CJ Deploys RapidCanvas AI to Target 75% Client Services Profit Lift
Ai4
CJ CEO Santi Pierini argues that over 80% of enterprise AI projects fail because leaders treat automation as an IT experiment rather than using modular tools to overhaul operational structures within single quarters.
## Background
Published on YouTube by Ai4 on August 27, 2025, this presentation features Mayur Kshetramade and Rahul Pangam of RapidCanvas alongside Santi Pierini of CJ. The session examines how modular, plug-and-play AI architectures can overcome fragmented retail data to optimize demand forecasting, dynamic pricing, and personalized marketing without requiring massive in-house AI teams.
## TL;DR
RapidCanvas COO Mayur Kshetramade opens by detailing how agentic AI and expert-in-the-loop workflows replace fragile spreadsheets with closed-loop demand forecasting, dynamic pricing, and automated operational flywheels. CJ CEO Santi Pierini then joins to explain how the affiliate commerce giant restructured its massive client services division with RapidCanvas to systematize deep institutional domain expertise and target a 75% gain in productivity and profitability. Pierini tackles modern retail fragmentation across retail media networks and social commerce like TikTok Shop, illustrating how AI-driven matchmaking and attribution help brands adapt to creator-led consumer journeys. Ultimately, the executives urge enterprise leaders to abandon sluggish, multi-year IT implementation cycles in favor of fast, single-quarter rollouts strictly anchored to tangible business outcomes.
## Atmosphere & style
The presentation carries an executive, no-nonsense tone that deliberately cuts through speculative AI hype with battle-tested enterprise realism. Transitioning from Mayur's structured case studies into a candid fireside-style discussion with Santi Pierini, the session feels both urgent and pragmatic, enlivened by colorful metaphors like riding the giant sandworm of Dune to tame retail disruption. It conveys the authentic tension of leading through a generational technology wave, capturing the dual sensation of executive terror and intense commercial excitement.
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=e3NDdNbytm8&t=4s) RapidCanvas combines an agentic AI platform with human experts in the loop to build tailored enterprise solutions and deliver measurable business outcomes.
- [02:18](https://www.youtube.com/watch?v=e3NDdNbytm8&t=138s) Market leaders adopting AI grow 2 to 2.5 times faster by maximizing predictive enterprise data and expanding margins without increasing headcount.
- [02:18](https://www.youtube.com/watch?v=e3NDdNbytm8&t=138s) AI-powered demand forecasting and dynamic pricing with margin guardrails achieve double-digit stockout reductions and protect profitability.
- [05:09](https://www.youtube.com/watch?v=e3NDdNbytm8&t=309s) Agentic AI workflows automate complex accounting write-backs and supply chain planning, driving up to 90% planning accuracy and an 80% reduction in spreadsheet work.
- [07:22](https://www.youtube.com/watch?v=e3NDdNbytm8&t=442s) Enterprise leadership at Publicis Groupe and CJ evaluates AI proposals strictly as core business transformation initiatives rather than standalone technology experiments.
- [10:25](https://www.youtube.com/watch?v=e3NDdNbytm8&t=625s) CJ partners with RapidCanvas to systematize veteran institutional expertise across client services, targeting a 75% increase in productivity and profitability.
- [11:28](https://www.youtube.com/watch?v=e3NDdNbytm8&t=688s) Commerce fragmentation across retail media and social channels makes AI essential for creator matchmaking, multi-touch attribution, and dynamic commissioning.
- [16:25](https://www.youtube.com/watch?v=e3NDdNbytm8&t=985s) Generational technology waves require leaders to embrace both immense excitement and the urgency to adapt rather than being disrupted.
- [17:21](https://www.youtube.com/watch?v=e3NDdNbytm8&t=1041s) Overcoming the high failure rate of enterprise AI requires democratizing tools across non-technical teams and strictly anchoring projects to concrete KPIs.
- [19:28](https://www.youtube.com/watch?v=e3NDdNbytm8&t=1168s) Replacing multi-year IT implementation roadmaps with rapid, single-quarter rollouts prevents AI solutions from becoming obsolete before delivering ROI.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=e3NDdNbytm8&t=4s) RapidCanvas Overview and the Power of Hybrid AI Transformation
Mayur Kshetramade, Chief Operating Officer at RapidCanvas, opens the session by outlining the company's core mission as a trusted AI transformation partner. RapidCanvas distinguishes itself through a practical hybrid model that merges an agentic AI platform with a dedicated "expert in the loop" approach. The engagement lifecycle begins with an immersive AI workshop where RapidCanvas meets with key customer stakeholders to diagnose specific business pain points and co-create an actionable transformation roadmap. From there, the team moves swiftly into designing and deploying tailored AI solutions alongside customer domain experts, focusing strictly on delivering measurable business outcomes.
RapidCanvas has built substantial operational momentum over its three years in operation. Backed by top-tier venture capital firms including Accel, the company is growing at approximately 4x year-over-year. Its leadership team brings extensive enterprise credibility, collectively holding six IPOs and five strategic exits, including a prior company sale to PayPal by one of the co-founders.
- RapidCanvas combines proprietary agentic AI workflows with human domain experts to accelerate deployment.
- Company leadership track record includes 6 IPOs and 5 strategic company exits.
- Mayur frames the session to cover field-tested AI use cases in retail, e-commerce, and CPG, followed by an executive discussion featuring Rahul Pangam (CEO of RapidCanvas) and Santi Pierini (CEO of CJ, the affiliate marketing platform powering over $20 billion in retail sales).
---
[02:18](https://www.youtube.com/watch?v=e3NDdNbytm8&t=138s) Decision Intelligence: From Demand Forecasting to Dynamic Pricing
AI has transitioned from an exploratory laboratory experiment into a proven revenue and profit engine deployed across pricing, revenue management, inventory planning, marketing, supply chain, and product development. Industry data reveals that forward-leaning change leaders effectively adopting AI are outpacing their peers by growing 2 to 2.5 times faster. These market leaders generally divide their focus across two primary vectors: maximizing the predictive power of existing enterprise data and utilizing agentic AI to augment human teams—driving margin expansion and operational scale while keeping headcount flat.
