A chasm between AI adoption and AI return showed up across multiple sources this year. Atlassian’s 2026 State of Teams research found that 89% of executives say AI increased the speed of work, and only 6% can point to organization-wide ROI. McKinsey’s 2026 research on enterprise AI found the same split. Individual productivity is up. Company earnings are largely unmoved. What’s missing from that conversation is the cause. The split between 89% and 6% isn’t a technology gap. Your tools work. The disconnect lives in three things that didn’t change when you bought them: the people who use them, the organization built around them, and the foundation underneath them.  

Robots & Pencils, an applied AI partner that deploys frontier AI for measurable business outcomes, today published The AI Productivity Paradox, a three-part series, written by Brendan Flynn, SVP Strategist, in the company’s Studio for Generative & Agentic AI.  

“You bought the licenses. That’s not the gap. The gap is a role that’s already changed and nobody’s noticed, an org chart that hasn’t caught up, and a foundation most companies don’t own. This series is about closing all three,” said Flynn.  

Why AI Pilots Stall: The People, Organization, and Foundation Gaps 

This series is written for operations, technology, transformation executives, and leaders of enterprise companies who already have AI tools live, have watched individual people get faster with them, and can’t point to a company-wide number that moved. Each article focuses on a specific gap and closes with two or three questions a leadership team can ask at its next meeting. 

• Part 1 of the series, “Your Teams are Already Changing Their Jobs, Have You Noticed?”, names a job most companies already have someone doing informally, and nobody has hired for on purpose. Flynn calls this person “The Orchestrator” because they recognize that a role has changed before the org chart does and turn one person’s workflow into something the rest of the team can use. 

• Part 2, “Common Mistakes in AI Organization Design (and how to fix them),” walks through three structural errors that stall AI initiatives after a promising pilot. Flynn pairs each mistake with solutions drawn from engagements across industries and operations, including Robots & Pencils’ own internal experiment, a four-week, company-wide event called RoboCon that put AI skill-building in the hands of seven cross-functional pods with no formal training. 

• Part 3, “Own the Foundation, Rent the Model,” argues that the asset that appreciates is the foundation underneath the frontier model. A good foundation includes a shared definition of the business, one door every AI request goes through, a register of every tool running, a way to test whether each one is working, and approval scaled to true risk instead of applied evenly to every decision. 

The full series, a 15-minute read, is available now. Start reading part 1 of The AI Productivity Paradox now. 

The post The AI Productivity Paradox: Why your teams have the tools but not the results.  appeared first on Robots & Pencils.

There are moments in life that stand out and change everything. Getting my driver’s license. The birth of my children. And today, when I got early access to Meta’s live voice calling feature.

The kids are still ahead. But the gap narrowed slightly when AI offered to deal with my dentist.

Where AI fits in my workday

As VP of Experience at Robots & Pencils, I spend a lot of time evaluating AI platforms and models for product delivery. Some are better suited to reasoning through complicated problems, others to writing code, exploring visual concepts, or analyzing information. We’re constantly testing where each fits into our process.

On any given day, I’m using these tools to untangle complex workflows, turn ideas into functioning prototypes, and solve problems that used to take entire teams weeks to work through.

But today, I asked AI to make a dentist appointment for me.

Apparently, I can help design complex digital products, but calling a dentist during business hours requires a rare alignment of celestial events.

The boring tasks are the big opportunity

That’s the opportunity I think the market may be underestimating.

We hear about AI curing diseases, accelerating scientific discovery, and transforming entire industries. Those ambitions matter. Meanwhile, millions of people would also appreciate getting six annoying things off their to-do lists.

An assistant that reliably handles those tasks earns a place in everyday life. And that could have enormous implications for how we design digital experiences.

When my agent talks to your agent

Right now, an assistant can make a phone call on my behalf. Looking ahead, Meta’s existing business and messaging relationships suggest a path toward my agent communicating directly with a business’s agent.

My agent asks my dentist’s agent for an appointment. They compare availability, work within my preferences, and book it.

No hold music. No voicemail tag. No website.

Commerce could work the same way. My TV breaks, so I ask my agent to find a replacement within my budget, confirm it fits my space, and arrange delivery. It brings me a recommendation for approval, then handles the purchase.

An afternoon of research and checkout flows becomes a few seconds of my attention.

What happens to the interface

For those of us in user experience (UX) and user interface (UI) design, that changes the assignment.

We’ve spent years helping people navigate websites, use apps, and complete transactions. But when an agent can handle the transaction, the customer may never encounter the interface we designed.

For that task, the company’s website or app becomes irrelevant to them.

A beautiful booking flow has limited value to someone whose assistant already booked the appointment. Every interface will face a higher bar: is this experience worth the customer’s time and attention?

The new brief: delegation, boundaries, and trust

Our work expands into designing how people delegate, set boundaries, grant permission, review decisions, and recover from mistakes. Businesses will need services that agents can discover, understand, and use reliably. Trust and clarity become essential parts of the experience, even when there’s very little screen involved.

People will still want to browse, explore, and make considered decisions. But they may expect routine tasks to happen simply by asking.

The question for companies is how well their experience works when the customer never opens the app.

I’ll be thinking about that.

Possibly in the dentist’s waiting room.

Want to see how agents change your customer experience?

Robots & Pencils runs AI briefings for leadership teams that want a clear read on where agents will touch their products, their customers, and their roadmap. Request an AI briefing, and we’ll bring the perspective, you bring the questions.

The post AI Called My Dentist. Now I’m Rethinking UX. appeared first on Robots & Pencils.

