Roarr Catalyst Group

I keep being called by US executives planning Indian GCC builds. The questions they ask have not changed in a decade. The answers have shifted structurally over the last 24 months, and most build agreements signed in 2025 do not yet incorporate that shift. Three patterns keep showing up in these conversations. ✅ The center's purpose is decided downstream of its location. ✅ The Product or Process decision is implicitly settled by the staffing model, the SLA structure, and the technology stack. By the time anyone asks the question explicitly, the architectural commitment has been made by default. The captive head's decision rights over operating-model architecture are being designed out of the center at the build stage. The default templates most advisors use place the captive head in an advisory role on architecture decisions. ⚜️ That was defensible in 2018, when operating model changes were slow. The compression now underway moves too fast for that role to keep up. SLAs are being structured to measure the work agentic AI will compress within five years. Throughput-based metrics that seem reasonable today produce centers whose economics will no longer exist by 2028. The compression now running on Process GCCs is the same one. It is just running on a five-year timeline instead of a twenty-five-year one. ♻️ The full piece, with composite examples from recent build conversations and the framework I now use to evaluate these decisions, is in this edition of The Revenue Circle.

The 6,000+ Indian GCC layoffs in FY2025 are early signals of operating-model compression on roughly 1,300 of the 2,117 centers. Industry reports treat all 2,117 as a single category. That framing obscures a structural fact: India hosts two GCC industries, not one. Product GCCs build intellectual property code, designs, patents, and product roadmaps. JPMorgan Powai, HSBC Tech, Goldman Sachs, Eli Lilly, Walmart Global Tech, Micron, Mercedes R&D. They are expanding. Process GCCs run transactions. Claims processed, tickets resolved, reports generated, and reconciliations completed. Most BFSI back-office captives. Most healthcare operations. Most shared services. The FY2025 layoffs at Technicolor (3,000), Ford (1,000), and Fidelity (500); and the Wells Fargo Chennai shutdown announced for Q4 FY27 relocation are distributed almost entirely on this side. Four structural forces are operating on the Process GCC category simultaneously: architecture capture by consulting firms, unit-of-production compression by agentic AI, decision-rights drift within the captive structure, and maturity self-report distortion in industry surveys. The forecast: between 250 and 400 Process GCCs will cease standalone operation by 2031. That is 12 to 18 percent of the current base. They consolidate with vendors, get absorbed into shared service centers, or close. The full argument is in this edition of The Revenue Circle. #GCC #Revenue #Captive #outsourcing #sharedservices #ODC

Anthropic just passed OpenAI in annualized revenue. $30 billion versus $24 billion. The company with 5% of ChatGPT's consumer user base is now generating more enterprise revenue than the company with 900 million weekly users. ❓ So what does this actually tell us about enterprise AI in 2026? ❓ Why did the consumer giant lose the enterprise race to the company most CIOs had not heard of two years ago? ‼️ What does Anthropic understand about enterprise procurement that OpenAI is still figuring out? And while we are asking uncomfortable questions, where exactly does Google fit in this conversation? They have the distribution advantage nobody else can match. They are also shipping a product portfolio so fragmented that the buyer cannot tell what to purchase. Vertex AI, Gemini Enterprise, Workspace AI, AI Studio, NotebookLM Enterprise. Is #Google's biggest competitor actually #Google? ✅ Here is the question I have been sitting with for weeks. While these three companies compete for enterprise revenue at the foundation-model layer, what happens to the hundreds of AI startups selling products built on top of these models? IDC research shows that for every 33 AI POCs enterprises launch, only 4 reach production. RAND tracked $547 billion in AI spending that produced nothing measurable in 2025. ❗ Who exactly is buying enterprise AI products right now, and why is most of the money producing nothing? I wrote about this in the latest edition of The Revenue Circle. The enterprise AI market has split into three categories. ✅ Two of them are winning. ❌ The third is losing badly, and most of the companies inside it have not yet accepted it. If you are building an AI product, leading sales at a frontier lab, or evaluating AI investments, the article is worth 8 minutes of your time. #CEO #AI #Enterprise #Sales #Revenue #GTM #PMF #product #VC