```mermaid
flowchart LR
accTitle: Decision Intelligence Pillars
subgraph DataSources["Enterprise Signals"]
A["CRM & In-Market Data"]
B["Competitor Price Feeds"]
end
subgraph UseCases["AI Decision Engines"]
C["Confidence Demand Forecasting"]
D["Precision Targeted Marketing"]
E["Margin-Guarded Dynamic Pricing"]
end
subgraph Impact["Measurable Business Outcomes"]
F["Double-Digit Stockout Reduction"]
G["Higher Retention & Campaign ROI"]
H["Protected Margins & Top-Line Lift"]
end
A --> C
A --> D
B --> E
C --> F
D --> G
E --> H
```
The first major decision engine focuses on demand forecasting with high confidence, replacing cumbersome ERP systems and static spreadsheets with live CRM and in-market signals. Deploying this approach has unlocked double-digit reductions in both overstock and understock conditions, directly translating into millions of dollars in annual bottom-line savings. In customer acquisition and marketing, deep learning models analyze customer behavior, recency, and competitor movements to execute targeted outreach, generating substantial lift in retention and campaign ROI. In dynamic pricing, AI integrates real-time inventory velocity, competitor pricing dynamics, and automated margin guardrails to ensure products are competitively positioned without eroding profitability.
- Demand forecasting models leverage live market signals to yield millions of dollars in inventory savings.
- Marketing automation engines analyze behavioral recency and competitor tactics to elevate campaign conversion rates.
- Dynamic pricing systems balance competitive positioning with hard profit-margin rules to avoid value-destructive discounting.
---
[05:09](https://www.youtube.com/watch?v=e3NDdNbytm8&t=309s) Agentic Workflows and the End-to-End Manufacturing Flywheel
Beyond pure decision-support systems, enterprise organizations are leveraging agentic AI to replace tedious, manual spreadsheet processes with automated workflow orchestration. In finance and accounting departments—where teams often drown in ledger reconciliations and file mapping—AI models leverage fuzzy logic to match complex vendor invoices against payment data. Crucially, the system does not stop at matching; it autonomously writes back and updates the core accounting books, dramatically accelerating month-end closes while eliminating human error. Similarly, in customer support and operations, AI agents triage incoming calls, emails, and voicemails, routing tickets instantly to specialized support or billing teams and enabling organizations to handle a 2x ticket volume with identical staffing levels.
```mermaid
flowchart TD
accTitle: Manufacturing and Supply Chain AI Flywheel
subgraph PlanningCycle["Forecasting & Production"]
A["AI Demand Forecasting"] --> B["Packaging & Production Scheduling"]
end
subgraph ExecutionCycle["Procurement & Logistics"]
B --> C["Raw Materials Inventory Check"]
C --> D["Automated Vendor Procurement"]
D --> E["Inbound Tracking & Schedule Adjustment"]
end
E --> A
```
In integrated supply chain and CPG environments, AI creates a closed-loop operational flywheel. The cycle begins with demand forecasting, which feeds directly into daily production and packaging scheduling. The platform continuously monitors raw material inventory, autonomously triggers purchase orders through procurement systems, tracks inbound delivery times, and dynamically reschedules manufacturing lines if component delays occur. This continuous feedback loop has demonstrated up to 90% planning accuracy alongside an 80% reduction in manual spreadsheet management.
- Finance teams use fuzzy-logic agents for end-to-end invoice reconciliation and automated ledger updates.
- Customer support operations achieve 2x ticket throughput without expanding headcount via intelligent AI routing.
- The manufacturing flywheel coordinates forecasting, raw material procurement, and line scheduling to attain 90% planning accuracy.
---
[07:22](https://www.youtube.com/watch?v=e3NDdNbytm8&t=442s) The Adoption Playbook and Building Executive Conviction
To implement AI successfully, Mayur outlines a four-step framework: start small with pilot programs tied directly to explicit business outcomes; rigorously prioritize use cases based on business impact and deployment velocity; ensure cross-functional alignment between business and IT stakeholders early to foster shared accountability; and focus on empowering business operators directly rather than creating bottlenecks around scarce data science and engineering resources.
Mayur then welcomes Rahul Pangam (CEO of RapidCanvas) and Santi Pierini (CEO of CJ) to discuss how executive leadership navigates AI hype and builds conviction around strategic bets. Rahul notes that executive teams are inundated with infinite AI possibilities, making prioritization difficult. Santi explains that within CJ (over 1,000 employees) and its parent organization Publicis Groupe (~110,000 employees), leadership refuses to evaluate proposals as mere "AI initiatives." Instead, every proposal must be framed strictly as a comprehensive business transformation initiative across sales, engineering, or client services.
In CJ's client services division—which represents approximately two-thirds of the company's total headcount—senior team members hold deep domain expertise developed over 5 to 10 years. Scaling the business historically meant adding headcount, but institutional domain expertise could not scale at the same rate. Santi highlights that CJ partners with RapidCanvas to capture, systematize, and scale this institutional intelligence across the entire enterprise, allowing junior staff to operate with the proficiency of veteran domain experts.
- Successful AI rollouts start with prioritized, high-impact pilots owned cross-functionally by business and IT teams.
- Publicis Groupe (110,000+ staff) and CJ evaluate all proposed AI projects strictly as business transformations rather than technical experiments.
- Two-thirds of CJ's organization sits in client services, making the systematization of 5-to-10-year domain knowledge via RapidCanvas a central driver of scalable growth.
[10:25](https://www.youtube.com/watch?v=e3NDdNbytm8&t=625s) AI as a Business Transformation Accelerant
Rather than treating artificial intelligence as an isolated technology experiment, CJ has integrated it directly into their operational restructuring. The company segmented its entire client services department into specialized units—dividing teams into roughly six distinct operational, publisher-growth, and client-growth groups, each with deep domain expertise. By pairing domain-specific personnel with automated workflows, the organization is targeting a 75% increase in productivity and profitability. The underlying philosophy is that AI must not function as "a hammer looking for a nail," but rather serve as an enabler and accelerant for broader structural change, matching the accelerating pace of modern business.
