1 · Source 2

Doyon's posts describe scoring roughly 500,000 Aircall calls and 500,000 video calls, and separately more than 11,000 MEDDICC calls across geographies, languages and company sizes.

• The findings are specific enough to carry a talk: MEDDICC rarely happens in one conversation and usually covers only two or three categories at first; reps often tick a framework box after hearing something as thin as a desire to improve CAC, while stronger sellers push for baselines, financial impact and ownership of the metric; and Decision Process scores lowest, with procurement requirements, legal timelines and approval paths surfacing late and stalling deals.
• For a sales-enablement buyer, that is a ready-made case for measurement and for the specific coaching an AI sales coach should deliver, and it points at the paper-process gap without needing product claims.

The tradeoff is that this evidence is about qualification and call quality, not about the handoffs between tools, agent design or CRM hygiene.

• A coach built on Doyon's material would be judged on conversation quality and pipeline rigour; a company that wants to talk about context moving between CRM, call recorder and forecast would still need Keating or someone closer to that argument.
• The posts are also the oldest reviewed here, published in February and March 2026, and they describe Doyon's own analysis rather than an independent study.
• Ask how the method transfers to mid-market deals and whether the scoring model is available.

Examples: Source 1 · Source 2

Tveter's posts walk through specific HubSpot releases — a Claude sales plugin with eight pre-built skills and a Deal Progression public beta — and explain what each one is for: a ten-minute call prep, a morning brief of open deals and stalled contacts, logging a call in plain language, drafting follow-up from deal history, a pre-review pipeline scan, contact lookup, imports.

• Tveter flags what teams get wrong: every deal record gains a Catch-Up tab and a default layout rename that reps should be told about, recording must be switched on for any of it to work, methodology configuration (MEDDIC, BANT, SPICED) and Deal Plans each need an admin to activate, and writes to the account are shown for confirmation first.
• There are also concrete usage opinions, such as starting reps on call logging and follow-up.
• For a sales-enablement company, this is the change-management and enablement layer that sits between buying an AI coach and reps actually using it.

The tradeoff is breadth and vendor dependence.

• The commentary is tied to HubSpot features and to one person's daily use, so it does not address buyers on other CRM platforms, the design of an AI agent, or the multi-tool execution gap Keating describes.
• It is also recent, September and October 2026, so there is little record to judge how consistent the analysis is over time.
• Ask what audience data, sponsorship formats, rates and speaking terms are available.

Examples: Source 1 · Source 2