I have been running forty enterprise AI demos for a mid-size technology company. ⛔ Ten of those went into proof-of-concept engagements with real budget behind them and genuine technical interest on the buyer side. Every single one of the ten pilots failed to convert into a production contract. ✅ The technology was not the problem. In almost every case, the product delivered what was promised, and the buyer acknowledged it. ⭕ The pilots failed because they were designed for evaluation and were not meant for production. The success criteria were so vague that everyone could claim victory without evidence to justify a production budget. Executive sponsors showed up at the kickoff and disappeared. It looks like the distance between the sandbox POC and the customer's actual infrastructure wasn't mapped until it was too late. 88% of AI POCs never reach production. The failure rate is structural, not technical. I examined what the companies converting above 20% do differently in the latest edition of The Revenue Circle. #GTM #AI #Revenue #Saas #sales #enterprise #PMF #AIEnterprise

Eight weeks ago, I published a diagnostic on why Q1 forecasts were already broken by January 5th. Three signals I named: ✅ Commit deals without validated buyer engagement since mid-December. ✅ Pipeline coverage built on phantom opportunities. ✅ Stage progression driven by seller activity, not buyer commitment. The responses since then confirmed the pattern. Multiple revenue leaders reached out, describing the exact sequence: deals that were "committed" in December returned in January as "requires re-engagement." Pipeline coverage that showed 3.2x on the board deck recalculated to 2.1x once someone removed dormant opportunities. Forecast calls that generated confidence in January led to revision conversations by mid-February. None of this was unpredictable. That is the point. ❓ Forecast accuracy does not collapse in the final week of the quarter. It degrades in the first week, when the structural gaps are visible, but the organisational incentive is to believe the number will hold. ❌ By the time March arrives, the intervention window has closed. What remains is narrative management explaining why the quarter missed rather than preventing the miss. The companies that acted on the January diagnostic auditing commitment deals for genuine buyer commitment, removing phantom pipeline, recalibrating coverage to reflect reality, entered March with a number they could defend. Not because they found more revenue. Because they stopped forecasting against assumptions that had already expired. The original diagnosis is linked below. If you are reading it now and recognising your own Q1, the question is not what happened this quarter. The question is whether the system that produced this outcome will produce the same result in Q2. #Revenue #AI #Sales #SaaS #Technology #innovation #Q1 #Forecast #gtm #Marketing

Claude releases a new feature at 2 PM. By 2:47 PM, someone on LinkedIn has already booked 50 sales calls using it. By 3:15, someone else fed their "100M-view post" into it and discovered a research breakthrough. By 4 PM, there's a carousel: "7 ways this changes everything." Fifty calls. "In forty-seven minutes. Do the math on that it's a call every 56 seconds. No configuration, no learning curve, no dialing." No download time. No configuration. No learning curve. No dialing. Just fifty calls. Materialized from a product announcement they read nine minutes ago. I work in enterprise revenue. I've spent thirty years in rooms where people make claims about the pipeline and then get asked to show the CRM. Fifty calls in 47 minutes would get you laughed out of a forecast review. But this isn't a forecast review. This is LinkedIn. And on LinkedIn, the claim doesn't need to survive scrutiny. It needs to survive scrolling. Three seconds of plausibility. 😁 That's the bar. The comment section does the rest, with two hundred people saying "this is incredible" under a claim that nobody checked, and the tool probably can't do. OpenAI ships a model update on a Tuesday morning. By Tuesday afternoon, someone has "completely rebuilt their sales process" with it. Have they? Open the CRM. Show me the pipeline. Show me one closed deal that moved because of what you built between lunch and the post going live. Nobody can, because the post was never about the tool. The post is about being seen as the person who moves fastest on the new thing. The performance of expertise, not the possession of it. This is the part that connects to something bigger than LinkedIn. The feed rewards speed over substance. First take over right take. 🔴 Claiming over the building. And it does this not because the algorithm is evil but because we all, me included, engage with the fast claim more than the slow truth. ⛔ The person who says "I'm still figuring this out" gets 4 likes. Say 'I'm still figuring this out' on LinkedIn and you'll get 4 likes and a dead post. ✅ Say 'I've already booked 50 calls,' and four hundred people show up to tell you you're a genius." Not about LinkedIn specifically. About what happens to professional identity when the platform incentivizes performance of competence over actual competence FEED REWARD THE FASTEST LIE About the version of yourself that the feed built, the one that always has a take, always has results, always moves first. I call it the Feed Self. It's the professional equivalent of the person in the coffee shop who photographs their meal before tasting it. ♻️ The book is called Feed Gods. March. And yes, I posted this on LinkedIn, which is exactly the kind of thing the book is about. I'm not outside this. Nobody is. The difference isn't awareness. It's about whether you're willing to look at the architecture rather than just perform within it. #book #feedgods #author #coach #writer #linkedinwrites