[11:28](https://www.youtube.com/watch?v=e3NDdNbytm8&t=688s) Riding the Dune Sandworm: Solving Modern Retail Disruption
While retailers have operated on razor-thin margins and faced technological disruption for decades, modern commerce has introduced unprecedented fragmentation through retail media networks (such as Amazon, Walmart, and Target) and social commerce platforms like TikTok Shop. Santi Pierini views AI not as the source of disruption, but as the essential cure. Borrowing an analogy from *Dune*, he explains that businesses can either let the giant sandworm consume and devour them, or learn to ride it to drive massive growth.
```mermaid
flowchart LR
accTitle: Modern Affiliate and Influencer Attribution Journey
subgraph Discovery["Consumer Need & Discovery"]
A["Consumer Problem<br/>(Plantar Fasciitis)"] --> B["Trusted Creator / Influencer<br/>(Patrick Mouratoglou)"]
end
subgraph Purchasing["Omnichannel Purchasing"]
B --> C1["Nike.com D2C"]
B --> C2["Retail Media (Amazon)"]
B --> C3["Social Commerce (TikTok Shop)"]
end
subgraph Attribution["AI-Powered CJ Engine"]
C1 & C2 & C3 --> D["Matchmaking, Attribution & Dynamic Commissioning"]
end
```
To illustrate the evolution of influence and commerce, Pierini points to long-time CJ client Nike—the pioneer of influencer marketing with Michael Jordan and Tiger Woods—and shares a personal anecdote. As an avid weekend tennis player dealing with arch and heel pain from plantar fasciitis, Pierini researched footwear solutions and initially purchased a pair of Nike Vapor shoes directly on Nike.com. However, modern journeys are far more fragmented: a buyer might see a recommendation from a trusted authority like tennis coach Patrick Mouratoglou on an Instagram Reel, or choose to fulfill the order via Amazon or a TikTok Shop.
Where legacy affiliate models tracked a few hundred trusted bloggers and cashback sites using basic spreadsheets, the ecosystem now spans tens of thousands of creators and hundreds of thousands of influencers across platforms like YouTube and Instagram. Managing this scale requires AI to handle core programmatic tasks:
- Matchmaking the right products and campaigns to the ideal creators.
- Tracking multi-touch attribution across fragmented retail media and direct-to-consumer endpoints.
- Calculating dynamic commission payouts based on granular conversion sources.
- Accurately forecasting campaign performance and consumer demand.
Underpinning this entire technological mesh is consumer trust. While retail has evolved from local butchers and bakers a century ago to digital influencers today, AI serves as the matchmaking engine that connects trusted advice to seamless purchasing experiences.
[16:25](https://www.youtube.com/watch?v=e3NDdNbytm8&t=985s) Generational Transformation: Navigating Excitement and Fear
Reflecting on recent discussions with 22 retail partners and enterprise customers in New York, Pierini observes overwhelming optimism across the market. He emphasizes that navigating this generational shift requires a healthy duality: "If you're not both extremely excited and terrified at the same time, you're probably not looking at the right things and not doing the right things day to day." Similar to previous step-function technology waves—the rise of the commercial internet, e-commerce, mobile computing, and social networks—AI represents an industry-defining transition.
[17:21](https://www.youtube.com/watch?v=e3NDdNbytm8&t=1041s) Practical Strategies for Enterprise AI Adoption and ROI
Addressing the historical reality that over 80% of enterprise AI initiatives struggle to demonstrate positive ROI due to poor change management and execution bottlenecks, Pierini outlines foundational strategies to translate executive optimism into measurable operational results:
- **Democratize Immediate Adoption:** Encourage experimentation across non-engineering divisions, including client services, implementation, and operations, rather than siloing tools within technical teams.
- **Avoid Reinventing Existing Tech:** Rather than building workflow infrastructure from scratch, leverage proven specialized platforms (such as CJ partnering with RapidCanvas) while remaining collaborative across the broader model ecosystem (OpenAI, Perplexity, Microsoft Copilot, Google Gemini).
- **Enforce Strict KPI Anchoring:** Reject vanity projects and demand concrete, testable business metrics tied to productivity, revenue, and client outcomes.
- **Compress Delivery Timelines:** Move away from the legacy consulting paradigm of multi-year, hundred-million-dollar ERP implementations—recalling Pierini's early career at Accenture deploying SAP—in favor of rapid, single-quarter overhauls of complete client service delivery models.
[19:28](https://www.youtube.com/watch?v=e3NDdNbytm8&t=1168s) Making AI a Business Enabler and Closing Remarks
Delivering targeted, agile value with AI is far more compelling and effective than getting bogged down in drawn-out, multi-year IT initiatives. A major risk with long-horizon implementations is that by the time an organization finally reaches the finish line, the market dynamics and underlying business needs have already shifted, rendering the anticipated benefits obsolete before they can ever be realized. Focusing on fast, iterative implementations ensures that organizations see tangible value quickly and stay aligned with current market conditions.
AI must ultimately serve as a true enabler of core business goals rather than being treated as "the thing" itself or siloed into an isolated side project. To drive real business impact, AI tools must directly support strategic outcomes rather than existing for technology's sake. Concluding the session, the hosts thank Santi Pierini for sharing his practical insights on driving AI adoption and value across retail and commerce.
- **Agile execution over long roadmaps:** Multi-year implementation cycles risk obsolescence, as market conditions often evolve before delivery.
- **AI as a strategic enabler:** Successful adoption treats AI as an instrument to achieve concrete business objectives rather than a standalone novelty or side initiative.
