The Science of Scaling · 21 May 2025
GTM in the AI Era w/ Andy Shorkey (CRO, Writer)
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Host Mark Roberge interviews Andy Shorkey, CRO of Writer, an enterprise full-stack generative AI platform. They cover how Writer grew from a brand-voice and style-guide tool sold to marketing and UX teams into a platform sold to two executive sponsors: the line of business and the CIO or head of AI. Shorkey also describes how Writer uses its own agents across go-to-market, including meeting prep, closed-lost analysis, SDR support and NDA review. Mark adds commentary on ICP tiers, Eric Ries's pivot types, and the moats native AI startups need against both incumbents and foundation models. The central argument is that enterprise AI sales now require a more complex, product-forward and technical motion. That motion is built around mission-critical use cases with defended ROI, and Mark suggests that proven enterprise implementations may be the moat for breakout native AI companies.
For founders
- Mark Roberge says native AI startups must assess their moat against two threats at once: incumbents building the feature in, and foundation models integrating up. He says probably more than half of the AI startups that raised over $100M and went to zero died because the foundation models integrated up.
- Mark Roberge has long told founders to start with smaller customers because enterprise deals take too long and drain cash. He now sees a pattern among breakout native AI companies where proven success in large enterprises becomes the moat, so going to the enterprise sooner may be the answer.
- Mark Roberge frames Writer's move from a single style-guide use case to a multi-use-case platform as a 'zoom-out' pivot, one of Eric Ries's four pivot types. He says such a pivot pulls in new buyers like the CIO and needs a different sales playbook and a different kind of salesperson.
- Andy Shorkey says Writer focuses on mission-critical use cases and won't sell software unless it can defend the ROI, helping customers build the business case both up front and on an ongoing basis.
- Mark Roberge recommends a three-tier ICP: green accounts you prospect proactively, yellow accounts you accept only when they come inbound and engaged, and a red tier you turn away even if they want to buy.
For revenue leaders
- Andy Shorkey says Writer's enterprise AI deals now have two executive sponsors. The line of business owns the outcome, while the CIO or head of AI standardizes platforms, so the playbook has to engage both.
- Andy Shorkey's pitch to CIOs is to crush their use-case backlog. He says every CIO Writer talks to has a massive backlog, and claims Writer delivers 50 to 100 agents a year where internal IT teams ship only a handful of apps.
- At Writer every rep demos the product, new hires are in the product in week one, and vertical-specific solution maps plus a library of use cases and demos make the pitch prescriptive.
- Writer uses its own agents across go-to-market: a meeting-prep app that generates the point of view, solution map and executive questions, plus SDR sidekicks, closed-lost agents, and an NDA agent that flags nonstandard terms so legal rarely touches NDAs.
- Mark Roberge's prep tactic is to 'do the meeting with the agent' beforehand, asking an AI about the person, the company, how they would evaluate you and your competitor, and how to sell against that competitor.
What was said 20, most useful first
Mark Roberge advises a three-tier ICP: proactively target perfect fits, accept engaged inbound from uncertain fits, and refuse poor fits even when they want to buy. Listen
Mark reacted to Andy's comment about not turning anyone away. He frames the ICP as having an engagement dimension: a green tier of A-plus fits you build email lists for and cold-outreach, and a yellow tier you aren't sure about but will sell to if they come inbound engaged. He says Andy's stance is probably a slight exaggeration and advises against accepting everyone. Some customer types will pull you down and add noise to your roadmap for a product you didn't build for them.
“This is my perfect green where I'm going to go and proactively sell to. This is my yellow where I'm not quite sure, but if they're engaged in inbound, I'll do it.”
Listen to the episode Strategy & market Link to this Report a problem
Mark Roberge recommends that non-technical go-to-market people spend an extra 25% of their time trying to automate parts of their own work with AI. Listen
Mark says people in sales, marketing, HR or finance should get down to the foundational tech level by playing with GPTs and AI apps at the configuration level. He suggests carving out an additional 25% of time to try automating something you do. The first few reps will take longer, but if you can cut the time for that task in half over the long term, he says that is huge. He cited Writer CEO May Habib's ability to go deep on LLMs and the stack as a selling founder as an example of this technical depth.
“They carve out an additional 25% of time to see if you can automate something that you do. It's going to take more time, the first few reps.”
A growing share of Writer's first conversations now start with a CIO or head of AI, pitched on crushing the IT backlog of AI use cases. Listen
Andy Shorkey says the entry point doesn't really matter, but a higher percentage of initial conversations are now with senior IT executives, because large enterprises are trying to build AI themselves. Writer's pitch is that as a full-stack platform it abstracts away the technical complexity and can deliver 10x to 20x the agents that internal teams can. He says Writer delivers 50 to 100 agents over a year, while internal IT teams drip out a handful of apps that may or may not be adopted.