Len made two recent trips. In July, he flew to Bogotá with Nick Waynik, SVP Global AI & Engineering, and Jeff Kirk, EVP Applied AI. In October, he went to Lviv with Waynik, Nate Carmon, COO, and Tony Antonelli, CFO. 

“Our teams in Lviv and Bogotá are a big reason we can move as fast as we do,” said Pagon. “I wanted to thank them in person, and you can’t do that on a video call. You sit down at lunch, you hear what’s working and what isn’t, and you get to know each other better as colleagues. We’re remote-first, and that’s not changing. We’re also going to keep getting on planes.” 

Lviv

From October 12 to 14, four members of the Robots & Pencils leadership team visited our Ukraine office: Len Pagon, Chief Executive Officer (CEO); Nick Waynik, Senior Vice President (SVP) of Global AI & Engineering; Nate Carmon, Chief Operating Officer (COO); and Tony Antonelli, Chief Financial Officer (CFO). They held meetings and working sessions in the office and shared lunches with the team. On Tuesday evening, everyone gathered for a bowling reception with lanes, billiards, and a healthy amount of friendly trash talk. 

Khrystyna Hrupska, Talent Operations Specialist in Ukraine, helped organize the visit. She summed it up for the team afterward, “There were many interesting discussions, productive meetings, new ideas, fruitful cooperation, and, of course, positive emotions.”  The bowling night amped up the fun and competition.  

Waynik put it plainly in his thank-you: “I appreciate all of the warmth and hospitality you’ve shown us, and I enjoyed meeting each and every one of you in person.” 

Engineers in Ukraine tell us they value clear expectations and work that holds their interest. Four members of the leadership team in the Lviv office put that team at the center of the company, where it belongs. 

Bogotá 

In July, Pagon, Waynik, and Jeff Kirk, Executive Vice President (EVP) of Applied AI, visited our Bogotá office. Our Colombia team is newer, and many of them met colleagues in person for the first time. Pagon, Kirk, and Waynik attended AWS Summit Bogotá with local team members.  

Engineers in Colombia choosing their next role weigh stability and a real place to work alongside people as heavily as the work itself. Three senior leaders flying in to meet the team by name answer both. 

The Proximity Premium 

Robots & Pencils is a remote-first company. Our work runs on Slack threads, standups, and SharePoint, and it runs well. A video call still carries information, and a shared lunch carries something more. It is the extra trust, speed, and context that show up when people share a room. 

Our Ukraine development center opened in early 2025, and our Bogotá office opened earlier this year. Both teams are growing fast.  

Interested in joining Robots & Pencils?  View open positions.  

The post Proximity Premium: Recap of Senior Leadership Visits to Our Ukraine & Colombia Offices   appeared first on Robots & Pencils.

Robots & Pencils, an applied AI partner that deploys frontier AI for measurable business outcomes, today announced the appointment of Helen Vetrano as Vice President of Data and Decision Science.  

Vetrano leads the company’s Data & Decision Science practice, which ties data quality and business intelligence to the velocity and performance of the agentic AI systems Robots & Pencils builds for clients. She brings 20 years of analytics experience, including agency work at Digitas and Cramer, followed by nearly 15 years in retail at Crocs, Talbots, J. Jill, and her own consultancy advising mid-market retailers on analytics and data strategy. Based in Boston, Vetrano anchors an investment in the East Coast that puts Robots & Pencils closer to clients in the region. 

“Agentic AI is only as fast and as capable as the data underneath it,” said Len Pagon, CEO of Robots & Pencils. “You bring agents to the data party, not the other way around. Helen is wicked smart, brings deep domain expertise, and is practical. Every client starts at a different level of data maturity, and she knows it. She prescribes the right solution for where they are, and that is how they get the business outcomes they are seeking.”  

Enterprise executives are under pressure to grow with flat budgets, optimize costs and keep security tight, all while delivering a platinum customer experience. Fall short and they risk handing market share to competitors. “Excelling in uncertainty requires great decisions, and great decisions are made with great insight,” said Vetrano. “This practice gives enterprise leaders the data foundations and intelligent systems to make confident, accelerated decisions.” 

Enterprises move fast and need answers quickly, but data systems often can’t keep up. “I’ve seen this challenge from both the commercial and IT side; AI can help close that gap, but only if the platform underneath is built to support it,” said Vetrano. “Robots & Pencils delivers resilient engineering focused on business outcomes. It’s an offering I would buy, and I joined to help scale it.” 

The practice meets the client at the point of their greatest need. “For some organizations that means surfacing critical context and a shared language across teams or developing the predictive models that matter,” said Vetrano. “For others it means just getting data into reports consistently. With a solid data foundation, clients can focus on outcomes, like planning where to spend the next marketing dollar, allocating resources where they are most needed, or intercepting a struggling student.” 

The practice builds client data foundations on leading platforms, including AWS, Databricks, and Snowflake, so people and agents reach a working system of truth quickly.  

“Helen brings a retail leader’s instinct for data, built on business outcomes,” said Jeff Kirk, EVP of Applied AI at Robots & Pencils. “She pairs that instinct with the business and data context platform work our clients depend on, so every agentic AI system we build starts on data clients can trust.” 

Clients start at different levels of data maturity. Talk with our Data & Decision Science team about where yours stands and what it takes to put agents on a foundation you can trust. Request a data briefing.

The post Robots & Pencils appoints Helen Vetrano as VP of Data & Decision Science  appeared first on Robots & Pencils.

Lay of the Land is a Cleveland video podcast built on long conversations with the people who build and shape Cleveland and Ohio. Jeffrey opens with the obvious question for a CEO who spent 17 years as an investor and board member before coming back to lead: why jump back into the arena?