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=e3NDdNbytm8&t=4s) expert in the loop — Mayur Kshetramade
> [07:22](https://www.youtube.com/watch?v=e3NDdNbytm8&t=442s) AI initiatives — Santi Pierini
> [10:25](https://www.youtube.com/watch?v=e3NDdNbytm8&t=625s) a hammer looking for a nail — Santi Pierini
> [16:25](https://www.youtube.com/watch?v=e3NDdNbytm8&t=985s) If you're not both extremely excited and terrified at the same time, you're probably not looking at the right things and not doing the right things day to day. — Santi Pierini
> [19:28](https://www.youtube.com/watch?v=e3NDdNbytm8&t=1168s) the thing — Santi Pierini
## Mentioned
RapidCanvas, CJ, Ai4, Mayur Kshetramade, Santi Pierini, Rahul Pangam, Accel, PayPal, Publicis Groupe, Amazon, Walmart, Target, TikTok Shop, Dune, Nike, Michael Jordan, Tiger Woods, Patrick Mouratoglou, Instagram, YouTube, Nike Vapor, Accenture, SAP, OpenAI, Perplexity, Microsoft Copilot, Google Gemini
## Video description
Mayur Kshetramade, Chief Operating Officer, RapidCanvas
Santi PIerini, CEO, CJ
Rahul Pangam, CEO, RapidCanvas
Abstracts
Retailers are drowning in fragmented data and disconnected tools. This talk explores how modular, AI-powered solutions are transforming retail decision-making—from forecasting demand to optimizing pricing and personalized marketing. I will share how leading brands are reducing stockouts, boosting ROI, and making faster decisions—without the overhead of in-house AI teams. CEO of CJ, Santi Pierini and CEO of RapidCanvas, Rahul Pangam, will share best practices on how to make the case for AI and drive adoption with high impact. You’ll learn what’s working today (and what’s not) across D2C, marketplaces, and omni-channel retail. Join me to see how plug-and-play AI can finally deliver real results—no legacy BI required. Let’s get real.
CJ部署RapidCanvas AI，目标提升客户服务利润75%
CJ首席执行官Santi Pierini指出，超过80%的企业AI项目之所以失败，是因为领导层将自动化视作IT实验，而非使用模块化工具在单个季度内重构运营架构。
RapidCanvas首席运营官Mayur Kshetramade首先详细介绍了智能体AI与专家在环工作流如何替代脆弱的电子表格，实现闭环需求预测、动态定价及自动化运营飞轮。随后，CJ首席执行官Santi Pierini阐述了这家联盟电商巨头如何借助RapidCanvas重构其庞大的客户服务部门，将深厚的机构领域专业经验系统化，并力争实现生产力与盈利能力提升75%的目标。Pierini还深入探讨了零售媒体网络以及TikTok Shop等社交电商导致的现代零售碎片化挑战，说明AI驱动的撮合匹配与归因分析如何帮助品牌适应创作者主导的消费者旅程。最后，高管们呼吁企业领导者放弃周期长达数年的迟缓IT实施模式，转向严格锚定明确业务成果的单季度快速落地方案。
RapidCanvas 结合自主 Agentic AI 平台与行业专家闭环协作模式（expert-in-the-loop），帮助企业构建定制化解决方案并推动可衡量的业务成果。 (0:04)
率先应用 AI 的领先企业通过高置信度需求预测与设有利润护栏的动态定价，实现两位数的缺货率下降，业务增速达到行业同侪的 2 至 2.5 倍。 (2:18)
Agentic AI 工作流正深度介入财务发票核对与端到端供应链排产，不仅使传统表格操作减少 80%，还达成了高达 90% 的排产计划准确率。 (5:09)
阳狮集团（Publicis Groupe）与 CJ 将 AI 项目严格评估为核心业务转型而非技术实验；CJ 携手 RapidCanvas 系统化沉淀资深专家的行业知识，目标将客户服务团队的生产力与盈利能力提升 75%。 (7:22)
面对零售媒体网络与社交电商渠道的日益碎片化，AI 正在扮演创作者精准匹配、多触点归因与动态分佣计算的核心引擎，帮助品牌应对现代商业重塑。 (11:28)
Reading summary
Enterprise artificial intelligence generates durable commercial returns only when deployed as an operational transformation rather than an isolated technical experiment, according to executives from RapidCanvas and CJ. Market leaders are bypassing multi-year IT deployments in favor of modular, KPI-driven agents that augment human teams and scale domain expertise.
Decision Intelligence and Agentic Workflows in Operations (2:18)
AI applications in enterprise environments have moved beyond experimental pilots into operational decision engines. RapidCanvas Chief Operating Officer Mayur Kshetramade notes that industry leaders adopting AI are growing 2 to 2.5 times faster than competitors by using existing data to drive margin expansion while keeping headcount flat. In retail and supply chain contexts, live market and CRM signals replace static spreadsheets to inform demand forecasting, delivering double-digit reductions in overstock and understock conditions and saving millions in inventory costs. Dynamic pricing engines incorporate real-time inventory velocity and competitor monitoring alongside automated margin guardrails to protect profitability.
Automating End-to-End Workflows in Manufacturing and Finance (5:09)
Beyond predictive recommendations, autonomous agentic workflows are replacing repetitive manual administrative tasks. In finance departments, fuzzy-logic AI agents reconcile complex vendor invoices against payment data and autonomously update core accounting ledgers, accelerating month-end closes. Customer support teams use intelligent routing to double ticket throughput with existing staffing levels. In manufacturing and consumer packaged goods, integrated flywheels connect demand forecasts directly to daily production scheduling, raw material checks, automated purchase orders, and dynamic rescheduling for delivery delays—achieving up to 90% planning accuracy and an 80% reduction in manual spreadsheet management.
Navigating Retail Fragmentation Across Media and Creator Channels (11:28)
Modern commerce faces compounding fragmentation from retail media networks like Amazon, Walmart, and Target alongside social commerce channels like TikTok Shop. CJ Chief Executive Officer Santi Pierini argues that managing this landscape requires AI to handle tasks once managed via spreadsheets across small networks of bloggers. With affiliate ecosystems now encompassing hundreds of thousands of digital creators, AI serves as the core matchmaking and attribution infrastructure. The engine pairs products with relevant influencers, tracks multi-touch customer journeys across direct-to-consumer and retail media endpoints, calculates dynamic commission payouts, and forecasts campaign demand while maintaining consumer trust.
Scaling Institutional Knowledge Through Operational Restructuring (10:25)
To counter the reality that over 80% of enterprise AI projects fail to generate positive ROI due to execution bottlenecks, CJ restructured its client services division—representing two-thirds of its 1,000-person workforce—into specialized units. Historically, client services scaled strictly through linear hiring, which could not replicate the institutional intelligence of personnel with 5 to 10 years of experience. By partnering with RapidCanvas to embed domain expertise directly into software workflows, junior staff can operate with veteran proficiency, supporting CJ's target of a 75% increase in productivity and profitability.
