# Molly Devin

Molly Devin writes about sales execution, RevOps, and how AI agents could help sales teams do complex deal work. Her posts draw on sales-leadership questions and Devin's early weeks at Fluint.

- Profile: https://b2binfluencer.marketing/creators/molly-devin/
- LinkedIn: https://www.linkedin.com/in/mollykatedevin
- LinkedIn followers: 2,390
- Topics: artificial-intelligence, sales, revenue-operations
- Based on 14 dated posts, latest 2026-10-01

## Summary

Useful for a discussion-led LinkedIn series on sales execution and AI agents for sales and RevOps audiences.

- Best for: B2B SaaS selling sales-AI or RevOps to sales leaders, enablement, and RevOps buyers.
- Not for: Consumer or high-reach paid sponsorship with rate card, impressions, video, demos, vendor comparisons.
- Check first: Confirm available formats, rates, terms, inquiry route, and recent reach and engagement results.

### Takeaways

- AI can write the business case but cannot read the deal.
- RevOps rollouts fail in a layer nobody owns.
- Agents that do not remember or coordinate are the bottleneck, not agent count.

## Main ideas

- **Knowledge is not judgment: AI can write the business case but cannot read the deal**: Devin's clearest position on sales AI: a model knows how to structure a business case — its sections, format, and general argument — but not which buyer needs a one-pager, that speed beats thoroughness because the champion presents to the CFO tomorrow, or that this deal calls for a risk-based frame against a competitor who leads with ROI. They distinguish knowledge — knowing what to do — from judgment: knowing what matters right now, based on patterns and outcomes the best reps carry without being able to explain them. That reframes the purchase: you are not buying documents; you are buying the judgment that produces them.
- **The recurring question: why does mid-funnel work not happen**: Sellers drawing $120K+ base salaries know they should write business cases, build account plans, and multithread into new stakeholders. But the work does not happen consistently, deals slip at stage 3, and pipeline dies. Devin lays out the problem and asks sales, enablement, and RevOps leaders whether the cause is people, process, capacity, or leadership — whether reps do not know what to do, or know but lack the time to execute. The post aims to collect answers, not deliver one.
- **RevOps rollouts fail in a layer nobody owns**: Devin's hypothesis: when a RevOps rollout fails, it is almost never the plan and almost never the rep. It is the space in between — a layer nobody owns, hires for, or names, where good strategy quietly fails to happen. They offer the idea as a theory to falsify, with a request for the strongest counterexample rather than agreement.
- **Agents that do not remember or coordinate are the bottleneck, not the number of agents**: Devin recounts a call with a former teammate now leading RevOps at a Series C company. The team had built 30-plus agents using Claude, custom GPTs, prompt libraries, and stitched-together agentic workflows. The test: does context survive between them? If the research agent learns the CFO caps projections at 8%, does the business case agent know? If a rep corrects the email agent once, does that correction carry forward? The answer was no: the agents are separate. Devin identifies memory and coordination as the genuinely hard problem and asks readers how they are approaching it.
- **Build versus buy: pricing the homegrown agent stack**: Devin walks through what "we'll just build our own agent" actually requires: 30-plus specialized agents covering research, discovery, business case writing, multithreading, executive messaging, and deal scoring; orchestration so they talk to each other; training on your methodology, case studies, and closed-won patterns; data storage linking CRM, transcripts, and documents; an ML model rather than only an LLM, to learn from your deal outcomes; and ongoing maintenance as deals change. They estimate 6 to 12-plus months of engineering, then contrast that with hiring one agent already trained on 150,000+ complex B2B cycles and live the same week. The frame is tool versus hire: a tool makes sellers faster at work they are already not doing; a hire does the work. The cycle figure and timeline are Devin's own, and the post argues a product position rather than an independent comparison.
- **RevOps is judged on finding efficient revenue, not on cleaner dashboards**: Devin argues that RevOps is judged not on cleaner dashboards but on whether it can find the next efficient dollar of revenue. They cite CFO attention to CAC payback and headcount efficiency, CEO appetite for ARPU growth without another pricing battle, and boards fixed on retention and net revenue retention. With payouts, FX, and issuing tapped out and margins flat, workforce spend leaking outside the organization's ecosystem becomes the unfilled lever. Devin asks FinTech RevOps leaders which lever would change how they are seen heading into 2026.
- **Ramp method: study closed-won deals before writing new plays**: Instead of product training and role plays, Devin spent a first week pulling every closed-won deal from the past 12 months and looking for patterns: what buyers actually cared about, what language champions used to sell internally, where deals accelerated or stalled, and what content moved deals forward rather than creating noise. The takeaways: the best sellers execute the same three or four moves consistently; buyers use different words than vendors do; and the data makes gaps obvious. Devin rules out building new frameworks from scratch, guessing at messaging, or waiting to feel ready, and offers to share the documents being built.

