Your marketing team’s ability to take advantage of AI agents depends less on the agent tools themselves and more on whether your existing platforms have the API infrastructure to support them. A new public dataset note 152 B2B APIs on their agent readiness, giving you a framework to identify where your stack can support automation and where it might be holding you back.
This matters a lot if you plan to replace some SaaS with a local, ambient-coded version. Many companies are doing just that, leading to growing concerns about a SaaSpocalypse. And for good reason. Earlier this year, about $300 billion in market value evaporated in software stocks after Anthropic released Claude Code and Claude Cowork, tools that allow AI agents to create and modify software autonomously. Deloitte, citing Gartnerpredicts that 35% of point product SaaS tools will be replaced by AI agents or absorbed into larger agent ecosystems by 2030.
“Vibe coding will lead to a renaissance of a million new SaaS applications. It’s (sic) already started,” said Jason Lemkin, founder of SaaStr. wrote on LinkedIn. But the flip side is that platforms with weak, human-focused APIs are most at risk. “The ‘SaaS apocalypse’ narrative is more about acceleration than extinction,” Jen Grant, CMO of conversational AI platform Quiq, told CMSWire. “The systems of record – your CRM, your data warehouse, your core operating platforms – aren’t going away. What’s changing is the layer above them.”
Scott Brinker, editor-in-chief of Martech Landscape and chiefmartec, and Frans Riemersma, founder of MartechTribe, offer a more measured view of their annual report. Martech Report for 2026. “Big changes are happening across the stack and in the marketplace. AI agents are real, and they’re reshaping the way marketers market and the way buyers buy. But hype still outpaces reality,” they write. “You should lean in and learn. But maintain a healthy skepticism of the AI equivalent of ‘crazy-eyed guns’ pushing for a radical rebuild of your entire stack. Experiment boldly, but evolve wisely.”
Their data backs it up: 90.3% of marketing teams already use AI agents somewhere in their stack, but companies are most often running agents integrated into their existing platforms (68%), not replacing them.
Check your own battery
For a marketing operations manager, the report card is more useful as a diagnostic framework than as a shopping list. Instead of asking whether to replace a platform, the most productive questions are:
- Does my current marketing automation platform have API limits that an agent could navigate in minutes during a segmentation workflow?
- Can my stack send real-time updates via webhooks, or will an agent need to constantly poll?
- Can I safely test agent workflows in a sandbox without risking live data?
- If an agent reattempts a failed action, will they create duplicate contacts or records?
These are criteria to include in a supplier renewal conversation. You don’t need to be a developer to ask them. You just need to know that the questions matter.
The layer moving above them is increasingly AI agents. And which platforms survive this change comes down to one factor: agent-friendliness of the API.
SaaStr’s AI Agent API Report Card evaluates 152 B2B APIs based on their ability to support autonomous agents. The thesis is simple, as SaaStr founder Jason Lemkin explains: After running more than 20 AI agents in production for 18 months, “the most important variable in determining whether a vendor stays or goes hasn’t been the user interface. It hasn’t been the price. It hasn’t been the brand. It’s the API.”
This means that the platforms at the bottom of the balance sheet – those whose APIs are the most difficult for an agent to use – are also the ones most directly exposed as the agent ecosystem evolves.
And the names at the bottom are not random.
Marketo scores 50 out of 100. Gainsight scores 47. Workday (a common data source for marketing operations) scores 42. ActiveCampaign scores 53. Mailchimp scores 57.
“They built empires with a human user interface as their product,” Lemkin wrote. “The API was an afterthought. The brand was the screen the rep logged in to every morning. It worked for 15 years. It doesn’t work now.”
Here’s what happens with several categories of solutions:
Marketing Automation
The marketing automation category is the clearest example of this. Marketo’s API scores five out of 10 for API design, five for webhooks, five for rate limits, five for SDKs and documents, and just four out of 10 for agent readiness. Marketo’s REST API allocates 50,000 calls per day, but that’s still modest by agent standards: a single agent running a segmentation workflow could get through that in hours, not days. The SaaStr Bulletin’s agent rating puts it bluntly: “Launchpoint’s outdated API architecture…and complex authentication make it one of the least agent-friendly marketing platforms.” »
Compare this to the AI and LLM APIs at the top of the list. OpenAI scores 90. Anthropic scores 90. These companies create APIs specifically for programmatic access because agents are their primary distribution channel.
Customer success
Customer success platforms face similar exposure. Gainsight scores 47 (C), with a 3 out of 10 for SDKs and Documents and a 4 out of 10 for Agent Readiness. Freshdesk obtains a score of 58 (C+).
The bright spot in the category is Intercom Fin, with a score of 70 (B) and a score of 9 out of 10 for agent preparation. It’s no coincidence that it’s the only platform in the category that has explicitly built an AI agent platform for third-party developers.
Business intelligence
Business intelligence tells a more nuanced story. Clay scores 75 (B+) with a 9 out of 10 in agent preparation – Lemkin called him “practically a visual agent builder.” But Gong scores 63 (B-), Outreach scores 65 (B-), and Hunter.io scores 60 (B-). The tools that sales teams rely on for pipeline intelligence and prospecting are not yet designed for the agents who will increasingly do this work.
RCMP
CRM is doing better but still has pockets of vulnerability. HubSpot gets an 80 (A-), Salesforce a 75 (B+). But Zoho CRM scores 57 (C+) and Freshsales scores 55 (C+). The gap between the category leaders and the rest is large – and growing.
The metric that matters most
Perhaps the most telling metric in the entire report card is officer readiness, one of six criteria evaluated.
Sales intelligence platforms average a rating of 7.0 out of 10 when it comes to agent readiness, which is decent, but significantly outpaced by Clay’s rating of 9. Marketing platforms only have a rating of 6.1 on average. Customer success averages 6.3.
Your customers are searching everywhere. Make sure your brand introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
@media (maximum width: 768 px) { .headline-responsive { font-size: 30px !important; line-height: 1.3 !important; } }
Agent readiness measures whether a platform has the basic protections that an autonomous system needs: a safe place to test without affecting live data, error messages that a computer can actually interpret, the ability to retry a failed action without creating duplicate records, and the ability to send updates in real time instead of waiting to be prompted. This is the dimension that determines whether an AI agent can do useful work the first time – or whether it will immediately hit a wall.
For platforms that scored below 5 in agent readiness – Marketo (4), Gainsight (4), Heap (3) – hitting this wall is almost certain.
What this means for practitioners
The question for marketing practitioners is not whether agents will leverage these tools. It’s a question of whether the current tool will be the one the agent leverages, or whether a more advanced API competitor will take its place.
HubSpot’s headless Spring 2026 push and Breeze AI Agent APIs suggest it sees the writing on the wall. Intercom’s Fin API platform opens its AI agent to third-party development. Salesforce’s Agentforce 360 and Agent Script gave it a 9 out of 10 for agent readiness.
Vendors that are already investing in their agent interface are the ones that are pulling out. Others – including some of the biggest names in martech – make themselves expendable.
The position The martech categories hardest hit by AI agents appeared first on MarTech.