Len talks about how Robots & Pencils became an AI-first company, why learning by doing beats training, how a one-week deadline changes the way teams work, and how one person’s skill becomes a whole team’s head start.  

He and Jeffrey also cover where lasting value will land as AI reshapes software and services, where leaders should start, and which parts of the work still belong to leaders, including data quality, culture, policy, and technical debt. The conversation closes with Len’s advice for this moment and his pick for a Cleveland hidden gem.  

Press play to hear it all. 

Looking to work for a leader who invests in your continued career growth? View open positions.

The post Len Pagon Appears on Cleveland’s “Lay of the Land” Podcast appeared first on Robots & Pencils.

Robots & Pencils, an applied AI engineering partner known for high-velocity delivery and measurable business outcomes, today announced the acquisition of Tiger Pistol, a marketing automation platform that helps enterprises, franchises, and resellers manage and scale digital advertising campaigns across Meta, TikTok, Amazon, and other channels. Tiger Pistol joins Robots & Pencils as its customer experience transformation studio practice, bringing a proven platform, a product development team, and deep marketing automation expertise. 

Building Reusable Agentic Patterns for Enterprise Teams 

Robots & Pencils builds AI into the systems enterprises already run on. Tiger Pistol built the platform that runs advertising for brands, franchises, and resellers. Each reached the same conclusion from its own side. Customer engagement work is becoming agentic, and the next stage belongs to companies that can scale agentic patterns. Patterns are rinse-and-repeat blueprints with working code, so teams rapidly adapt a proven solution for each client instead of rebuilding from scratch, getting clients to known business outcomes in compressed timeframes and with less risk. 

Tiger Pistol is the first example. Its platform already handles the repeatable work of advertising for global franchises and resellers. Robots & Pencils provides additional agentic systems that keep improving those campaigns as results come in, and the same pattern adapts to the next brand, franchise network, or reseller. 

Robots & Pencils is moving from a pure services model toward repeatable, product-accelerated solutions that combine software, agentic patterns, and expert delivery. 

Frontier AI and Forward-Deployed Engineering Meet a Proven Product Team 

The acquisition of Tiger Pistol accelerates this strategy by adding a proven product development capability, culture, platform, and deep domain expertise in marketing automation. With more than four million digital advertising campaigns launched, the platform has been in market for more than 15 years. Robots & Pencils adds applied AI and the frontier AI engineering that builds it into the systems a business already runs on. Its forward-deployed engineering approach puts engineers side by side with a client’s team to build and launch solutions.  

“Brands need integrated, outcome-driven agentic systems that work at scale,” said Len Pagon Jr., chief executive officer (CEO) of Robots & Pencils. “By converging Tiger Pistol’s product development strengths with our applied AI, forward-deployed engineering depth, and delivery model built on AWS, we’re giving clients a rinse-and-repeat path to measurable business results, live in as little as 30 to 45 days.” 

Marketers who once launched campaigns and reviewed results after the fact will have agents adjusting campaigns as results arrive, and creative moving from idea to published faster.  

“Marketers came to Tiger Pistol to accelerate scale and speed to market, while maintaining brand control,” said Paul Elliott, CEO of Tiger Pistol. “Now they can transform the entire customer experience.” 

Tiger Pistol customers will now be able to: 

• Augment Tiger Pistol’s automation with AI agents that continuously optimize campaigns based on performance data, brand standards, and media intelligence. 

• Automate media, ads, video creation, and publishing through Tiger Pistol’s Creative Automation Studio, paired with generative models informed by historical performance, audience response, and brand coherence. 

• Draw on Robots & Pencils’ Applied AI Studio and forward-deployed engineers and AI builders to rapidly design, build, and deploy adjacent AI solutions alongside Tiger Pistol’s tools. 

• Productize and harden Robots & Pencils’ other Agentic Customer Experience Patterns like their Voice-2-Voice patterns they have implemented for a number of other clients. 

• Access AWS-aligned innovation including advanced data science, scalable cloud infrastructure, and reusable agentic patterns through Robots & Pencils’ status as one of 11 inaugural AWS Pattern Partners selected from the global AWS Partner Network, and its selection as an AWS Business Value Realization (BVR) Motion Launch Partner. 

• Accelerate outcomes through Robots & Pencils’ flywheel approach to taking new capabilities live in as little as 30 to 45 days, enabling rapid experimentation, deployment, and measurable business outcomes. 

Together, Robots & Pencils and Tiger Pistol give marketers one engine for running campaigns, producing the creative, and building the AI that connects both to the rest of the business. The combined organization will keep investing in reusable AI assets, agentic patterns, and solution accelerators, with select offerings available in AWS Marketplace. 

To learn how Robots & Pencils takes AI from pilot to live, request an AI briefing. Learn more about Tiger Pistol’s AI solution for digital advertising at scale at tigerpistol.com. 

The post Robots & Pencils Acquires Tiger Pistol to Scale Reusable Agentic Patterns into Products   appeared first on Robots & Pencils.

Robots & Pencils, an applied AI partner that deploys frontier AI for measurable business outcomes, today announced it is returning as a sponsor of the Agentic AI and the Student Experience Summit, hosted by Arizona State University (ASU) October 20 to 22, 2026, at the Omni Tempe Hotel. Elliott Hedman, Robots & Pencils Lead AI Strategist for Education, will join a panel session, “Using AI to help professors start the conversations students avoid,” on how agentic AI can help faculty notice when a student is struggling or facing an academic integrity concern, and start the conversation before it’s too late. Hedman holds a Ph.D. in psychophysiology, AI, and user research from the Massachusetts Institute of Technology (MIT). 

