SimplifyX Unveils AI Agent Layer to Automate Legacy CRM Workflows

Idea to Execution: Crafting AI Agents That Work With SimplifyX™ — Ai4

Ai4

Enterprise software suites like Salesforce and Microsoft Dynamics are shifting into passive backend mainframes as autonomous multi-agent networks take over front-end operational execution.

AI-assisted summary of Idea to Execution: Crafting AI Agents That Work With SimplifyX™ — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published by Ai4 on YouTube on August 24, 2025, this presentation features Deepak Bandi, EVP and GM of SimplifyX™, introducing the company's agentic AI platform. The session focuses on transforming fragmented manual enterprise workflows into intelligent, automated systems operating across existing SaaS stacks.

## TL;DR Deepak Bandi argues that legacy enterprise CRMs like Salesforce and Microsoft Dynamics are evolving into backend systems of record, functioning much like modern mainframes. Using a typical procurement scenario, he illustrates how operations teams still suffer from high friction due to manual phone calls, PDF quote parsing, and spreadsheet wrangling. SimplifyX solves this by deploying a multi-agent orchestration layer—featuring specialized Voice, Web, Doc, UI, and Chat agents managed by a master Workflow Agent—to execute end-to-end tasks on top of existing enterprise tools. This approach enables deep workflow automation across regulated sectors like banking, healthcare, manufacturing, and retail without ripping out core IT infrastructure.

## Atmosphere & style The presentation delivers a fast-paced, interactive conference vibe, kicking off with lighthearted CRM logo trivia and branded swag giveaways to warm up the crowd. Bandi maintains an engaging, pragmatic tone, blending sharp critiques of SaaS license bloat with relatable humor about how modern enterprises still run entirely on Excel. The talk feels dynamic and solution-oriented, smoothly pivoting from operational frustrations to an ambitious architectural pitch for autonomous multi-agent systems.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) Legacy enterprise SaaS and CRM platforms like Salesforce, Microsoft Dynamics, Pega, and ServiceNow are shifting from front-end engagement layers to backend systems of record, functioning as modern mainframes. - [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) Enterprises struggle with bloated custom builds, high licensing costs, and shelfware, often paying for software where 90% of features remain untouched. - [02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) Despite major investments in modern software, critical enterprise workflows still rely on disjointed, manual routines involving spreadsheets, email inboxes, and phone calls. - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) SimplifyX deploys a multi-agent orchestration layer directly on top of legacy CRMs and ERPs without requiring companies to replace their existing software investments. - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) The Workflow Agent acts as the central orchestrator, delegating multi-step operational tasks across specialized AI agents. - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) Specialized Voice Agents autonomously place outbound calls to vendors and negotiate terms based on incoming order requirements. - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) Web Agents browse external sources to assess vendor credibility and reviews, while Doc Agents ingest and extract structured data from PDF quotes received via email. - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) UI Agents dynamically generate front-end interfaces on demand, while Chat Agents supply direct access to internal enterprise knowledge bases. - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) SimplifyX targets high-compliance and regulated sectors including banking, healthcare, manufacturing, and retail.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) The Modern Mainframe: Rethinking SaaS and Enterprise CRMs

The presentation opens on an interactive note as Deepak Bandi quizzes the audience on enterprise CRM logos, rewarding correct answers identifying Salesforce and Microsoft Dynamics with SimplifyX branded water bottles. This leads directly into a broader critique of legacy enterprise software suites like Salesforce, ServiceNow, Pega, and Microsoft Dynamics. Highlighting the provocative assertion by Microsoft's CEO that "SaaS is dead," Bandi clarifies that while these platforms are not vanishing overnight—given the millions of dollars enterprises have invested in them—their role is undergoing a fundamental architectural shift.

Rather than serving as the active front-line engagement layer where day-to-day employees spend their time, these legacy systems are evolving into the modern equivalent of mainframes. Much like the core banking and healthcare mainframes that still run invisibly in the background managed by small, specialized IT teams, CRM platforms will become backend record systems. Bandi points out how platforms originally built as simple sales intake and lead engines became over-engineered custom software monsters, weighed down by exorbitant subscription fees and shelf licenses where companies routinely pay for software while leaving 90% of the features untouched.

- **Legacy CRM giants mentioned:** Salesforce, ServiceNow, Pega, Microsoft Dynamics, and Oracle. - **The Mainframe Shift:** SaaS platforms are moving from the front-end user engagement layer to backend systems of record maintained by dedicated administrative teams. - **Enterprise SaaS friction:** Bloated custom builds, high licensing costs, and massive feature underutilization ("shelfware").

