Salesforce unveils enterprise AI harness to govern agents across the business
New CapabilitiesSix-capability framework adds model routing, security, and policy controls for enterprise AI deployments
Yesterday: Salesforce unveils Trusted Enterprise AI HarnessNew here? Follow stories to track developments over time. Create a free account to get updates when stories you care about change.
Overview
Updated 57 minutes agoEnterprises are putting AI agents to work across their business systems, which creates a new problem: controlling what those agents can see, do, and spend. Salesforce introduced its answer on September 10, 2026 — the Trusted Enterprise AI Harness.
The harness is a software layer that wraps AI in six capabilities: context, agency, action, governance, security, and models. It gives agents a shared view of the customer and business, routes work to the right model, and applies consistent security and compliance rules. The goal is letting agents act across systems without companies building that control layer separately for every agent.
Much of the underlying tech already exists inside Salesforce's products, including MuleSoft's API management, Data Cloud, and Agentforce. The unified experience rolls out in early fiscal year 2028, which starts February 2027. Pricing arrives closer to general availability.
The move positions Salesforce as the central management layer for enterprise AI, not just another vendor of AI tools. Rohan Kumar, Salesforce's chief platform and engineering officer, said what will differentiate an enterprise is the trusted, proprietary context it brings to AI — and the ability to turn that into action.
Why it matters
Enterprises deploying AI agents face a governance decision: adopt Salesforce's harness or assemble point tools from model vendors and startups.
Questions about this story
Free account needed to ask — your question is kept and asked for you right after sign-up. Answers are public.
No questions yet — be the first to ask.
Key Indicators
Voices
Curated perspectives — historical figures and your fellow readers.
Play
Exploring all sides of a story is often best achieved with Play.
Higher or Lower
A number from this story, against one from elsewhere in the news — guess which is bigger, then keep the chain going. 5 rounds, 3 strikes; a miss costs a strike and resets your streak.
Keyboard: ↓/L lower · ↑/H higher
0 points — sign up to put that on the leaderboard.
Connections
Sixteen names from the news. Find the four hidden groups of four. Four mistakes max.
Sign up to keep a daily streak — a new puzzle lands every day.
Exit debate?
Your progress in this debate will be lost.
- 1 Two AI personas square off on this story.
- 2 You predict who'll win each round — correct picks earn XP.
- 3 One crossfire question is yours to fire. Pick it carefully.
Couldn't generate a topic
Select Your Champions
Choose one persona for each side of the debate
DEBATE TOPIC
Choose personas with different perspectives for a more dynamic debate.
Select debater for this side:
No debate personas available right now.
Select debater for this side:
No debate personas available right now.
Who's Got This Round?
Make your prediction before the referee scores
The referee scores both sides on
Round Results
Set the Crossfire
Pick the question both personas must answer in the final round
Debate Oracle! You called every round!
Sharp Instincts! You know your debaters!
The Coin Flip Strategist! Perfectly balanced!
The Contrarian! Bold predictions!
Inverse Genius! Try betting the opposite next time!
XP Breakdown
Prediction History
People Involved
Organizations Involved
Enterprise cloud software vendor best known for CRM, now pivoting hard toward AI agents.
Platform for connecting applications, data, and devices through APIs.
Salesforce's platform for building and running AI agents.
Timeline
March 2018 September 2026
-
Salesforce unveils Trusted Enterprise AI Harness
Latest Product launchSalesforce announced the six-capability AI harness and control plane; unified rollout starts early fiscal 2028.
-
Agentforce launches at Dreamforce 2024
Product launchSalesforce launched Agentforce, its agentic AI platform, positioning the company for autonomous agents.
-
Salesforce announces MuleSoft acquisition for $6.5 billion
AcquisitionThe deal brought MuleSoft's API management and integration tools, which anchor the harness's action layer.
Historical Context
3 moments from history that rhyme with this story — and how they unfolded.
Microsoft Active Directory (1999-2000s)
As Windows servers spread through corporate networks in the late 1990s, Microsoft introduced Active Directory to centralize identity and policy. It became the layer that governed who could access what.
Enterprises standardized on Active Directory for identity management.
The control layer anchored Microsoft's dominance of enterprise IT management.
Whoever owns the identity and governance layer for AI agents gains a controlling position in the enterprise AI stack — the dynamic Salesforce is chasing.
The API management wave (2008-2018)
When companies broke software into microservices and adopted software-as-a-service, APIs multiplied faster than teams could manage them. Vendors like MuleSoft, Apigee, and Kong sold governance layers to track, secure, and control access to those connections.
Enterprises bought or built API gateways to manage connections between applications.
MuleSoft became an enterprise standard and was acquired by Salesforce for $6.5 billion in 2018.
AI agents are multiplying across enterprises the way APIs did. The harness is Salesforce's attempt to own the governance layer for agents the way MuleSoft did for APIs.
Cloud cost management and FinOps (2015-2020)
As businesses moved workloads to AWS, Azure, and Google Cloud, tracking and controlling cloud spending became a real problem. A wave of startups — CloudHealth, Turbonomic, Apptio — sold cost management tools to fix it.
Cost management tools became standard practice for cloud-heavy enterprises.
Cloud providers built native tooling and bought or displaced the startups.
The AI Control Plane explicitly manages inference costs and agent performance, applying the FinOps model to AI agents.