Higher education leaders spent the last two years piloting generative AI chatbots and assistants across advising and student support. Most of those pilots answer a question a student already knew to ask. Few reach the student who disengages and never asks anything at all, which is the harder problem advising and student success offices are now asking vendors to solve. 

“Most AI tools wait for a student to come ask for help. The best tools find the students who are struggling and step in before they ask,” said Hedman. “That’s what I’m walking through in Tempe: what it takes to build an agentic AI system that notices when a student may need help, reaches out before they are asked to, connects them to the right campus resource, and follows through until they get it.”  

“Agentic AI in higher education is measured by whether students get help before they have to ask, and whether faculty and staff trust it enough to point students toward it,” said Jason Lacy, SVP, Client Partner at Robots & Pencils. “That’s ultimately the standard proactive student support has to meet.” 

Senior Delivery Manager Alison DuBose and Technical Lead Chetan Thayur will join Lacy and Hedman on-site for the three-day summit, meeting with higher education leaders throughout the event by request. 

Robots & Pencils has delivered production systems at ASU across enrollment, financial aid, media, credentialing, and operations, including MediaPlus, ASU’s enterprise media platform built with AWS, a Canvas enrollment system that cut course creation and enrollment from three days to thirty minutes, a 24/7 financial aid chatbot, and the CredID digital credentialing platform. Today, the team is building Nova Sol, an agentic platform that helps ASU reach students who need support, alongside Degree Planner and the Food is Medicine wellness portal.  

Higher education leaders attending the summit can request time with the Robots & Pencils team by contacting Lacy directly at jason.lacy@robotsandpencils.com.

The post Robots & Pencils to Sponsor and Speak at ASU’s Agentic AI and the Student Experience Summit  appeared first on Robots & Pencils.

Part 1: The People | Part 2: The Organization | Part 3: The Foundation

Somewhere in your company right now, a handful of people have become several times more productive with AI. Not because of the training program everyone had to take or the Copilot licenses that were purchased. On their own, on real work, because it made their week better.  

If you asked them, they would tell you it feels like work that used to take them days, now completes in minutes. They are in every organization. 

But your company’s productivity numbers didn’t move. That gap is the paradox in its most personal form.  

Atlassian’s 2026 State of Teams research puts a number on it: 89% of executives say AI has increased the speed of work, but only 6% can point to organization-wide ROI. Individual speed is everywhere. Organizational results are rare. The interesting question isn’t whether AI works; your people have already answered that. It’s why individual 10x doesn’t add up to organizational 10x. 

Where the gains go 

The gains are real. They’re just trapped. Three things trap them, and none of them is technical. 

First, the gains live in individuals. The prompts, the workflows, the personal knowledge base, the judgment about when to trust the AI output and when not to, all of it sits in someone’s chat history and someone’s head. When that person is on vacation, the gain is on vacation. When they leave, it leaves with them. 

Second, nobody can see the gains from the outside. The report still lands on Friday. The analysis still arrives before the meeting. The work looks identical to what it looked like a year ago. People are using the time they saved to do more of the old job, or to go home at a reasonable hour. Neither shows up in a productivity dashboard. 

Third, people have a reason to keep quiet. The honest reaction to suddenly being several times faster is not pride. It’s something closer to this feels like we shouldn’t be doing this. If the work that used to take a week now takes an hour, what does that say about the week? About the role? About the headcount? Your most proficient people may be under-reporting, not over-claiming. The paradox has a visibility problem before it has anything else. 

What we learned running it on ourselves 

This summer we ran an experiment at Robots & Pencils called Robocon. Four weeks, company-wide, across seven cross-functional pods we were given challenges to build and publish working AI skills to a shared internal library, then show them off at a live event. No course. No training. Not just engineers, but everyone in the company. Build something real and put it in front of people. 

Two lessons came out of it, and neither was the one we expected. 

The first was that the barrier wasn’t skill. Teams lost hours, in some cases days, getting environments, repositories and file-share access to work before they could build anything. Once these hurdles were cleared, people began creating at an alarming rate. Everyone in the organization from marketing, sales, product, project managers, design, engineering, came out the other side with real AI proficiency. The expensive part of proficiency wasn’t teaching people. It was clearing the path so they could do the work. 

The second lesson landed harder. The library filled up fast, and plenty of good work shipped and then sat there, because nobody’s job was to turn what one person built into something the rest of the company used. Not everything built was worth keeping, and the pieces that were, had to be deliberately identified, hardened and surfaced to the rest of the company. Sorting them was a separate act, done by different people, with a different skill. 

That’s the pattern, and we think it generalizes. The event produced proficiency in individuals. Amplifying it across the organization was a second job, and it didn’t happen on its own. 

Roles aren’t redesigned. They are discovered.  

Most advice about AI and the workforce runs top-down: redesign the roles, then deploy the tools. In our experience it works the other way around. People pick up the tools, use them to create value for themselves first, and the role starts to change underneath them. The job description is the last thing to move. AI is moving from individual productivity to multiplayer, collaborative productivity.  

That’s the good news, because it means the change is already happening in your company. The challenge is that discovery doesn’t distribute itself. A role that has quietly changed in one person’s hands stays there unless someone does three things on purpose. 

Find it. Who has already changed how they work? You won’t learn this from a survey; you learn it by asking a different question: not “are you using AI?” but “what part of your job have you automated?” People who have stopped doing something manually are the ones whose role has evolved. 