---

[02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) Deconstructing the Friction in Manual Enterprise Workflows

To demonstrate why traditional engagement models are failing, Bandi steps through a concrete packaged goods enterprise scenario involving three distinct personas: the ordering customer, internal operations employees, and third-party delivery vendors. When an incoming purchase order arrives, employees named Jessica and Darren must manage the entire procurement and fulfillment lifecycle through fragmented, manual interventions.

```mermaid flowchart TD accTitle: Legacy Manual Workflow subgraph S1["1. Intake & Entry"] A["Customer Purchase Order"] --> B["Manual Entry in Microsoft Dynamics"] end subgraph S2["2. Vendor Coordination"] B --> C["Manual Outbound Phone Calls"] C --> D["Vendors Email PDF Quotes"] end subgraph S3["3. Processing & Analysis"] D --> E["Download & Data Extraction"] E --> F["Manual Excel Consolidation"] end ```

Jessica manually opens Microsoft Dynamics CRM to create opportunities and quotes from scratch before both employees pick up the phone to dial individual vendors to confirm logistics and scheduling. Once vendors respond with disparate PDF proposals over email, Jessica and Darren are forced to monitor inboxes, download files, manually parse unstructured data, and paste numbers into spreadsheets. As Bandi quips, "without an Excel there is no enterprise company running even today," highlighting how critical business logic in the AI era remains tethered to outdated, high-friction manual routines.

---

[04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) The SimplifyX Agentic Architecture and Future Workflow

To replace disjointed manual touchpoints without ripping out existing investments, SimplifyX introduces a multi-agent orchestration layer that sits directly on top of legacy platforms such as Salesforce, Microsoft Dynamics, Oracle, Pega, and ServiceNow. Rather than requiring workers to navigate complex CRM interfaces, specialized AI agents autonomously coordinate and execute the entire multi-step process from start to finish.

```mermaid flowchart TD accTitle: SimplifyX Agentic Orchestration Layer subgraph User["Inputs & Triggers"] PO["Purchase Order Received"] end subgraph Agents["SimplifyX Agentic Layer"] WA["Workflow Agent<br/>(Master Orchestrator)"] VA["Voice Agent<br/>(Vendor Calls & Negotiation)"] WebA["Web Agent<br/>(Vendor Reviews & Web Lookup)"] DA["Doc Agent<br/>(PDF Quote Parsing)"] CA["Chat Agent<br/>(Knowledge Base Access)"] UIA["UI Agent<br/>(Dynamic UI Generation)"] end subgraph Core["Backend Systems of Record"] Legacy["Legacy Systems<br/>(Dynamics, Salesforce, Oracle, Pega)"] end

PO --> WA WA --> VA WA --> WebA WA --> DA WA --> CA WA --> UIA WA --> Legacy ```

In this autonomous paradigm, receiving a purchase order immediately triggers the Voice Agent to place outbound vendor calls and negotiate terms directly. Concurrently, Web Agents act as autonomous operators browsing external web sources like Google reviews to verify vendor reliability, while Doc Agents monitor email channels to ingest and structure PDF quote data automatically. The UI Agent eliminates bespoke software coding by dynamically generating interfaces—akin to tools like Lovable and v0—while Chat Agents provide direct access to institutional knowledge bases, all unified by the overarching Workflow Agent. Bandi concludes by emphasizing SimplifyX's strong footprint across heavily regulated environments including healthcare, manufacturing, banking, and retail, inviting attendees at booth K2 to transition away from manual CRM and spreadsheet wrangling.

- **Voice Agent:** Initiates automated outbound calls to vendor networks and conducts dynamic negotiation based on order parameters. - **Web Agent:** Autonomously browses search engines and reviews to evaluate vendor quality and reliability. - **Doc Agent:** Continuously monitors inboxes, downloads vendor PDF proposals, and extracts relevant line items and structured quotes. - **UI & Chat Agents:** Generates user interfaces on demand and provides real-time access to organizational knowledge repositories. - **Workflow Agent:** Serves as the central orchestration engine coordinating specialized agents and synchronizing data back to underlying CRM/ERP databases. - **Target Industries:** Deep deployment focus across regulated sectors such as banking, healthcare, manufacturing, and retail.

## Notable quotes > [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) SaaS is dead — Deepak Bandi

> [02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) without an Excel there is no enterprise company running even today — Deepak Bandi

## Mentioned Deepak Bandi, SimplifyX, SimplifyX AI, Salesforce, Microsoft Dynamics, ServiceNow, Pega, Oracle, Microsoft, Lovable, v0, Google

## Video description Deepak Bandi, EVP and GM, SimplifyX™, SimplifyX™

Abstracts

In today’s fast-paced business environment, managing multi-step processes and dynamic decisions across departments is a significant challenge. SimplifyX AI is a powerful Agentic platform designed to tackle these challenges head-on. In this engaging session, we’ll showcase how SimplifyX transforms disjointed manual workflows into seamless, intelligent systems—driving efficiency, transparency, and scalability.