Agile Deployment Strategies Over Multi-Year IT Roadmaps (17:21)
Pierini and RapidCanvas CEO Rahul Pangam emphasize that enterprise leadership, including CJ's parent firm Publicis Groupe (comprising roughly 110,000 employees), evaluates proposals strictly as commercial transformations rather than technology exercises. Pierini cautions against the traditional consulting model of multi-year, hundred-million-dollar ERP rollouts, noting that shifting market dynamics routinely render long-horizon projects obsolete before completion. Successful enterprise adoption requires democratizing tools across non-technical operators, avoiding building commoditized tech from scratch, anchoring every pilot to explicit commercial KPIs, and executing over single-quarter deployment timelines.
The presented framework positions modular AI as an operational necessity for managing fragmented modern retail, but its success depends on rigorous change management. Organizations must resolve the operational friction of upskilling non-technical staff and replacing legacy spreadsheet habits with automated workflows before these productivity gains can be fully realized.
[0:04] Mayur Kshetramade: expert in the loop
[7:22] Santi Pierini: AI initiatives
[10:25] Santi Pierini: a hammer looking for a nail
[16:25] Santi Pierini: If you're not both extremely excited and terrified at the same time, you're probably not looking at the right things and not doing the right things day to day.
[19:28] Santi Pierini: the thing
Original transcript
[0:04] U hi I'm mayor I'm chief operating
[0:06] officer at Rapid Canvas. Rapid Canvas is
[0:09] a trusted AI transformation company that
[0:12] has a unique hybrid approach of
[0:14] combining the power of our agentic AI
[0:17] platform with what we call as expert in
[0:20] the loop approach and we we do it very
[0:22] practically. This is all about helping
[0:24] our customers uh to deliver measurable
[0:27] outcomes through AI solutions. So we
[0:30] typically start off with what we call as
[0:32] AI workshop where we meet our customer
[0:35] stakeholders, understand their
[0:36] challenges and come up with a
[0:39] transformation roadmap as such and then
[0:41] we very quickly move into uh developing
[0:44] designing the solutions using our
[0:46] platform and our experts and of course
[0:49] experts from our customers as well and
[0:51] it is all about delivering the outcomes
[0:53] that can be measured. Um rapid canvas
[0:56] has been around for about 3 years. We
[0:58] are backed with some of the top tier VCs
[1:00] including Axel. Um we have we are
[1:03] growing thanks to our customers about 4x
[1:05] every year and amongst our leadership we
[1:09] have about six IPOs, five strategic
[1:12] exits including one of our founders last
[1:15] exit to PayPal. Anyway, what we're going
[1:17] to cover today is u we are going to I'm
[1:20] going to give you a walkthrough of few
[1:22] practical use cases in AI that we have
[1:25] used with our clients that we have
[1:27] deployed with our clients in retail uh
[1:29] e-commerce and CPG space. U I'm also
[1:33] going to share some tips of what we have
[1:35] learned how to start, how to accelerate,
[1:38] how to scale and so on. And then I'm
[1:40] going to invite uh two CEOs on the
[1:43] stage. one Rahul Pangam CEO of Rapid
[1:46] Canvas and then Santi CEO of CJ. CJ is
[1:49] world's largest affiliate marketing
[1:51] platform uh driving retail sales up to
[1:54] 20 billion dollars$2 billion dollars
[1:56] actually uh for top retailers in the
[1:58] world and Rahul and Santi are going to
[2:00] give us more tips around how to really
[2:02] drive a business case internally within
[2:05] organization for AI solutions and you
[2:08] know hopefully and not hopefully I'm
[2:10] very sure they're going to leave us with
[2:12] lot of inspiration to take action so
[2:14] with that in mind let's jump into it
[2:18] AI in retail, CPG and e-commerce, it's
[2:21] not uh an experiment anymore. It is a
[2:24] true revenue and profit engine. It's
[2:25] being deployed across all functions. Be
[2:28] it uh pricing and revenue department, be
[2:30] it be demand forecasting, be it
[2:32] inventory planning, be it marketing, be
[2:34] it supply chain, be it product
[2:37] development, you know, it's it's all all
[2:39] over and it's making an impact. And in
[2:42] fact the adopters, the change leaders
[2:44] who are using AI effectively are
[2:47] actually growing two to two and a half
[2:49] times faster than their peers. This is
[2:52] the research.
[2:54] So what are these change leaders exactly
[2:58] doing? And I'm going to walk you through
[3:00] a few use cases. But the way most of our
[3:03] customers in this space are approaching
[3:05] problem is in two ways. Number one, they
[3:08] are focusing on how to use their
[3:10] existing data to maximize its potential
[3:13] to drive the most intelligence out of
[3:15] it. And the second group is how to use
[3:20] agentic AI to complement what humans are
[3:23] already doing and then drive scale,
[3:25] drive efficiency
[3:27] um and drive stronger ROI, profit
[3:30] margins and so on by keeping the
[3:32] headcount flat. So let's jump into the
[3:34] first three use cases that are more
[3:36] around driving smarter business
[3:38] decisions and growing what matters most.
[3:41] Uh one of my favorite ones is about
[3:43] demand forecasting with confidence. you
[3:45] know moving away from lot of SAP tools,
[3:48] ERP tools, Excel spreadsheets and really
[3:51] really using the data you got on
[3:53] historical CRM in market signals and so
[3:56] on and allowing for a accurate demand
[3:59] forecasting and one of our customers who
[4:01] has done that has not only seen a
[4:03] double-digit reduction in overstock and
[4:06] and undertocking which has like strong
[4:08] financial impact and they have seen
[4:10] actually millions of dollars in annual
[4:11] savings. One of our customers is has
[4:14] deployed AI in marketing by really
[4:16] understanding the customer signals,
[4:18] their behavior, their recency, inmarket
[4:21] signals, what a competitors are doing
[4:24] and so on and creating a targeted
[4:25] outreach program that allows them not
[4:28] only to boost their retention of
[4:30] existing customers but acquire more and
[4:32] more customers more effectively and see
[4:35] a very high lift in the campaign ROI.