Harvest it. Turn what one person does into something others can use. Not a training deck or show and tell, but the actual workflow, the actual prompts, data and tools they have connected, and the actual judgment calls about when the output can be trusted. Then be honest about what holds up. Our own library taught us that a lot of good individual work is not, in fact, reusable. The discipline is in keeping what is. 

Make it official.  Redefine the role around what changed and retire the old work. Skip this and people end up twice as fast in a job still defined as if they weren’t. The time saved gets refilled with more of the old job, and the productivity number never moves. 

We did this to ourselves. A client team that used to be seven to ten people is now three to five. AI didn’t take the seats — the people kept them and got more done. Each person now works with AI the way they’d work with a strong analyst: it drafts, researches, checks and runs the routine parts, and they spend their time on the judgment calls. Once that was true in practice, we rewrote how we define what a team is on paper: who’s on it, what each role owns, what nobody does by hand anymore. 

The job nobody has hired for 

Boris Cherny, who leads the Claude Code team at Anthropic, has described how engineering and product roles on his team have melted into six archetypes: the Prototyper who generates ideas most of which don’t ship; the Builder who turns a validated idea into production; the Sweeper who simplifies and removes; the Grower who iterates for scale; the Maintainer who owns the mature system; and the Orchestrator, who knows which of the other five a team should be in, and when to move between them. He notes the Orchestrator is rarer than the other five combined, and that it has no established job title. Most people are doing 2 or 3 of these archetypes today. 

We think that last role is the one the productivity paradox is missing. The Orchestrator is the person who does the finding, harvesting and socializing to teams. They notice that a role has changed before the org chart does.  

Nobody at this table has hired one. Most organizations don’t know it’s a job. But somebody in your company is already doing a version of it informally — the person other people go to when they want to know how so-and-so got that done so quickly. That person is your first Orchestrator. They just don’t have the title yet. 

Questions to take back 

Your people are already changing their jobs. The organizations that get results from AI aren’t the ones that designed the change from the top. They’re the ones that noticed it from the bottom and made it official before it evaporated. 

Two questions for your leadership team this week: 

1. Who in your company has already changed their job without telling you? 

2. And whose job is it to notice? 

Up next in this series, Part 2: Organization

Individual proficiency is the easy part. The harder part is turning it into something the organization can run on without you in the room. In Part 2, we break down the three structural mistakes that stall AI pilots before they reach production, and what to build instead. Read Part 2: Common mistakes in AI org design (and how to fix them).

Sources referenced: Atlassian, State of Teams 2026 (89% / 6% figures). Boris Cherny, public posts on role archetypes (X.com). 

About the Author

Brendan Flynn is SVP, Strategist at Robots & Pencils where he heads industry strategy within the Generative & Agentic AI Studio.

Key Takeaways

• Individual AI proficiency and organization-wide ROI are not the same thing. Atlassian’s 2026 State of Teams research found 89% of executives report AI has increased the speed of work, while only 6% can point to organization-wide ROI.
• Productivity gains stay trapped in three places. They live inside individuals who never document how they work, they stay invisible to standard reporting because the output looks the same as before, and they stay inside people who have a reason to stay quiet about how much time they have saved.
• Robots & Pencils ran a four-week, company-wide AI build event called Robocon. The barrier to proficiency was never skill. It was clearing access to the environments and tools people needed to start building.
• Roles change from the bottom up. People adopt AI, get faster, and the job description catches up later, if anyone is watching for it.
• Turning individual proficiency into organizational capability takes three deliberate steps: find who has already changed how they work, harvest what they are doing into something reusable, and make the new role official on paper.
• Anthropic’s Boris Cherny describes a role missing from most org charts, the Orchestrator, the person who notices when a role has changed and helps it spread. Most companies already have one working informally and have not named the job.

FAQs

What is the AI productivity paradox?

It is the gap between how fast people say AI has made their work and how rarely that speed shows up in company-wide results. Atlassian’s 2026 State of Teams research found 89% of executives report faster work from AI, while only 6% can point to organization-wide ROI.

Why don’t individual AI productivity gains show up in company-wide numbers?

The gains stay trapped in three places. They live in one person’s prompts and judgment instead of a shared process, they are invisible to standard reporting because the output looks the same as before, and people often stay quiet about how much faster they have become.

What did Robots & Pencils learn from running Robocon, its internal AI build event?

The barrier to AI proficiency was never skill. Once people had access to the environments and tools they needed, proficiency spread quickly across the company. The harder problem was turning one person’s work into something the rest of the organization could use.

What three steps turn individual AI proficiency into organizational capability?

Find who has already changed how they work, harvest what they are doing into a reusable workflow, and make the change official by redefining the role and retiring the old work it replaced.

What is an AI Orchestrator?

A term from Anthropic’s Boris Cherny for the person who notices when a role has changed because of AI and helps spread that change across a team. Most organizations already have someone doing this informally, without the title.

The post Part 1: The AI Productivity Paradox – Your Teams are Already Changing Their Jobs, Have You Noticed?  appeared first on Robots & Pencils.

Part 1: The People | Part 2: The Organization | Part 3: The Foundation

You have a few AI pilots that have worked. The results are promising, and the business case makes sense. You spin up a dedicated AI team, give them a mandate, and hand off the work. 

Six months later, nothing has shipped to production. The business side says the AI team doesn’t understand the constraints. The AI team says the business doesn’t understand what’s possible. Meanwhile, the productivity gains that looked obvious in the pilot have vanished. 

It’s the most common failure pattern we see, and it’s rarely a technical one. It’s organizational. 