You’ll discover how:

Smart agents automate multi-step tasks—like invoice processing, customer support routing, or strategic planning—by interpreting context, managing dependencies, and learning from outcomes.

Modular orchestration enables teams to combine AI agents for cross-functional workflows without needing code—e.g., initiating outreach, verifying data, and escalating when needed.

Human-in-the-loop controls give teams confidence and visibility, enabling feedback at key steps for trust and compliance.

Real-time monitoring dashboards capture metrics, errors, performance, and ROI so teams can continuously refine operations.

Seamless integrations with GPT-based large language models, internal knowledge bases, SaaS systems, and RPA tools allow deep automation across the entire technology stack.

🔍 Why This Matters Boost productivity by offloading repetitive decision-making and coordination.

Accelerate project outcomes using adaptive, autonomous agents that dynamically adjust plans.

Enhance scaling of workflows across teams without exponential resource costs.

Ensure compliance and accountability via auditable logs and human oversight.

[Source](https://www.youtube.com/watch?v=l9_-mAzQrvY)

中文简报

SimplifyX推出AI Agent层，实现传统CRM工作流自动化 随着自主多Agent网络接管前端业务执行，Salesforce和Microsoft Dynamics等企业软件套件正逐渐转变为被动的后端大型机系统。 Deepak Bandi认为，Salesforce和Microsoft Dynamics等传统企业CRM正演变为后端记录系统，其运作方式极似现代大型机。他以典型的采购场景为例，阐明运营团队目前仍因人工电话沟通、解析PDF报价单以及处理电子表格而承受极高的协作摩擦。SimplifyX通过部署多Agent编排层来解决这一痛点——由核心Workflow Agent统一调度专门的Voice、Web、Doc、UI及Chat Agent，在现有企业工具之上执行端到端任务。该方案无需推翻核心IT基础设施，便可在银行、医疗、制造和零售等强监管行业实现深度工作流自动化。 Salesforce、Microsoft Dynamics 等传统企业 SaaS 与 CRM 正逐渐从前端交互层退化为类似于“现代大型机”的后端记录系统，企业常为此承担高昂许可费用，却有高达 90% 的功能未被使用。 (0:04) 即便企业投入巨资部署现代化系统，采购和履约等关键业务流依然高度依赖电话沟通、邮件 PDF 报价单以及 Excel 手工录入等碎片化的高摩擦日常操作。 (2:22) SimplifyX 在现有 CRM 和 ERP 系统之上构建了多智能体（multi-agent）编排层，无需企业推倒重构已有系统投资，即可接管跨部门业务执行流程。 (4:05) 系统由 Workflow Agent 充当中央编排中枢，协调自主外呼谈判的 Voice Agent、核查声誉的 Web Agent、提取 PDF 报价的 Doc Agent 以及按需生成界面的 UI Agent 协同作业。 (4:05)

Reading summary

Deepak Bandi of SimplifyX argues that legacy enterprise SaaS platforms are shifting from front-line user interfaces into backend systems of record, enabling autonomous multi-agent AI layers to coordinate manual enterprise workflows. Legacy SaaS Shifting to Modern Mainframes (0:04) Bandi contends that legacy enterprise software suites such as Salesforce, Microsoft Dynamics, ServiceNow, Pega, and Oracle are undergoing an architectural shift rather than disappearing. Because enterprises have invested millions of dollars into these systems, Bandi compares their trajectory to core banking and healthcare mainframes: rather than serving as the active front-line engagement layer where staff spend their working hours, they are transitioning into backend systems of record managed by specialized administrative teams. He notes that these suites have evolved into over-engineered systems burdened by steep subscription fees and shelfware, with organizations routinely leaving roughly 90% of licensed features unused. Persistent Friction in Manual Fulfillment Lifecycles (2:22) To demonstrate the breakdown in current engagement models, Bandi outlines a typical packaged goods procurement workflow involving customers, internal operations, and third-party delivery vendors. When an incoming purchase order arrives, employees must manually create opportunities and quotes in Microsoft Dynamics, conduct outbound telephone calls to individual vendors, monitor inboxes for emailed PDF quotes, and manually extract and consolidate unstructured proposal data into spreadsheets. Bandi observes that despite enterprise software investments, critical business logic across enterprises remains tethered to manual interventions and spreadsheet management. Autonomous Multi-Agent Orchestration Architecture (4:05) Rather than replacing existing CRM or ERP databases, SimplifyX introduces a multi-agent orchestration layer directly on top of legacy platforms. The architecture uses a central Workflow Agent to coordinate multiple specialized autonomous agents. Upon receiving a purchase order, a Voice Agent places outbound calls to negotiate logistics directly with vendors, while Web Agents browse search engines and review sites to assess vendor quality. Concurrently, Doc Agents monitor email inboxes to download and extract line items from PDF quotes, UI Agents dynamically generate user interfaces to avoid custom coding, and Chat Agents retrieve institutional knowledge, synchronizing the completed workflow back into underlying enterprise systems. Bandi positions SimplifyX as a solution for heavily regulated sectors—specifically banking, healthcare, manufacturing, and retail—designed to eliminate manual CRM and spreadsheet handling without dismantling established enterprise infrastructure.

