Demand Agents: AI Enters the Enterprise

As the year comes to a close, AI seems to be the only thing analysts and technology vendors want to talk about. This newsletter is no different. Media hype and boardroom pressure are driving a narrative that doesn’t reflect day to day reality in most businesses. Individual users and specific teams may be seeing some success with generative AI. Yet, for all the experiments and pilots, we're still only just getting a picture of where the technology can deliver a return on investment.

Understanding the Limitations

AI deployment is slower than expected because the benefits of the technology are not evenly distributed. AI is great at spreading baseline knowledge across a team and in providing suggestions for troubleshooting. That makes the technology great at raising the output of weak performers. However, for the top performers, the benefits are a lot less predictable. Most subject matter experts simply don’t need the assistance, unless they want to use a chatbot as a sounding board.

Part of the problem is cost. The technology is expensive and unreliable. It requires close monitoring. As a result, many of the promised efficiency gains fail to materialise, which limits how the technology can be used. It makes little sense to use AI workflows for processes that can already be implemented using traditional automation solutions. Some people are starting to use AI for automation, but due to inconsistent output, that is not an ideal solution.

Choosing the Right Model

There are workflows where the technology makes sense, but only if you’re using the right tools. In recent months, Claude and Gemini have become the go-to models for enterprise use cases. ChatGPT works fine as a content generation tool, but has fallen behind in its ability to understand context and parse complex requests. That affects a lot of in-application AI tools, which typically leverage ChatGPT as the foundational model. Microsoft Copilot is a notable example, but it extends to the likes of HubSpot Breeze too.

Regardless, business leaders still demand that AI is use in automation. However, the tools available to operations teams simply aren’t good enough to make that happen. Apps have embedded AI across their entire product portfolio, but such features are rarely used. Specialised AI assistants need dedicated training to be effective, which few developers have bothered with. As a result, many business users would rather copy and paste content from Gemini or ChatGPT instead of using built-in features.

The Rise of Agents

In house agents are emerging as a trend. However, their scope is limited. They exist to collect input for deterministic workflows, or to summarise complex content for quick reference. Think project recaps or internal surveys. In a marketing operations context, that means agents are being used to collect information for campaign intake processes. That allows the agent to fill in the gaps that always exist within any campaign brief. The actual campaign setup is then automated using traditional automation, or carried out manually.

Technology vendors have adapted their tools to this approach. Automation solutions are adding AI decision steps to their platforms. Chatbot builders can be used to collect structured information and kickoff workflows in downstream tools. It’s this mix of capabilities that allows teams to deploy AI into production. For now, these projects are just for internal use. No one trusts AI for customer facing automation.

Corporate Expectations

Meanwhile, Wall Street are incorrectly betting that internally developed agents will replace SaaS software. That’s a losing gamble. Investors massively underestimate the complexity of the average enterprise technology platform. Few businesses will be willing to invest the necessary resources to develop and maintain core platforms in-house. That’s the reason SaaS software took off in the first place. Vibe coding will be used to maintain existing in house point solutions, but not much beyond that.

Nevertheless, many boardrooms are still fixated on the idea that AI can be used to automate end-to-end processes. Unrealistic AI mandates will remain a corporate reality into next year. That needn’t be a problem. The distinction between AI automation and rule based automation is barely understood in many C-suites. Simply deploying more automation should be sufficient to meet the KPIs being passed down from on-high.

Salesforce Expands AWS Partnership

Amazon are the forgotten player in enterprise AI. Their Bedrock platform hosts a wide range of LLMs for use within developer applications and AWS services. Last week, Agentforce became the newest AI service available to AWS Bedrock customers, allowing more flexible LLM choices and joint billing for businesses building AI agents using Agentforce.

Operationalising ABM in the Real World: Your Step-by-Step Guide

Finding it difficult to navigate the complexities of ABM. A new eGuide from CRMT Digital cuts through ABM hype, explaining how to implement and run an ABM program in practical terms. Boost ROI with practical approaches to account-based targeting, personalisation and engagement.

Marketing Operations Roadmap Matrix

A Marketing Operations Roadmap is an essential strategic tool that aligns marketing priorities with your overall business goals. Any roadmap begins with understanding where your marketing organisation is right now, and then defining where you want to get to. This free Marketing Operations Roadmap Matrix from CRMT Digital allows you to do exactly that.