Fourteen people are in this coffee shop. Eleven are on their phones. I counted because I'm on a screen too. The difference between the guy two tables over isn't awareness. He probably knows exactly what he's doing. We all do. Knowing doesn't change the behaviour. That's the part nobody wants to sit with. 🔴 A couple came in together twenty minutes ago. Ordered together. Sat down. Then both went somewhere else same table, separate feeds. One of them half-laughed at something on their screen. The other didn't ask what was funny. That moment when curiosity about the person in front of you gives way to whatever the algorithm serves next, that's not a phone habit. That's a structural change in how humans allocate attention to each other. I've spent two years researching how this happened. Not the technology, the technology is obvious. What isn't obvious is the person who got replaced. Gradually. Voluntarily. The version of you that used to sit in a place like this and get bored. Stare out of a window. Let a thought arrive without requesting it. That person doesn't show up anymore. ✅ Someone else does. Same face. ✅ Different attention. ✅ Shaped by a feed that trained them to need something new every eleven seconds. I call it the Feed Self. The version of the platform on which it was built. There's another version underneath the one that existed before the feed. It's still there. It just stopped being invited. ♻️ The book is called Feed Gods. It comes out in March. It's the most difficult thing I've written because the subject includes me. I'm not watching this from outside. I'm sitting in the same coffee shop, on the same kind of screen, fighting the same architecture. I just happened to have spent two years learning how it was built. I looked up after writing this paragraph. Twelve are on their phones now. One more than when I started. I went back to my screen. #Attentioneconomy #Linkedin #author #books #feedgods

What is your customer actually buying when they buy an AI agent? Not your product. Not your model. The billable unit. The thing that shows up on a procurement line item. Can you name it? Salesforce couldn't. They shipped three different answers in 18 months. ❓ Does your customer's bill increase when your product performs better? If yes, you have built a pricing model that punishes adoption. The customer bought the agent to reduce costs. Your commercial architecture charges them more for succeeding. ❓ Can your buyer's CFO forecast your cost within 15% for four consecutive quarters? 73% of enterprise AI deployments went over budget. If the answer is no, your deal cycle just added months. Not because of the product. Because of procurement math. ❓ What happens to your revenue when inference costs drop another 80%? GPT-4 pricing fell 75-80% in 18 months. If your pricing is anchored to compute, your unit economics degrade as your usage grows. Revenue is built on a cost curve that only moves in one direction. ❓ If your agent succeeds, does your customer need less of your other products? Salesforce is seeing 10% seat reductions across 90 enterprise accounts. Their AI product is cannibalizing their own per-seat revenue. This quarter. Not hypothetically. ❓ What percentage of your customers' actual costs appears on your price sheet? API pricing is 20-30% of the real deployment cost. The rest of the orchestration, infrastructure, implementation, and governance is where the trust gap opens, and retention dies. Seven questions. Attached below. No answers included. The answers depend on things I don't know about your business. But if your leadership team sat in a room and couldn't answer five of these with specificity, the pricing model isn't the problem. It's the commercial architecture underneath it. #AI #Revenue #pricing #API #AgenticAI #GTM #marketing #technology #innovation #futureis #therevenuecircle

There is a question I have started asking in every pipeline review conversation over the past year, and it consistently produces the most uncomfortable silence in the room. ✴️ "Of your current commit deals, how many originated from a conference where the prospect agreed to a follow-up because you were standing in front of them?" The silence is not because people do not know the answer. It is because they do know, and they have been avoiding what that answer means for their forecast. ‼️I went deep on this. Pulled industry conversion data, studied a specific pipeline of forty event-sourced accounts over several months, and tracked what actually closed versus what occupied forecast space. The gap between what the event pipeline looks like in a CRM and what it actually produces is wider than most revenue leaders realise, and it has a specific structural cause that has nothing to do with follow-up execution. The full diagnostic is in the article. #GTM #Sales #SaaS #AI #PMF #Marketing #innovation #technology #futureis #revenue #conferences #events