精炼阅读摘要

SimplifyX 演讲者 Deepak Bandi 提出，传统企业级 SaaS 正在转变为仅作数据沉淀的现代“大型机”，而企业业务流仍被人工与电子表格割裂；SimplifyX 主张在既有底层系统之上构建多智能体编排层，通过自主协作的 AI 代理接管前端操作。 SaaS 定位转变：从前端操作层退居为后台系统 (0:04) Bandi 援引了微软首席执行官关于“SaaS 已死”的论断，并指出企业在 Salesforce、Microsoft Dynamics、ServiceNow、Pega 和 Oracle 等平台上投入了数百万美元，这些系统不会迅速消失。然而，它们正在经历架构层面的核心重构：这些传统软件不再是员工日常主要交互的前端操作界面，而是退化为类似于银行与医疗核心业务中静默运转的现代“大型机”（Mainframe）。这些原本为简单销售线索管理设计的工具已演化为过度定制的复杂软件，伴随着高昂的许可费用，且有高达 90% 的功能因企业仅将其用作后台记录系统而处于闲置状态。 传统流程摩擦：手动录入与电子表格的数据断层 (2:22) 为了印证传统交互方式的弊端，Bandi 以包装消费品行业的订单履行场景进行了说明。当外部采购订单送达时，内部运营人员必须先在 Microsoft Dynamics CRM 中手动创建商机与报价单，随后通过电话逐一联络外部供应商确认物流排期。收到供应商发送的非结构化 PDF 邮件后，员工需要手动下载文件、比对数据并将其粘贴至 Excel 汇总。Bandi 表示“如今没有 Excel 就没有运转的企业”，这表明即使在人工智能时代，关键业务逻辑与数据流转依然深度依赖繁琐且易出错的手工操作。 多智能体架构：在记录系统之上构建自主编排层 (4:05) 为避免推倒重来已有的巨额 IT 投资，SimplifyX 提出了架设在现有 CRM 与 ERP 之上的多智能体协同网络，由工作流智能体（Workflow Agent）充当中央协调引擎。在订单触发后，语音智能体（Voice Agent）能自主外呼供应商并基于订单参数协商条款；网络智能体（Web Agent）在外部网络及评价渠道审核供应商资质；文档智能体（Doc Agent）自动监听邮件并解析 PDF 提案提取明细；UI 智能体类似 Lovable 或 v0 按需动态生成界面，配合调用内部知识库的聊天智能体（Chat Agent），共同实现跨系统数据自动回写与全流程闭环。 Bandi 强调，SimplifyX 的多智能体技术主要落地于医疗健康、制造、银行和零售等高合规性要求行业，旨在保持原有记录系统底层结构不变的前提下，替代传统跨系统搬运数据的手工操作与电子表格维护流程。