[4:39] Another use case which I really like is
[4:41] also about dynamic pricing. How to use
[4:43] signals of your own data, what's
[4:45] selling, what's not, how to see what
[4:48] understanding and combining that with
[4:49] what competitors are doing at any point
[4:51] of time. And then really also applying
[4:54] your smart margin rules so that you're
[4:56] not selling something at a loss all the
[4:58] time and bringing all of that together
[5:00] with the power of AI to really drive
[5:02] revenue increase and also driving
[5:05] customer retention in the process.
[5:09] Next three use cases are all about AI
[5:11] agents. How you use workflow
[5:13] orchestration, agentic AI and really
[5:16] moving away from spreadsheets after
[5:18] spreadsheets after spreadsheets. I'm
[5:21] sure a lot of people know what I'm
[5:22] saying here, right? And and one of my
[5:24] use cases that I really like is actually
[5:26] in finance department where lot of
[5:28] financing people in accounting and are
[5:30] really busy in reconciliation of books
[5:33] of mapping one file to another and so
[5:35] on. And one of my favorite use cases,
[5:37] all of these are, by the way, my
[5:38] favorite use cases. I love all of them.
[5:40] But one of this is about how one of our
[5:42] customers is using uh AI and fuzzy logic
[5:46] to match invoices to payments data and
[5:50] not only stopping at matching but also
[5:52] taking a step further and updating the
[5:54] accounting books. So when you do it end
[5:56] to end through AI, you not only now have
[5:59] very high matching accuracy but also
[6:01] reduction in lot of manual work and
[6:03] books are getting closed very fast.
[6:06] Another one is in customer support and
[6:08] operations where they are using AI to
[6:11] categorize all of these calls and emails
[6:14] and voicemails that are getting through
[6:16] for their customer support center and
[6:18] doing using AI to really triaging it
[6:20] very quickly understanding if it's a
[6:22] tech support if it's a account issue if
[6:24] it's related to pricing whatever it is
[6:27] and having the right people work on it
[6:29] faster and really saving time and
[6:31] headcount being able to scale they are
[6:33] they are doing two more 2x X of their
[6:36] support tickets with the same uh staff
[6:39] that they have. And the last example is
[6:41] about how you can use AI right from
[6:44] creating a flywheel. You start from
[6:46] demand forecasting to go into really
[6:49] understanding which packages need to
[6:50] start manufacturing today versus
[6:52] tomorrow. Then using AI to understand do
[6:55] you have all the raw materials ready
[6:57] where you need to order some of those.
[6:59] doing that automatically through
[7:01] procurement, understanding when they're
[7:03] going to come in, changing the packaging
[7:05] and scheduling if you need to, and
[7:07] coming back into demand forecasting and
[7:09] doing that over and over again. So
[7:11] that's like a fantastic use case that
[7:13] has led to 90% planning accuracy and 80%
[7:17] less use of spreadsheets and stronger
[7:19] and more confident decisions. So how do
[7:22] you then you know get working and what
[7:25] are some of the tips? First tip is start
[7:27] small, scale fast. Start with a pilot
[7:30] program, something that directly ties to
[7:32] a business use case.
[7:34] Prioritize the what matters. We always
[7:37] work with our people, with our customers
[7:39] during this workshops and rank all the
[7:42] use cases based on different priorities.
[7:44] What is a higher business impact? What
[7:45] can be done faster versus not and so on.
[7:48] Make it crossunctional. The more
[7:50] business and technology teams are
[7:52] working together right from the
[7:53] beginning, not only you're creating
[7:55] momentum for success, but you are
[7:57] getting shared accountability right at
[8:00] the beginning. build a clear business
[8:02] use case which is what Santi and Rahul
[8:05] are going to help us with that and with
[8:07] the thought of empowering more business
[8:09] leaders because now with AI you do not
[8:12] have to keep having this bottleneck of
[8:15] not having data scientists not having
[8:16] engineers and actually empower business
[8:18] users to do a lot more so I'm very
[8:21] excited to have Rahul and Santi two CEOs
[8:24] join us and and help us understand this
[8:26] a little bit more and give us more tips
[8:28] Rahul and Santi please
[8:31] So, Santi, thank you for taking the time
[8:32] to join us on stage here. Uh, hopefully
[8:35] your mic is on as well.
[8:36] >> Good.
[8:38] >> Um, you know, as we talk to clients, one
[8:40] of the things we hear consistently is
[8:42] there is so much of inputs people are
[8:46] getting around AI, what AI can do, where
[8:48] it can, it almost seems like there's
[8:50] limitless possibilities.
[8:52] And you as a CEO, how do you build
[8:54] conviction around where you want to make
[8:55] the big bets? And when your team comes
[8:58] to you with ideas of oh we should try AI
[9:00] in this department or that function or
[9:03] that area how do they kind of articulate
[9:07] that in a way that helps build
[9:08] conviction. So just trying to understand
[9:10] how do you get to this uh what are the
[9:12] right ways to use AI what are the right
[9:15] use cases the right uh business areas
[9:18] and how do you build that conviction
[9:20] >> that's good so first of all we don't
[9:22] look at them as AI initiatives we look
[9:24] at them as business initiatives and in
[9:27] every area of the company I should note
[9:29] too that publicist group is about
[9:31] 110,000 employees CJ's over a thousand
[9:34] and every employee has an awareness of
[9:36] AI and how it might help facilitate what
[9:39] they're trying to accomplish. But when
[9:40] they bring us a an initiative, it's
[9:42] about a business transformation. Sales
[9:44] and marketing could be in our client
[9:46] success group, could be in our
[9:47] engineering team. And so I'll give you
[9:50] one example. Probably one of the most
[9:51] impactful is around our client services
[9:54] organization. It's about twothirds of
[9:57] the employees are delivering client
[9:59] services. Um, brilliant people that have
[10:01] been with the company for a long time, 5
[10:03] to 10 years, domain expertise that's
[10:05] been compiled. The problem is if we try
[10:07] to grow our organization, we have to
[10:10] grow our headcount. We have to grow the
[10:11] number of people, which is fine, but we
[10:13] can't necessarily grow that expertise
[10:15] along and keep pace with. So, we use AI
[10:18] and we're working with rapid canvas for
[10:20] this to generate all that intelligence
[10:22] inside our organization and make it more
[10:25] operational through the use of AI in the
[10:27] platform. So, for example, we've
[10:29] segmented all of our client services
[10:31] teams into specialists. And there's
[10:34] specialists that are operationally
[10:35] focused. There are specialists that are
[10:38] uh more publisher and um growth focused.
