Salesforce Spends $3.6B on AI Customer Service Agent Platform as NewCore Raises $66M for Enterprise Agent Identity

Salesforce is acquiring Fin, the AI customer service platform formerly known as Intercom, for $3.6 billion — the largest pure AI agent acquisition of 2026. The CRM giant plans to fold Fin's technology and team into Agentforce, its enterprise platform for building custom AI agents. Fin's core product uses AI agents to handle customer inquiries autonomously across chat, email, and voice channels. Salesforce sees agentic customer service as the primary on-ramp for enterprises deploying AI agents at scale. The deal also validates Fin's business model, which charges per conversation resolution rather than per seat — a pricing structure that reflects how agent-based services are reshaping enterprise SaaS.

Two other agent infrastructure companies raised significant rounds on the same day. NewCore announced $66 million in Series B funding to build identity and access management systems specifically for AI agents. The company argues that enterprises deploying dozens or hundreds of agents will need to manage them the same way they manage human employees — with credentials, permissions, audit trails, and approval workflows. Malaysia-based Respond.io raised $62.5 million for its AI agent-powered customer messaging platform, which also charges per conversation rather than per seat. Combined, the three deals pushed over $3.7 billion into agent infrastructure in a single day.

AWS published two significant pieces of agent developer infrastructure. Strands Evals is a new open-source tool that detects and diagnoses AI agent failures in production, categorizing them with confidence scores and surfacing root causes. The second piece walks through building research agents with Deep Agents and Bedrock AgentCore, a pattern that creates isolated execution environments for multi-step agent workflows. Both target the operational side of agent deployment — not just building agents, but running them reliably at scale.

The Anthropic Fable 5 and Mythos 5 shutdown story continues to produce new angles. The Verge reports that Anthropic spent the weekend fighting the White House over the order to block foreign access to its newest models. Cybersecurity experts are protesting the restrictions as "dangerous," arguing they limit defensive AI capabilities. The Decoder reports that European leaders are using the incident to push for sovereign AI infrastructure. The story is now in its fourth day of generating new developments.

Source-linked headlines

Salesforce acquires AI customer service platform Fin for $3.6B

TechCrunch · June 15

Salesforce buys Fin (formerly Intercom) to supercharge Agentforce, its enterprise platform for building and deploying custom AI agents that automate business tasks.

Why it matters: This is the biggest signal yet that enterprises are treating AI agents as a core infrastructure category, not an experiment. The $3.6 billion price tag sets a benchmark for agent-native companies.

As AI agents become employees, NewCore emerges with $66M to give them identities

TechCrunch · June 15

NewCore raises $66 million for an identity and access management platform built specifically for AI agents — treating them as credentialed entities with permissions, audit trails, and lifecycle management.

Why it matters: Every company running more than a handful of agents will need this. Identity management for agents is a category that barely existed a year ago and is now attracting serious venture capital.

Malaysia's AI agent-powered messaging app Respond.io raises $62.5M

TechCrunch · June 15

Respond.io's AI agent platform handles customer inquiries at scale and charges per conversation, not per seat — a pricing model that swaps seats for resolutions.

Why it matters: Per-conversation pricing is becoming the default for agent-based services. If this model scales, it changes how enterprise software is bought and sold.

AWS launches Strands Evals for AI agent failure detection

AWS Machine Learning Blog · June 16

AWS releases an open-source tool for detecting, categorizing, and diagnosing AI agent failures in production, with confidence scoring and root cause analysis.

Why it matters: Production agent reliability is the next bottleneck after building agents. AWS is betting that open-source diagnostic tooling becomes essential infrastructure for agent operations.

Inside the fight over Claude Mythos 5

The Verge · June 15

Anthropic spent the weekend battling the White House over an export-control order that forced it to block foreign access to Fable 5 and Mythos 5.

Why it matters: The Fable/Mythos story keeps generating new layers — government pressure, corporate competition, cybersecurity pushback, and now a global sovereignty debate. It's reshaping how AI companies think about model releases and geopolitical risk.