Adept AI funding and software actions
Adept AI reached a $1 billion post-money valuation in March 2023. The Series B round included $350 million from General Catalyst and Spark Capital. Other investors included Microsoft, NVIDIA, Atlassian Ventures, and Workday Ventures. Total funding for the San Francisco company is $415 million. Adept AI founders include David Luan and Ashish Vaswani, plus Niki Parmar. Adept AI previously raised $65 million in a Series A in April 2022. Investors in that round included Addition, Greylock, Andrej Karpathy, and Chris Re. The company employs between 60 and 120 people. Adept AI develops the ACT-1 model to automate software workflows. This technology translates natural language into actions on websites and enterprise tools. It automates tasks such as browsing and data entry or form filling. Adept AI targets software like spreadsheets, browsers, and CRMs. The system allows AI to perform actions without manual clicks.
Salesforce deepens enterprise AI partnerships
Salesforce and OpenAI expanded their partnership at Dreamforce 2025. This collaboration brings AI capabilities to ChatGPT and Slack, along with the Salesforce platform. Users can use GPT-5 to build agents directly in the Salesforce Platform. Salesforce positions Slack as the "Agentic Operating System" for the modern enterprise. ChatGPT and Slack are available for use, plus the Salesforce platform. OpenAI’s Codex agent also works in Slack to help developers with coding tasks. Salesforce also partnered with Stripe and OpenAI to create the Agentic Commerce Protocol. The partnership between Salesforce, Stripe, and OpenAI creates a complete, end-to-end solution that empowers merchants to drive revenue growth and build deeper customer loyalty directly within platforms where shoppers already reside. This protocol ensures payment security and merchant control. Salesforce also integrates Gemini models into the Atlas Reasoning Engine. For regulated industries, Salesforce uses Anthropic’s Claude. Claude operates within the Salesforce trust boundary. Research from Salesforce reveals that 48% of shoppers who already use AI are open to having an AI agent make purchases on their behalf. Reddit used Agentforce to handle advertiser support. This deployment achieved 46% case deflection. It reduced resolution times from 8.9 minutes to 1.4 minutes. Reddit reported a 20% increase in advertiser satisfaction. Salesforce uses the Einstein Trust Layer to provide security. This layer includes data masking and toxicity detection. You should ignore the hype and focus on reducing operating costs. Smaller agencies waste 70% of Salesforce capability and overpay on licensing.
Agent maturity and market valuations
The AI software market shows a valuation gap between company categories. AI-native companies command a median 21.2x revenue multiple in venture capital rounds. This is four times the 5.5x median for legacy SaaS. AI-enabled companies earn an 8.5x median multiple. In M&A transactions, AI-native companies see a median 11.5x revenue multiple. Legacy SaaS companies see a 3.8x median multiple.
| Company Category | VC Revenue Multiple | M&A Revenue Multiple |
|---|---|---|
| AI-Native | 21.2x | 11.5x |
| AI-Enabled | 8.5x | 7.0x |
| Legacy SaaS | 5.5x | 3.8x |
Cursor reached $4 billion in annualized revenue in June 2026. SpaceX completed its $60 billion all-stock acquisition of Anysphere, the parent company of Cursor, in August 2026. Cursor uses its own Composer models. Cognition reported a $1 billion run rate in September 2026. Cognition also reported that Devin usage grew 50% month over month for six straight months. Replit targets a $1 billion run rate before the end of 2026. Factory raised $200 million at a $5 billion valuation in September 2026. Factory sells enterprise agents called Droids. Salesforce classifies agent maturity into five levels. Level 0 focuses on fixed rules and repetitive tasks. Level 1 agents fetch and analyze data. Level 2 agents execute workflows in one domain. Level 3 agents coordinate workflows across different systems. Level 4 agents use multiple agents supervised by a central AI. Will the distinction between AI-native and AI-enabled companies persist as models improve?