[10:41] There are client growth teams. So we've
[10:43] broken into about six pieces all with
[10:45] their own domain expertise. And now
[10:46] we're able to grow and get much more
[10:48] efficient and trying to tie all that
[10:50] back to KPIs. We're we're shooting for a
[10:51] 75% increase in productivity, profit,
[10:55] and so on through the use of AI. But
[10:57] again, it's not an AI project per se.
[10:59] It's a business transformation project
[11:01] enabled by AI.
[11:03] >> Got it. So, not just think about AI as
[11:05] like a hammer looking for a nail, but
[11:07] like you know, think about the broader
[11:09] business transformation. And where does
[11:11] AI enable you to to make that more
[11:13] faster, better, more realistic?
[11:16] >> It's an accelerant. It's it's an
[11:18] enabler. Um the pace of business is
[11:20] faster and and AI is partly responsible
[11:22] for that increase in pace. So, it's the
[11:25] natural enabler to help you keep pace
[11:26] with that change. Yeah.
[11:28] >> Perfect. Um Santi you travel around the
[11:30] globe mayor spoke about the retail
[11:33] revenue support you meet a lot of CEOs a
[11:35] lot of teams uh they are doing their own
[11:37] AI right all the retailers and uh your
[11:40] partners and customers and as you talk
[11:43] to them what are some of the things
[11:44] you're hearing where you think they feel
[11:46] it's truly disruptive uh in their
[11:49] business uh especially in retail
[11:51] >> it's good um retailers have been
[11:53] disrupted by technology forever right
[11:56] it's the uh it's the low margin piece
[11:58] where they're trying to match consumer
[12:00] needs, consumer demand to the products
[12:01] and services they have. Um, I tend to
[12:04] look at AI in retail not necessarily as
[12:07] the disruptor, but as the cure for the
[12:09] disruption that other channels are
[12:11] starting to enforce into retail. And
[12:13] I'll give you some examples. They're
[12:14] dealing with retail media networks like
[12:16] Amazon, Walmart, and Target. They're
[12:18] dealing with social commerce around Tik
[12:20] Tok shops. They're dealing with all
[12:21] these changes, but AI is the most
[12:24] logical enabler to be able to fix some
[12:26] of those problems. I'm going to borrow
[12:27] steel an analogy from a good friend of
[12:29] mine where you can look at if you're
[12:31] familiar with the Dune movies, the Dune
[12:33] book. Um, you can you can look at that
[12:35] giant sandworm as someone that can
[12:38] consume you and devour you or someone
[12:40] that you can ride and be part of and
[12:42] really enable growth and change. And so
[12:44] in our world, I look at and I'll give
[12:46] you an example. So Nike is a customer of
[12:49] CJ's has been in from the inception. If
[12:51] you think of Nike, they're really the
[12:52] original influencing marketing team.
[12:54] They they understood that Michael Jordan
[12:56] and uh and Tiger Woods could sell a lot
[12:59] more product than they could just by
[13:00] designing cool cool shoes. So Nike in a
[13:04] traditional affiliate marketing world
[13:06] and these are examples of all the
[13:07] disruptive um changes that are happening
[13:09] in that field. I'm I'm going to use a
[13:11] personal example. I'm a big weekend
[13:13] tennis player. I've been playing for
[13:15] many years. Love the sport. But I
[13:17] started to develop arch pain and heel
[13:19] pain. And just like anyone in this room,
[13:21] you'd go online, start to say,"Well, why
[13:24] do I have heel pain? Why am I starting?"
[13:25] Well, planter fasciitis may be a
[13:27] problem. What can I do about that? Well,
[13:29] there's lots of um techniques. You can
[13:31] use orthotics, but really the best bet
[13:33] is to use a much, you know, a tennis
[13:35] shoe that's designed to support your
[13:38] arch better. So, long story short, I
[13:40] took that recommendation. I followed the
[13:42] link back to nike.com. I bought a bought
[13:44] a pair of Nike Vapor 12s and lo and
[13:47] behold planer fasciitis got better.
[13:49] That's a commerce journey that was we're
[13:51] all familiar with that we've done many
[13:53] many many times but now a disruptive
[13:55] technology comes in. I want to use this
[13:58] influencer that I'm very you know I
[14:00] trusted source of information a guy
[14:02] named Patrick Monlue. He used to be the
[14:04] coach for Serena Williams. Great with
[14:05] product, great with technique. Well, he
[14:07] recommended the same shoes. So if I
[14:09] clicked on his real link or his
[14:11] Instagram link and went back to Nike and
[14:13] bought, somebody had to facilitate that
[14:16] transaction and understand that the
[14:17] source was now an influencer, that's CJ,
[14:20] you know, all along paying commissions,
[14:21] making sure that happens. What if I had
[14:23] chosen though to buy from a retail media
[14:25] network like Amazon? Again, consumer
[14:28] wherever they are to wherever they want
[14:29] to buy. We had to facilitate that. What
[14:32] if it was a Tik Tok shop and the same
[14:33] type of influence? So you're starting to
[14:35] understand it gets more and more and
[14:36] more complex,
[14:38] but the premise is still the same
[14:40] trusted source of information that I
[14:42] believe will give me good good advice
[14:44] and feedback all the way back to where I
[14:46] want to purchase, how I want to
[14:48] purchase. And if we look at AI as an
[14:50] enabler in that world, um, taking it
[14:52] back to the CJ example, the first
[14:54] scenario, we've been doing that for over
[14:56] 20 years. We've had trusted sources,
[14:58] bloggers, people that uh that were
[15:00] trusted in terms of their their advice,
[15:02] comparison sites, cashback, normal, we
[15:05] can do that. And it maybe involves
[15:07] hundred uh adi trusted advice in that
[15:10] category. You can manage that on a
[15:12] spreadsheet. Back to my ear's point. Now
[15:13] you start to add tens of thousands of
[15:16] influencers to this and hundreds of
[15:18] thousands of creators through every
[15:20] mechanism from YouTube to to all the
[15:22] social media channels. Without AI, we
[15:24] could not matchmake the right product to
[15:27] the right uh campaign to the right
[15:30] influencer or creator ever. We couldn't
[15:33] measure the results of those campaigns.