“There's not a CIO we talked to that is not staring at a massive backlog. I can help you crush that backlog”
Listen to the episode Positioning & marketing Link to this Report a problem
Native AI startups must defend against both incumbents and foundation models, and that foundation models integrating up probably killed more than half of the failed $100M+ AI startups. Listen
Mark says that besides the usual risk of legacy incumbents building the product into their platforms, native AI companies face a new risk from foundation models expanding up the stack. He describes a growing graveyard of AI startups that raised over $100 million and are now at zero. He gives 'probably more than half' as his estimate of how many failed for this reason. He says investors and job seekers must analyze the long-term moat against both pressures.
“there's already a growing graveyard of native AI startups that raised over $100 million and they're now at zero. And for probably more than half, it was because the foundational models integrated up.”
Listen to the episode Strategy & market Link to this Report a problem
Mark Roberge, who has long warned founders against going enterprise too early, now suggests native AI companies may need to reach the enterprise sooner. Listen
Mark says founders traditionally jumped to the enterprise too soon. They assumed a big logo would bring others, underestimated how long a million-dollar deal with Goldman Sachs takes, and ran out of cash first. He pushed them toward smaller companies, partly because you can't walk into Goldman Sachs with a buggy first product. He says that is still somewhat true, but that 'maybe' going to the enterprise sooner is the answer in AI, a pattern he sees at Writer.
“But eventually you go to the enterprise, maybe you go there sooner, and that becomes the answer.”
Listen to the episode Strategy & market Link to this Report a problem
Writer indexes its ICP on verticals even though it sells a horizontal platform, starting mid-market and moving up. Listen
Andy Shorkey says indexing the ICP on verticals is a little unusual for Writer because the product is a horizontal platform serving many verticals. Customers start mid-market and expand upward. Intuit and UHG were early customers, and Writer still runs a mid-market business that does well and moves with more velocity. He says he does not turn away any customer segment.
“We do index our ICP on verticals, which is a little bit unique.”
Listen to the episode Strategy & market Link to this Report a problem
Writer targets mission-critical use cases and won't sell software unless it can defend the ROI, helping customers frame the business case both initially and on an ongoing basis. Listen
Andy Shorkey says Writer deliberately gravitates toward mission-critical workloads rather than the long tail of smaller use cases in customers' backlogs, though he notes that 'river of nickels' does add up. Writer helps customers frame the business case at the start and keeps defending the value delivered over time. He credits this focus on value and customer success for customers promoting Writer to others.
“We go after Mission Critical workloads and use cases. And we are not selling a lick of software unless we are defending the ROI and the value.”
Listen to the episode Sales process & deals Link to this Report a problem
The agentic AI wave has led executives to declare their companies AI-first, which eases the need to prove ROI for every single use case. Listen
Shorkey says agents have made it much easier for executives to see the scale of the unlock than earlier productivity tools like copilots and ChatGPT did. As a result, more executives say they are an AI-first organization that will aggressively embed AI. He says this organizational commitment alleviates 'a little bit' of the burden of locking down an ROI for each use case.
“so that alleviates a little bit of the burden on like you know, locking down an ROI for every single use case, because organizationally, they're adopting this”
Enterprise AI platform deals at Writer have two executive sponsors: the line of business owns the outcome, and the CIO or head of AI owns platform standardization. Listen
Writer originally sold mainly to line-of-business buyers such as marketing and UX teams. As demand for AI surged, CIOs and heads of AI became more prevalent sponsors because they had to get their arms around all the tooling options. Andy Shorkey says Writer evolved its playbook to engage CIOs, heads of AI and other technical stakeholders directly, while continuing to drive outcomes for the line of business, which could be the CRO, CFO or CMO.
“ultimately in our world, we have two executive sponsors.”
Listen to the episode Sales process & deals Link to this Report a problem
Andy Shorkey has seen open hostility between CMOs and CIOs over who controls AI platforms, and Writer argues that AI transformation requires them to collaborate. Listen
In some large enterprises the CMO has a lot of budget and power over the martech stack, and Writer enters with strong sponsorship there. As AI platform ownership shifts toward a centralized CIO, Shorkey says he has seen 'absolute disdain' between the two, partly because IT has historically been seen as slow or a bottleneck. Writer's position is that AI transformation won't accelerate unless business and IT collaborate tightly. He says the relationship between the two varies dramatically by organization.
“I've seen front and center that there's absolute disdain between the CMO and CIO in terms of the battle over those platforms.”
Listen to the episode Sales process & deals Link to this Report a problem
A 'zoom-out' pivot, where one narrow use case succeeds and the company expands around it, is a positive signal, but it requires a new sales playbook and a different kind of seller. Listen
Mark uses Eric Ries's four pivot types from Lean Startup: zoom-in, zoom-out, customer segment and customer need. He says the last two can be harder to navigate. He describes Writer's move from a simple, probably cheaper style-guide product sold to a head of branding into sales, finance and HR use cases as a zoom-out pivot. That expansion brings the CIO into the buying group, so companies must adapt both the playbook and the salesperson profile to the more complex decision-making unit.
“So we have to adapt our playbook and our salesperson to this more complex DMU in this zoom-out pivot.”