Detailed notes — zh

## Background 该内容由行业峰会频道 Ai4 于 2025 年 8 月 24 日发布在 YouTube 上，主讲人为 SimplifyX 执行副总裁兼总经理 Deepak Bandi。视频简介指出，SimplifyX 致力于打造强大的企业级 AI 智能体平台，旨在通过模块化编排、多智能体协同与人机协同机制，将断裂繁琐的跨部门手动流程改造为端到端自动化的高效智能系统。 ## TL;DR 演讲首先指出传统企业 SaaS（如 Salesforce 和 Microsoft Dynamics）因功能臃肿和高昂维护成本，正在逐步退化为仅供后台存储的“现代大型机”。主讲人以包装消费品行业的订单处理为例，深入剖析了员工在 CRM 录入、人工电话询价、PDF 邮件跟踪及 Excel 离线汇总之间反复搬运数据的现实痛点。对此，SimplifyX 提出了覆盖在底层记录系统之上的智能体架构，通过调度语音、文档、网络、UI、对话和工作流等专用智能体，实现供应商自动谈判、报价解析与数据回填的完整业务闭环。 ## Atmosphere & style 演讲开场借助竞猜 CRM 界面送水壶的小互动迅速活跃了现场气氛，拉近了与观众的距离。随后主讲人直击企业痛点，用“没有任何一家公司能完全脱离 Excel 运转”等接地气的生动表达引发了台下强烈的行业共鸣。整体节奏紧凑轻快、论证清晰有力，在犀利解构传统企业软件困境的同时，展现出智能体重塑未来工作流的务实与前瞻性。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 传统 SaaS 和 CRM 系统虽不会彻底消失，但正在退化为类似于银行底层记录系统的“现代大型机”（Mainframes）。 - [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 随着 AI 智能体接管用户交互层，普通员工不再需要直接操作臃肿复杂的 CRM 界面，SaaS 正在丧失其核心交互入口地位。 - [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 企业正面临严重的 SaaS 功能蔓延与闲置软件许可（Shelf Licenses）浪费，很多系统高达 90% 的采购功能从未被实际使用。 - [02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) 传统跨部门履约流程高度断裂且低效，极度依赖人工录入 CRM、逐个拨打电话、手动下载邮件 PDF 附件并使用 Excel 离线汇总。 - [02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) 在 AI 时代，企业必须淘汰依赖人力反复搬运数据的运作模式，利用智能体架构重构端到端业务流。 - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) SimplifyX 构建了覆盖在 Oracle、Salesforce、Dynamics 等底层记录系统之上的企业级智能体架构（Enterprise Agentic Layer）。 - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 语音智能体（Voice Agents）能够实现高并发自主外呼，甚至可直接与外部供应商进行多轮价格谈判及交期确认。 - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 文档智能体与网络智能体协同运作，可秒级提取邮件报价 PDF 结构化数据，并自主抓取分析供应商网络口碑与质量评价。 - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) UI 智能体（UI Agents）能够根据实时业务需求动态生成轻量交互界面，颠覆耗时费力的传统定制软件开发模式。 - [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 工作流智能体（Workflow Agents）作为中枢统一调度各类专用智能体，全面赋能医疗健康、制造、银行等强监管行业实现闭环自动化。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) SaaS 的未来演进：从交互层退化为“现代大型机” 演讲开场时，主讲人通过向现场观众提出两个关于 CRM 平台的看图竞猜互动来活跃气氛，猜对的观众可以获得来自 SimplifyX 的纪念水壶。在观众先后认出 Salesforce 和 Microsoft Dynamics 之后，主讲人顺势切入现代企业软件的现状：当大家谈起 ServiceNow、Salesforce、Pega 和 Microsoft Dynamics 这类主流企业级 CRM 与 SaaS 平台时，必须面对一个现实——微软 CEO 曾提出 "SaaS is dead"（SaaS 已死）的论断。 关于这句话背后的深层含义，SimplifyX 的核心观点是：这些 CRM 与 SaaS 系统绝不会凭空消失，毕竟各大企业已经砸下了数百万美元的重金建设这些平台。但它们的角色正在发生颠覆性的演变——它们将逐渐退化为类似于银行或医疗系统中的**“现代大型机”（Mainframes）**。 ```mermaid flowchart TD accTitle: 传统 SaaS 向底层记录系统退化演进图 subgraph Past["过去：SaaS 作为交互与应用层"] A["员工直接交互"] --> B["Salesforce / Dynamics"] B --> C["沉重臃肿的定制化逻辑"] end subgraph Future["未来：智能体作为交互层"] D["员工 / 客户"] --> E["SimplifyX 智能体编排层"] E --> F["底层记录系统 (Mainframe 化)\nSalesforce / Dynamics / Oracle / ServiceNow"] end ``` 这种角色的转变意味着： - **失去交互层地位**：普通员工不再直接登录和操作庞大臃肿的 CRM 界面，未来只需要极少数专业技术人员在后台维护这些系统，就像今天少数团队维护传统大型机一样。 - **功能过度蔓延与膨胀**：以 Salesforce 为例，其初衷只是用于线索录入和报价管理的销售工具，但企业的 IT 部门为了图省事，不断在上面堆叠各种定制化业务应用，最终让系统变得不堪重负。 - **高昂的成本与资源闲置**：企业支付着极其昂贵的 SaaS 订阅费与闲置软件许可（Shelf Licenses），许多公司购买的系统功能有 90% 根本从未被使用，甚至员工都不知道它们的存在。 --- ### [02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) 传统企业业务流程的痛点：以包装消费品行业为例 为了具象化当前企业面临的繁琐手工流程，主讲人以包装消费品（Packaged Goods）行业的订单处理场景为例展开剖析。该业务场景通常涉及三类角色：下单的客户、负责推进履约的企业内部员工（例如案例中的员工 Jessica 和 Taran），以及负责供货与运输的外部供应商。 在传统的工作流程中，当客户提交采购订单（Purchase Order）后，员工面临的是极其断裂且低效的手工操作链条： 1. **手动录入系统**：Jessica 需要在 Microsoft Dynamics CRM 中手动创建销售机会（Opportunity）并生成报价单（Quote）。 2. **人工拨打电话询价**：Jessica 与 Taran 必须逐个给供应商（例如 John）打电话沟通细节，核实能否在次日下午 3 点前按要求送达。 3. **邮件与大文件跟踪**：供应商发来海量包含报价与方案的 PDF 邮件，员工需要时刻盯紧收件箱，逐份手动下载 PDF 附件。 4. **Excel 离线计算汇总**：员工再将 PDF 中的报价数据手动复制并提取到 Excel 表格中进行比对与核算。“在当今的企业界，没有任何一家公司能完全脱离 Excel 运转。” 主讲人强调，在 ChatGPT 与 AI 时代，企业不能再继续沿着这种依赖人力反复搬运数据的老路运作，必须引入自动化的智能体架构来解决这种断裂的业务流程。 --- ### [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) SimplifyX 智能体平台：重塑端到端自动化业务流 针对传统流程的痛点，SimplifyX 构建了一套覆盖在 Oracle、Microsoft Dynamics、Salesforce、Pega 以及 ServiceNow 等底层记录系统之上的**企业级智能体架构（Enterprise Agentic Layer）**。底层系统继续作为稳定记录层存在，而上层的多模态智能体则协同接管所有的业务交互与执行。 ```mermaid flowchart TD accTitle: SimplifyX 智能体平台架构及协作流程 subgraph WorkflowLayer["编排控制层"] WF["Workflow Agent 工作流智能体\n(全局协调与业务流构建)"] end subgraph AgentLayer["能力智能体矩阵"] direction LR VA["Voice Agent\n语音外呼/谈判"] WA["Web Agent\n网络调研/评价分析"] UA["UI Agent\n界面快速生成"] DA["Doc Agent\n文档/PDF 数据提取"] CA["Chat Agent\n企业知识库检索"] end subgraph BottomLayer["底层企业记录系统 (Record Layer)"] direction LR SF["Salesforce"] MS["Dynamics"] ORA["Oracle"] SN["ServiceNow"] PEGA["Pega"] end WF --> VA & WA & UA & DA & CA VA & WA & UA & DA & CA --> BottomLayer ``` 该架构中包含六大核心智能体角色，彼此协同配合： - **语音智能体（Voice Agents）**：订单生成后无需人工打电话，AI 可根据供应商名单发起高并发外呼，甚至具备多轮价格谈判与交期确认的能力。 - **网络智能体（Web Agents）**：类似于自主操作员，能够自主浏览 Google 等公开网络，抓取并分析供应商的网络口碑与客户评价，辅助质量决策。 - **UI 智能体（UI Agents）**：摆脱耗时费力的传统定制软件开发，类似 Lovable 或 v0，根据业务需求动态构建轻量交互界面。 - **文档智能体（Doc Agents）**：监控邮件进件，自动下载供应商报价 PDF，秒级提取关键结构化数据。 - **对话智能体（Chat Agents）**：直连企业知识库，随时随地为流程各环节提供问答支持与规则检索。 - **工作流智能体（Workflow Agents）**：负责中枢编排，串联并调用上述各类专用智能体，实现端到端的业务流程闭环。 在 SimplifyX 赋能的未来工作模式下，收到采购订单后，AI 智能体自主呼叫供应商并完成价格谈判，随后持续监听收件箱、下载并解析报价 PDF，最终将结构化结果回填至后台系统中。该平台广泛适用于**医疗健康（Healthcare）、制造业（Manufacturing）、银行业（Banking）等受强监管行业**以及零售业（Retail），全面替代原先依赖人工操作 CRM 与 Excel 的低效模式。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) SaaS is dead — Deepak Bandi > [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 现代大型机 — Deepak Bandi > [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 失去交互层地位 — Deepak Bandi > [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 功能过度蔓延与膨胀 — Deepak Bandi > [00:04](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=4s) 闲置软件许可 — Deepak Bandi > [02:22](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=142s) 在当今的企业界，没有任何一家公司能完全脱离 Excel 运转。 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 企业级智能体架构 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 语音智能体 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 网络智能体 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) UI 智能体 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 文档智能体 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 对话智能体 — Deepak Bandi > [04:05](https://www.youtube.com/watch?v=l9_-mAzQrvY&t=245s) 工作流智能体 — Deepak Bandi ## Mentioned Deepak Bandi, SimplifyX, Salesforce, Microsoft Dynamics, ServiceNow, Pega, Oracle, ChatGPT, Lovable, v0, Excel, Google ## Video description Deepak Bandi, EVP and GM, SimplifyX™, SimplifyX™ Abstracts In today’s fast-paced business environment, managing multi-step processes and dynamic decisions across departments is a significant challenge. SimplifyX AI is a powerful Agentic platform designed to tackle these challenges head-on. In this engaging session, we’ll showcase how SimplifyX transforms disjointed manual workflows into seamless, intelligent systems—driving efficiency, transparency, and scalability. You’ll discover how: Smart agents automate multi-step tasks—like invoice processing, customer support routing, or strategic planning—by interpreting context, managing dependencies, and learning from outcomes. Modular orchestration enables teams to combine AI agents for cross-functional workflows without needing code—e.g., initiating outreach, verifying data, and escalating when needed. Human-in-the-loop controls give teams confidence and visibility, enabling feedback at key steps for trust and compliance. Real-time monitoring dashboards capture metrics, errors, performance, and ROI so teams can continuously refine operations. Seamless integrations with GPT-based large language models, internal knowledge bases, SaaS systems, and RPA tools allow deep automation across the entire technology stack. 🔍 Why This Matters Boost productivity by offloading repetitive decision-making and coordination. Accelerate project outcomes using adaptive, autonomous agents that dynamically adjust plans. Enhance scaling of workflows across teams without exponential resource costs. Ensure compliance and accountability via auditable logs and human oversight. [Source](https://www.youtube.com/watch?v=l9_-mAzQrvY)