After working through this with clients across finance, manufacturing, and operations, we’ve landed on three structural mistakes that kill AI initiatives before they scale. Here’s what they look like, and what we’ve found works. 

Mistake 1: Centralizing AI Decisions Away from the Business 

The error: You create an AI Center of Excellence or a dedicated AI team, and suddenly every AI decision routes through them. They’re the gatekeepers. The business waits. 

Why it fails: The people closest to the problem, the ops manager, the finance lead, the customer service director, can’t move fast. They hand requirements to the AI team, the AI team interprets them, requirements get misunderstood, timelines slip, and by the time something ships the business context has already shifted. 

What works instead: Push AI decisions closer to the business. You still need shared standards — how AI workflows get evaluated, how decisions get logged, how risk is governed — but the business unit lead, who owns the outcome, should decide when an AI workflow is ready to go live in their function. 

We made the same call inside our own delivery organization. Instead of standing up one central AI practice that every client team has to route through, we organized around small cross-functional pods, each aligned to a specific client, each deciding for itself how AI gets used in that engagement. A pod answers to shared standards, a hiring bar, a set of proficiencies, not to a gatekeeper reviewing its every move.  

Mistake 2: Treating Upskilling as Training, Not Proficiency Building 

The error: You run a one-week training program. Everyone learns to use Claude, Copilot, or ChatGPT. Then you expect productivity to jump. 

Why it fails: Training teaches how to use a tool. It doesn’t build organizational proficiency. The operations manager learns Claude syntax, goes back to their desk, and does the same job exactly the way they used to, just a little faster. Productivity gains are marginal, executives see no ROI, and everyone quietly concludes AI didn’t work. 

What works instead: Build proficiency through doing, not classrooms. 

Start by asking what this person stops doing, what new responsibilities they take on, and how their relationship to the work changes. Answer those honestly and you’ve effectively restructured how that function operates, at which point upskilling stops being a training event and becomes proficiency built through doing the work. 

A client we worked with had an immediate unlock the first time we had them use an orchestrated workflow we built for their forecasting process. The response was emphatic, “This will replace 90% of the meetings we have, I can simply ask AI questions about my forecast, and it knows my entire book.”  

Zero training involved, they got it immediately. That is powerful. This frees them up people to do what they do best, build relationships to close deals.  

Mistake 3: No Feedback Loop Between Operations and AI 

The error: The AI team builds an AI workflow, the business puts it to work, and six months later nobody can say whether it’s moving the outcomes that matter. 

Why it fails: Without real feedback you can’t improve: performance drifts, edge cases pile up, the business loses confidence, and the AI team never learns what’s needed. 

What works instead: Build operational feedback into the rollout from day one. Who’s watching how it’s performing against the outcomes you care about? Who surfaces problems? How fast can you respond? 

On another engagement, we skipped the weekly-sync approach entirely and went tighter. The day two account managers first tried the assistant live, on their own real data, their feedback went straight into a ranked list for engineering before the day was over. Four fixes made the build before the end of the same day the feedback was received. A handful of other requests got logged as legitimate but not urgent. A couple of ideas got an explicit “not this round,” with the reasoning written down so nobody had to relitigate it later.  

The Pattern Underneath 

These aren’t technology problems, they’re proficiency problems: companies think they’re buying an AI system when what they’re building is organizational capability. 

The fix comes down to three structural shifts, and they reinforce each other.  

Push AI decisions closer to the business, because feedback only flows fast when the people doing the work own the decision to change it.  

Redefine roles and workflows through doing instead of training, because software that keeps evolving forces people to build proficiency in real time, and that’s where the actual transformation happens.  

Close the feedback loop, then keep it tight, because iteration speed is the discipline: every cycle, the system improves, and the organization learns alongside it. 

When those three things hold, AI proficiency becomes the muscle of how the organization operates day to day, and the productivity gains stick instead of fading out after the pilot. None of it requires a heroic AI team. It requires distributed decision-making with clear guardrails around it. 

If your AI tools are live but proficiency, and the productivity gains that are supposed to come with it, still haven’t shown up, this is the first place to look. 

Questions to take back 

1. What did your last AI pilot leave behind that the next one can reuse, or are you paying to start over?  

2. Of the AI you’re running today, which of it can someone prove is working? 

Up next in this series, Part 3: Foundations

Fixing the organization gets you further, but it does not answer the harder question underneath it. Once the model itself becomes a commodity, what should you actually own? In Part 3, we lay out the five-part foundation every company needs, whether it builds its own AI or simply buys it. Read Part 3: Own the foundation, rent the model.

About the Author

Brendan Flynn is SVP, Strategist at Robots & Pencils where he heads industry strategy within the Generative & Agentic AI Studio.

Key Takeaways

• Most AI initiatives fail for organizational reasons, not technical ones. The common pattern is a pilot that succeeds, a dedicated AI team that takes over, and six months later nothing has reached production.
• Centralizing AI decisions in one team slows everything down. The business unit that owns the outcome should decide when a specific AI workflow goes live, inside shared standards.
• Training teaches a tool. It does not build proficiency. Real capability comes from changing how the work gets done, not from a one-week class on a chatbot.
• Without a feedback loop between operations and the team building the AI, nobody can say months later whether a workflow is actually improving the outcomes that matter.
• The fix is three structural shifts that reinforce each other: push AI decisions closer to the business, build proficiency through real work, and close the feedback loop early, then keep it tight.

FAQs

Why do AI pilots that work so well often fail to scale?

Because the failure is usually organizational, not technical. If AI decisions route through one central team, if training substitutes for real proficiency, or if there is no feedback loop back to the business, the gains a pilot proved rarely carry into production.