Listen to the episode Strategy & market Link to this Report a problem
Writer hires sellers for versatility and the ability to inspire and evangelize, because buyers aren't replacing old tech and need to be led. Listen
Shorkey calls Writer's motion typically the most complex selling motion its sellers will see, even compared with his own background in complex platform selling. The motion spans line of business and IT, and is use-case and vertical oriented. Because the product is transformational and isn't displacing legacy tools, he says sellers must lead the customer. He expects the motion to become more technical while staying grounded in business outcomes.
“This will typically be the most complex selling motion that our seller will see.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Every seller at Writer demos the product, and new hires are put into the product in their first week. Listen
Andy Shorkey describes Writer as very product-forward. Reps demo hands-on rather than relying on someone else to run the demo. He notes this is a shift for some enterprise sellers, who are used to bringing a solutions engineer to every call, and says that's not the case at Writer. Writer supports this with solution maps for each vertical it serves and a library of use cases and demos, so reps can be prescriptive. Shorkey says the demo itself shows the platform's ease of use and power.
“Everyone demos at Writer. We are very product forward. We've got a bunch of new hires here in New York, and we get them dipped into the product like week one.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Proven success implementing AI in large enterprises is becoming the moat for breakout native AI startups, because later buyers follow the reference rather than a cheaper copycat. Listen
Mark says the few breakout native AI companies share a pattern of proven enterprise implementations, which he calls 'no easy task' given IT, security, legal and change-resistant employees. Once a vendor succeeds at a name brand like Goldman Sachs, he argues word gets around, and a copycat with the same story at half the price won't win the next bank. He says he is still trying to frame this idea. Andy Shorkey responded only that Writer's customers advocate for it because of the value delivered.
“Bank number two doesn't care. They're just like, they made it happen at Goldman, just go with them.”
Listen to the episode Strategy & market Link to this Report a problem
Writer reps prepare for first meetings with an internal agent that generates the point of view, a solution map and role-specific executive questions. Listen
Andy Shorkey says a rep tells the app which account they are meeting, using the example of a sporting goods company. The app then lays out a point of view on the customer, the customer's orientation toward AI, and five areas mapped to Writer's capabilities. It produces a solution map, the point of view, and a script or question set tailored to each executive's role. He says it essentially gives the rep the first call.
“And so it essentially would create a solution map. It would create the point of view. It would also create the script or the questions for the executives, depending on their role and responsibility.”
Mark Roberge recommends preparing for sales meetings by running the meeting with an AI agent first. Listen
Mark suggests asking an AI the same open-ended questions you would ask in the meeting: about the person, their company, and their organizational and personal needs. He also suggests asking how they would evaluate your company and your competitor, how to sell your product against the competitor, and how the competitor would sell against you. He says this makes reps far more prepared than was possible before AI.
“I think just coming back to like, what are you going to do in the meeting and just do it now?”
Writer runs NDAs through an agent that flags nonstandard terms, so its legal team rarely touches an NDA. Listen
Andy Shorkey says Writer handles a high volume of NDAs. Incoming NDAs run through an internal app that flags anything nonstandard, and within the governance set around the NDA template reps simply 'turn and burn' them. AI also gives the legal team a jumpstart on contract reviews by quickly assessing key standard terms. He says this lets a relatively small legal team move fast enough that new hires are surprised by contract turnaround.
“our legal team rarely will touch an NDA because it'll run through the app and it will flag anything that is nonstandard”
Writer uses its own platform across go-to-market, including SDR sidekicks, closed-lost agents, enablement design, and account plans and points of view. Listen
Andy Shorkey says reps build many apps organically. The enablement team uses the platform to ideate, run needs analysis and create the right enablement materials for each team. Writer also runs SDR sidekick agents and closed-lost agents. He describes the goal as go-to-market productivity beyond individual productivity: giving AEs time back on work like developing account plans and points of view.
“we've got SDR sidekicks, we've got closed lost, you know, agents”
This quote could not be matched to the transcript. Treat it as a paraphrase.
Writer prioritizes prospecting toward ICP accounts that show signals such as website activity and LinkedIn engagement from its typical buyer personas. Listen
Asked whether AI has changed targeting, Andy Shorkey says Writer layers its own applications and workflows on top of the AI built into its existing go-to-market stack to find warm signals. He gives the team high marks for prioritizing ICP accounts with activity on the website and for tying into LinkedIn to find personas in the roles that typically gravitate toward Writer.
“going after ICP where there's activity where there's signal around website.”
Listen to the episode Pipeline & demand generation Link to this Report a problem
Writer's product grew from a brand-voice and style-guide tool into a knowledge-graph platform connected to enterprise data, which widened its use cases across functions. Listen
Andy Shorkey says early customers used Writer for brand-voice consistency and compliance, mostly in marketing and UX teams generating a lot of content. That evolved into Writer's knowledge graph, which connects into an organization's enterprise data. That in turn opened use cases such as content supply chain, market commentary in financial services, pipeline-generation kits and account-based apps for AEs, and legal and brand-compliance review in CPG and retail.
“that quickly evolved into our knowledge graph, which obviously is is essentially the connective tissue that allows us to connect into the enterprise data for an organization.”