## Work to read

- [Knowledge is not judgment: what an AI-written business case gets wrong](https://www.linkedin.com/posts/mollykatedevin_if-youre-a-sales-leader-watching-your-team-activity-7445520047445331968-gYx9)
- [The layer nobody owns: where a RevOps rollout quietly fails](https://www.linkedin.com/posts/mollykatedevin_theory-im-testing-id-love-to-be-wrong-activity-7457602292670754817-rbcc)
- [Thirty agents that do not talk to each other](https://www.linkedin.com/posts/mollykatedevin_week-3-at-fluint-on-a-flight-to-denver-activity-7434296469358313472-LSzf)
- [Build versus buy: 30 agents or one trained hire](https://www.linkedin.com/posts/mollykatedevin_week-4-update-at-fluint-because-i-know-activity-7437488749753556993-6iYs)
- [RevOps is judged on revenue, not cleaner dashboards](https://www.linkedin.com/posts/mollykatedevin_revenueefficiency-customerretention-marginpressure-activity-7374075217737949184-hw6i)
- [Mid-funnel work that is not happening: people problem or capacity problem?](https://www.linkedin.com/posts/mollykatedevin_ive-been-thinking-a-lot-about-something-activity-7429521514904842240-VFsS)
- [Day three on the job: studying closed-won deals instead of role plays](https://www.linkedin.com/posts/mollykatedevin_official-day-3-how-im-ramping-without-activity-7430265913490333696-CazB)

## Brand relationships

- **Fluint** (employment)

## Partnerships

Molly Devin’s LinkedIn posts are organic commentary on sales execution, RevOps, AI agents, and their early weeks at Fluint. They contain no sponsored placements, affiliate links, or disclosed commercial terms, and publish no media kit, rate card, sponsorship policy, format list, or official inquiry route. There’s no package to price or schedule; brands should ask directly about available formats, rates, terms, and how to inquire.

## Alternatives

- Omprakash Karuppanan: Pick Karuppanan when the campaign has to show the mechanics of the fix, not just name the problem. The two posts make the same mid-funnel argument Devin's series would open with, but from inside a live ABM programme: five named AI skills for deal reporting, audience targeting, account research, field marketing, and stall diagnosis; the inputs each one reads, including CRM closed-lost and closed-won data, meeting transcripts, website engagement, ad accounts, and webinar Q&A; and one audited figure: 31 stalled accounts reviewed, with 17 showing a trigger event in the last 60 days that nobody acted on. If the deliverable is a usable artefact readers can copy into their own stack, these posts go further than a debate post, and the diagnosis matches Devin's: AI saturated the top of the funnel; the edge sits where deals quietly stall.
- Jose Aleman: Choose Aleman when the campaign targets revenue operations and planning buyers rather than AI tooling. The posts answer questions that audience actually argues about: why hiring ten reps does not add ten quotas to the forecast; why a five-month sales cycle makes a 90-day ramp assumption mathematically impossible; why doubling headcount without adding pipeline just slices the same pipe into smaller pieces; why a top rep handed a smaller patch reads success as punishment; and why boards can plan around a miss they saw coming in week three but not a surprise they learn about in week eleven. If the sponsor is selling forecast accuracy, capacity planning, or deal-risk discipline to a VP of Revenue Operations or a finance-adjacent buyer, this framing sits closer to the decision than an argument about what agents cannot do.
- Bryan Byler: Choose Byler when the campaign needs executive and partner-ecosystem credibility plus a vertical story. The two posts document a leadership track at a HubSpot partner, described in the author's own terms: sales org rebuilt, team scaled from 3 to 13, close rate up 15%, average deal size up five times, more than 200 deals closed, and partner standing moved to second in North America in 2023 and second globally in 2024. The second post states the position that move set up: HubSpot for healthcare, specialist partners over generalist consultancies, more than 150 completed projects, and three to five EMR integrations a month where top partners do three to five a year. For a campaign aimed at HubSpot partners, healthcare technology buyers, or growth leadership, that mix of operating record and explicit market thesis is a better fit than a peer discussion series. The posts assert outcomes rather than asking for counterexamples.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/