Should a company centralize its AI decisions in one team?

No. Centralizing every AI decision in one gatekeeping team creates a queue the business waits behind. Shared standards should be centralized. The decision to put a specific AI workflow into production should sit with the business unit that owns the outcome.

What is the difference between AI training and AI proficiency?

Training teaches someone to use a tool. Proficiency changes how they do the job. A short training session on a chatbot rarely moves productivity. Asking what a role stops doing and what it takes on instead, then building that into daily work, does.

What happens without a feedback loop between the business and the AI team?

Performance drifts, edge cases pile up, and nobody can prove months later whether the AI workflow is helping. A tight feedback loop, where real users test the workflow on real data and issues go straight into a ranked list for the team building it, catches problems while they are still cheap to fix.

What three changes help AI initiatives scale past the pilot stage?

Push AI decisions closer to the business unit that owns the outcome, build proficiency through real work instead of classroom training, and close the feedback loop between operations and the team building the AI, then keep it tight.

The post Part 2: The AI Productivity Paradox – Common Mistakes in AI Organization Design (and how to fix them)  appeared first on Robots & Pencils.

Part 1: The People | Part 2: The Organization | Part 3: The Foundation

Count the AI tools running in your company right now. The ones actually in use. The Copilot licenses, Enterprise Claude accounts, Salesforce Agentforce. The pilot the finance team built with a contractor. The thing the marketing group pays for on a corporate card. The internal assistant somebody in operations put together over a long weekend. 

If you got past ten, you’re typical. If you can name who owns each one, what it costs, and whether it’s working, you’re rare. Atlassian’s 2026 research found that only 6% of executives can point to organization-wide AI ROI, and the reason isn’t that the tools don’t work. It’s that there is nothing to measure them with. Twelve tools, twelve logins, twelve vendors, no shared memory. Every pilot starts from zero. The company gets faster at individual tasks and no smarter as an organization. 

Why this matters more than it did a year ago 

Something shifted in the middle of 2026 that most executive teams haven’t fully absorbed: the frontier models became close to interchangeable. The gap between the leading providers narrowed to the point where, for most business tasks, which model you use matters far less than what you’ve built around it. The models are becoming commoditized. Prices fell. Switching got easier. The model has become something you rent. 

That changes what the durable asset is. When the model is a commodity, the thing you wrap around it, your business context, the guardrails, your way of knowing whether it’s working, is the only part that appreciates. 

The question for executive teams isn’t “which AI should we buy?” It’s “what should we own, regardless of what we buy?” Engineers call this the harness. We’ll call it the foundation. You rent the model. You own the foundation. 

What the foundation is, in plain terms 

Strip the architecture diagrams away and the foundation is five things. None of them is exotic. Most companies have a version of each for their financial systems and none for their AI. 

A shared definition of your business. What “a customer” means in your company. What “an order” is, what “a batch” is, what “on time” means, and the history of each, because most systems of record overwrite yesterday and remember nothing. Written down once, in a form every tool can use, so the finance team’s assistant and the operations team’s forecast are talking about the same thing. Without this, every AI tool learns your business from scratch, and each one learns it differently. 

One door. Every AI request in the company goes through a single point, so cost, usage and risk show up in one place. Without this, you cannot answer the CFO’s question about what AI is costing you, and you cannot swap vendors without rewriting everything that touches them. 

A register. A list of every AI tool running in your company, bought or built, each with a named owner. Without this, you don’t know what you’re governing. Most companies discover the size of their AI footprint the first time something goes wrong. 

A way to know if it’s working. Before anything goes live, and every week after. Not a vendor’s demo; your own test, on your own data, against what happens. Without this, “the pilot worked” is an opinion, and six months later nobody can say whether the thing in production still does. This is your evaluation framework. 

Approval in proportion to risk. Who signs off on what, at what level of consequence, with the same rules for the tools you bought as for the tools you built. Without this, governance either doesn’t exist or exists as a committee everything waits behind. 

That’s it. Five things. If you have them, you can add the eleventh tool in a week and know what it costs and whether it works by the second week. If you don’t, the eleventh tool is another island. 

A nervous system, not a headquarters 

Here is where executives get worried, and rightly. “Own the foundation” sounds like “centralize AI,” and centralizing AI is one of the fastest ways to fail. We wrote about that in the second article in this series: the AI Center of Excellence that becomes the place where the business goes to wait. 

The foundation is not that. Think of it as a nervous system, not a headquarters; it’s the core of what makes your business, yours. A nervous system doesn’t decide where the hand goes. It makes sure the hand can feel, that the signal gets back to the brain, and that the whole body knows what the hand just learned. The center owns the foundation, the shared definitions, the one door, the register, the tests, the rules. The business units own the decisions: what to build, what to buy, whether it goes live in their operation. Standards are shared. Decisions are local. 

Get this distinction wrong in either direction, and your AI initiatives won’t scale. Centralize all decision making and you have a bottleneck. Decentralize your foundation and you have twelve islands with twelve definitions of a customer. The companies getting results have done the unglamorous thing: they’ve been disciplined about what’s shared and explicit about what’s not. 

The tools you already bought count 

Most organizations are not in the business of building an agent-building program. You’ve bought licenses. That doesn’t exempt you from defining the foundation; it’s the strongest argument for it. 

Governance that only covers what you build misses most of what you run. The Copilot seats, the vendor’s embedded AI, the SaaS tool that quietly added a model last quarter: all of it consumes your data, produces outputs your people act on, and costs your organization money, and almost none of it shows up on the same budget line item as the internal pilot.  