Source quotes

[0:04] Deepak Bandi: SaaS is dead [0:04] Deepak Bandi: 现代大型机 [0:04] Deepak Bandi: 失去交互层地位 [0:04] Deepak Bandi: 功能过度蔓延与膨胀 [0:04] Deepak Bandi: 闲置软件许可 [2:22] Deepak Bandi: 在当今的企业界，没有任何一家公司能完全脱离 Excel 运转。 [4:05] Deepak Bandi: 企业级智能体架构 [4:05] Deepak Bandi: 语音智能体 [4:05] Deepak Bandi: 网络智能体 [4:05] Deepak Bandi: UI 智能体 [4:05] Deepak Bandi: 文档智能体 [4:05] Deepak Bandi: 对话智能体 [4:05] Deepak Bandi: 工作流智能体

Source quotes

[0:04] Deepak Bandi: SaaS is dead [2:22] Deepak Bandi: without an Excel there is no enterprise company running even today

Original transcript

[0:04] Perfect. I I'll start with the two [0:05] questions first and whoever uh answers [0:08] will get a bottle from Simplify Extreme [0:11] here. So, can someone name what CRM [0:13] platform this is? Salesforce. Okay, we [0:16] have one winner here. Uh let's see where [0:19] this clicker is. Awesome. Cool. Perfect. [0:21] And the second one. There you go. One [0:24] other person. Anyone? [0:27] Dynamics. Okay. Cool. Perfect. So uh [0:30] when you think of enterprise CRM SAS [0:32] platforms you know these are the common [0:34] ones comes to mind right there is [0:36] service now there's Salesforce there's [0:38] PEGA there is Dynamics but one of our [0:41] own uh the CEO of Microsoft basically [0:45] mentioned SAS is dead I'm sure everyone [0:48] has you know heard about this but this [0:50] has like so many layers of meaning when [0:53] he mentioned SAS is dead our take is [0:56] it's not like these CRM s are going away [0:59] because you know to be honest uh many of [1:02] these enterprise companies have invested [1:03] millions of dollars bringing these [1:05] platforms so they're not really going [1:07] away what they're going to become is the [1:10] main frames so which means this these [1:13] CRM platforms are not the ones where the [1:15] employees of the organizations will be [1:18] interacting with so these won't be the [1:19] engagement layer anymore so you'll have [1:22] like small set of teams who is managing [1:24] the CRM platform very similar to how we [1:26] manage mainframes today I mean if you [1:28] think of any banking industry, [1:30] healthcare industry, everyone has [1:31] mainframe like that's still there. You [1:33] have a small set of team managing that's [1:35] kind of our point of view where the SAS [1:37] platforms are going to go. Now the the [1:39] and the reason for that is think of like [1:42] Salesforce for a second, right? [1:43] Salesforce was built as a sales engine [1:45] or a sales CRM tool where you intake the [1:47] leads uh basically run quotes whatnot. [1:50] But someone in the IT team mentioned oh [1:53] we have Salesforce why can't we leverage [1:56] that to build a prior some custom you [1:58] know application and build that on [2:00] Salesforce. That's basically what [2:01] happened and then of course Salesforce [2:02] expanded every other company expanded [2:04] that that's basically the issue with the [2:06] some of these platforms. Other thing is [2:08] they're very it's a very expensive [2:10] platform. It's very high SAS fees and [2:12] shelf licenses. I mean many people buy [2:14] these platforms 90% of the features you [2:16] don't even use. you don't even know to [2:18] be honest but it's like sitting there [2:19] doing nothing. So let let's pick a use [2:22] case so we can understand this a little [2:24] bit more better. So let's I'm just [2:26] picking a use case of package goods [2:28] here. So when you think of package goods [2:30] there are three persona comes to mind. [2:32] One is a customer. So in this case let's [2:34] say a customer is placing an order. So [2:36] that's basically the persona of the [2:38] customer. Then you have employees [2:40] working at this packaged goods company [2:42] who as soon as you get the purchase [2:43] order then they need to contact the [2:46] these different vendors so they can [2:48] actually deliver the products back right [2:49] so that's kind of the life cycle of the [2:51] use case so we have Jessica and Taran [2:54] who are uh employees working at a [2:56] packaged goods company as soon as you [2:58] get an order Jessica is using Microsoft [3:01] dynamic CRM in this case she goes in [3:04] manually creates an opportunity creates [3:05] a quote like everything is completely [3:07] manual [3:09] Once that is done, they both get on [3:11] phones. They are supposed to now call [3:12] every single vendor saying, "Okay, John, [3:15] can you deliver this package? You know, [3:17] I need to be here at 3 p.m. tomorrow [3:20] making, you know, the these are kind of [3:21] typically how these packaged goods [3:22] company work, right? So, they make their [3:24] own phone calls right now." After that, [3:26] they get a bunch of quotes that in an [3:29] email. So, they get a a large PDFs. All [3:31] of these proposals they're getting in [3:33] emails and Jessica and Darren, they have [3:34] to then manually download this PDF. keep [3:37] watching the you know emails extract all [3:40] of the data and then of course without [3:42] an Excel there is no enterprise company [3:44] running even today without an Excel so [3:45] you have to have a Excel use case so of [3:47] course you bring in all of that data [3:48] into Excel run a code I mean this is [3:50] kind of traditional way of how most of [3:52] these companies work today now again our [3:55] point of view is there has to be a [3:56] better way you can't in the era of chat [4:00] GPT and AI whatnot you can't keep [4:02] running the way the business you know [4:03] the way you're doing it today and our [4:05] solution is this agentic platform [4:07] platform where the the idea here is what [4:11] we have done is we actually kind of took [4:13] all of those enterprise platform if you [4:14] see that underlying layer there Oracle [4:16] dynamic Salesforce PGA uh service now as [4:20] a back to my initial point they're not [4:21] going away anywhere they will stay as it [4:23] is but we're building that enterprise [4:25] agentic layer on top of those for [4:28] example what voice agent does is as soon [4:31] as u let's say as soon as a purchase [4:33] order comes in the AI can actually make [4:35] phone calls to all the relevant vendors [4:37] that you have you can scale that by you [4:40] know so much volume web agents today [4:42] these vendors nobody's actually looking [4:44] at okay let me look into Google and see [4:47] the reviews are people talking good [4:49] about it are these guys delivering [4:50] nobody is looking at reviews at all [4:52] right so web agents think of them as [4:54] operators where it you can it can access [4:57] um kind of browse Google you know look [4:59] into the reviews whatnot UI agents we [5:02] want to get out of this custom software [5:04] building uh you may have heard Lovable [5:07] V0ero where UI agents allow you to build [5:09] the entire interface. On the right side [5:11] here bottom we have doc agents where as [5:13] soon as you get the PDF sorry you get [5:15] the codes the doc agent can extract the [5:17] data. Chat agents if you want chat [5:19] capabilities of course it's no-brainer [5:21] right is nothing but a knowledge base at [5:22] any point of time you want some kind of [5:24] knowledge base right so and finally the [5:26] workflow agent interacts basically with [5:28] all of these agents and basically build [5:29] that whole entire use case entire [5:32] business process for you. Now this is [5:35] kind of how it looks in the future world [5:36] when you use simplify x platform where [5:38] as soon as the um purchase order comes [5:40] in the AI actually makes a call it also [5:43] is uh you know it's uh kind of trained [5:45] to actually even negotiate uh in this [5:46] particular use case once that is done [5:49] it'll keep watching at the email it'll [5:51] download the PDF it'll extract the PDFs [5:53] and bring it back into whatever system [5:55] you're doing. So we are literally [5:56] building the entire platform that [5:58] orchestrates your entire workflow any [6:01] industry that is we we work in [6:02] healthcare we work in manufacturing uh [6:04] we work in banking the couple of mostly [6:07] regulated industries is where our focus [6:09] of course retail is one of them uh [6:11] that's basically what I wanted to talk [6:13] about today again we are in the booth K2 [6:15] so in case if you have any if you're [6:17] using Microsoft using Salesforce or [6:19] Excels come to us and uh hopefully we [6:22] can help. Thank you.