[15:35] We couldn't tie the right commission
[15:37] base back to that. We couldn't forecast
[15:38] these campaigns. That's all driven by
[15:40] AI. So, in this case, we're writing that
[15:43] sandworm to help us solve all of these
[15:45] problems rather than being consumed by
[15:46] it.
[15:47] >> As a fan of the Dune movies, this is an
[15:50] amazing analogy, right?
[15:52] >> Thank you for that, Jordan. By the way,
[15:53] the analogy is good. And that's my I'll
[15:55] just finish with the one thing that AI
[15:57] can't replace is trust. And that has
[16:00] always been the basis of any consumer
[16:02] journey. A hundred years ago, you'd go
[16:03] to a merchant that you trusted. You go
[16:05] to a baker that you trusted. You go to a
[16:06] butcher you trusted. Same thing is
[16:08] happening online. And AI just helps us
[16:10] find and matchmake trusted sources back
[16:14] to purchases that we want to make. And
[16:15] I'm oversimplifying a little bit, but
[16:17] that's our job is to make that happen
[16:18] and make it seamless and to feel like a
[16:20] very personal experience even though
[16:22] there's a lot of technology and
[16:24] expertise going into that. Yeah.
[16:25] >> So there's a lot more sort of promise
[16:27] and and hope and optimism and how this
[16:29] is going to be beneficial rather than
[16:32] sort of overcoming the anxiety of like
[16:34] what this is going to do to my business.
[16:35] >> 100%. We were at a industry conference
[16:38] last week in New York and I met with
[16:39] over 22 22 partners, customers, and
[16:43] every one of them were more excited than
[16:45] afraid. And this is a phrase I like to
[16:46] use. If you're not both extremely
[16:48] excited and terrified at the same time,
[16:50] you're probably not looking at the right
[16:52] things and not doing the right things
[16:53] dayto day. And I think we should all be
[16:54] excited and a little bit terrified. But
[16:57] it's a great opportunity to do things.
[16:59] This is transformative. This is
[17:01] generational transformation that we're
[17:02] seeing right now. And I' I've personally
[17:04] never been more excited about being part
[17:06] of it.
[17:07] >> Yeah. And uh you know we have these
[17:09] waves right like step function changes
[17:11] in in technology whether it's the
[17:14] internet when everything went online and
[17:16] e-commerce whether it was mobile with
[17:18] the iPhone and Android whether it's
[17:19] social
[17:21] >> um and you have ideas you have theory
[17:25] you have approaches you have optimism.
[17:28] How do you convert that to realism where
[17:30] the rubber meets the road? and you know
[17:32] what are the tips for
[17:35] making it adoption successful. So I've
[17:37] been in AI for a long time like last 15
[17:39] years and we know that 80 plus% of
[17:42] projects have a hard time meeting ROI
[17:44] simply because not because the idea is
[17:46] bad or it's not the right thing to do
[17:48] for the business but just the
[17:50] implementation adoption change
[17:51] management all those things kind of get
[17:53] to be tricky right so what advice would
[17:56] you have for companies as they embark I
[17:58] mean I don't think anybody's not going
[18:00] to embark on transformation right so how
[18:02] do you make that real how do you make it
[18:04] uh successful how do you get the right
[18:06] outcomes uh in terms of adoption.
[18:09] >> So jump in, start now uh if you haven't
[18:12] already. Um
[18:14] it's a lot of fun. The entire
[18:16] organization can participate. You're
[18:17] going to see change in areas you never
[18:19] anticipated. Um it won't just be
[18:21] engineering. It can be client service.
[18:23] It can be your implementation teams
[18:25] partner where it makes sense. Um we've
[18:28] been involved in AI at CJ for over two
[18:30] years. We've done some great internal
[18:32] projects we built, but we turned to
[18:33] Rapid Canvas when it came time for all
[18:35] of the workflows and all that. We don't
[18:36] need to reinvent technology that already
[18:38] exists in the marketplace. There's a lot
[18:40] of trusted advisors. Most of my
[18:41] conversations last week were
[18:43] collaborative where we would learn more
[18:45] about what they're doing with Copilot or
[18:47] Gemini. We tell them what we're doing
[18:48] with OpenAI and Perplexity. And it's
[18:50] become a much more strategically uh you
[18:54] know cohesive conversation. um make
[18:58] everything based on real business
[19:00] outcomes so that you're not looking at
[19:02] wow AI is fun we should we should do a
[19:04] project because it's interesting and
[19:05] exciting of course but make sure there's
[19:07] KPIs make sure there's business results
[19:09] that you're demanding and test and
[19:11] measure constantly um I my first role
[19:14] out of college was with Accenture and it
[19:16] was not unusual to do a four-year
[19:18] project hundreds of millions of dollars
[19:19] to implement SAP I think going forward
[19:22] that's going to be you have one quarter
[19:24] to completely transform perform your
[19:26] entire client services delivery and
[19:28] you're going to do that using AI. That's
[19:30] a lot more interesting than, you know, a
[19:32] multi-year project. And I think you can
[19:34] see the benefit. A lot of times those
[19:35] long projects, the benefit was already
[19:38] changed. The market's changed before you
[19:40] ever get to the end of that journey
[19:41] where you've implemented. Um, and then I
[19:43] say make AI an enabler, not not just the
[19:47] thing, not just a side project that
[19:48] you're trying to do. It has to be an
[19:50] enabler for your business goals. It's
[19:51] probably more than three things, but
[19:54] >> I appreciate it. Thank you so much
[19:55] Santi.
[19:56] >> Thank you all.