The register is how it gets there. The one door is how you see what it costs, the tokenomics. The test is your evaluation framework to find out whether the expensive bundle you renewed in January is moving you closer to your north star. 

This is the accountability answer for a company that has bought AI and can’t see the return. You don’t need to have built anything to need a foundation. You need one because you bought things. 

What it looks like when it’s done right 

One company we worked with, a manufacturer with operations across several regions, had systems of record that kept no history. Each month overwrote the last. That’s common, and it’s fatal for AI, because a model that can’t see yesterday can’t learn anything about tomorrow. 

The first thing the foundation did was remember. Before any forecast, before any agent, the team built a place where the business’s own history accumulated: what was ordered, what shipped, what the plan said versus what happened. Alongside that history came a connector into the main system of record, a way to test outputs against reality, a handful of reusable patterns for how AI intakes data and asks a human for approval, and a written record of every architectural decision and why it was made. 

None of those was the deliverable. The deliverable was a way for their users to interact with the data that was locked behind unavailable systems to the business unit. When the second use case came along, completely unrelated to this specific function, it needed almost none of that built again. When the business wanted to expand into another region, the only thing that changed was the connector. Each use case shipped faster than the last. That compounding of information is what the foundation is for, and it’s the same story we told in the second article about pilots that leave something behind. This is what “behind” looks like. 

Governance that doesn’t mean slow 

One more thing the foundation does, and it’s the one that makes the rest survivable: it lets you put the controls where the consequences are, instead of everywhere. 

Not every AI decision needs a human gate. An assistant that drafts an internal summary needs a basic check and nothing more. A model that changes a price, approves a stage in your workflow or touches a customer, needs an independent second opinion, ideally from a different system than the one that produced the answer, and a named person who signs off. The gates go at the high-consequence moments, not spread evenly across every step so that everything moves at the speed of the slowest approval. 

That’s what most AI governance gets wrong. It treats every use the same, which means either everything is slow or nothing is checked. Matching oversight to risk is the only approach that is sustainable, and you can only pull it off if the foundation exists. You need the register to know what’s running, the one door to see it and manage the costs, and the evaluation framework to judge its performance. 

The muscle to build 

If you intend to run many AI systems, whether you build them or buy them, the capability your organization needs to develop isn’t building more agents. Agents are getting easier to build. The capability you must develop to scale AI usage is the foundation: knowing what you’re running, knowing what it costs, knowing whether it works, and knowing who decides. Evaluation, governance, and the foundation they run on. 

That’s the muscle we’ve learned to prioritize before anything scales — it’s the difference between compounding every new AI effort and starting from zero each time. 

Questions to take back 

1. Who owns your business context and where does it live? If the answer is “in the systems” but the systems keep no history, or “in the data team” but every tool defines a customer differently, you don’t own it yet. 

2. If you swapped AI partners tomorrow, what would break? If the answer is “everything,” you’ve been building on rented land.  

3. Which of your AI tools could we measure this quarter? Cost, usage and whether it’s working? If the answer is “the one we built,” you’re governing a fraction of what you run. 

Request an AI Briefing

AI has already changed how your people work. Request an AI Briefing with Robots & Pencils to see exactly where your organization sits on the productivity paradox, and what to fix first, across the people, the organization, and the foundation.

About the Author

Brendan Flynn is SVP, Strategist at Robots & Pencils where he heads industry strategy within the Generative & Agentic AI Studio.

Key Takeaways

• Most companies cannot say what AI tools are running, who owns them, or whether they work. Atlassian’s 2026 research found only 6% of executives can point to organization-wide AI ROI, largely because there is nothing in place to measure it with.
• Frontier AI models are becoming commodities. The gap between leading providers has narrowed enough that the model matters less than what a company builds around it.
• The foundation is five things: a shared definition of the business, one door every AI request goes through, a register of every AI tool with a named owner, a way to test whether each tool is working, and approval sized to the risk of what it touches.
• The foundation works like a nervous system, not a headquarters. Standards, shared definitions, the one door, the register, and the tests stay centralized. Decisions about what to build or buy stay with the business units.
• A company does not need to build its own AI to need a foundation. Buying tools is reason enough, since governance that only covers what a company builds misses most of what it actually runs.
• Companies with a foundation in place can add a new AI tool in about a week and know its cost and performance within two. Companies without one start every new tool from zero.

FAQs

What does it mean to own the foundation instead of the model?

AI models are becoming commodities as leading providers converge in capability, so switching between them is getting easier. What a company builds around the model, its business definitions, its governance, its way of testing outcomes, is the part that keeps its value no matter which model runs underneath it.

What are the five parts of an AI foundation?

A shared definition of core business terms, a single point every AI request goes through, a register of every AI tool with a named owner, a repeatable way to test whether a tool is working, and an approval process sized to the risk of what the tool touches.

Does a company need a foundation if it only buys AI tools and does not build any?

Yes. Governance that only covers tools a company builds misses most of what it actually runs, since most AI tools in a typical company are bought, not built. A foundation gives a company a way to see cost, usage, and performance across every tool, regardless of where it came from.

Does owning the foundation mean centralizing every AI decision?

No. The foundation centralizes standards, shared definitions, and the way tools are tested and tracked. Decisions about what to build, what to buy, and when it goes live in a specific business unit stay local to that unit.

What is the payoff of having a foundation in place?

A company with a foundation can add a new AI tool and know what it costs and whether it works within about two weeks. Without one, every new tool starts from zero and stays an island, disconnected from everything the company has already built.

The post Part 3: The AI Productivity Paradox – Own the Foundation, Rent the Model  appeared first on Robots & Pencils.