AI deployment teams emerge as test of SaaS resilience

Enterprise AI providers are committing large sums and staff to customer-embedded engineering teams, a shift that Igor Kostyuchenok, SVP of engineering at Mbanq, presents as evidence that software-as-a-service is being reshaped rather than displaced. In a Finextra community opinion post, Kostyuchenok cited rapid growth in forward deployed engineer roles and multibillion-dollar programmes at major technology groups as signs that AI adoption depends on integration with existing enterprise systems.

Kostyuchenok wrote that job postings for forward deployed engineers rose by more than 1,000% year on year through early 2026. He said senior roles at frontier AI labs can reach total compensation of $520,000 to $780,000, and described Palantir as the largest-volume hirer for the operating model it helped popularise.

The central claim is that generative AI models do not become enterprise products until they are tied to workflows, data permissions, controls and compliance requirements. Forward deployed engineers are the personnel used to do that work: they sit close to customer operations and adapt technology to the systems already used inside banks, insurers, industrial groups and other large organisations.

AI labs and cloud providers expand deployment teams

According to Kostyuchenok, OpenAI increased its forward deployed engineering team from two people to 52 engineers in 2025 and later launched the OpenAI Deployment Company, described by him as a $4 billion venture backed by TPG, Bain Capital and Brookfield. He also said Anthropic’s Applied AI team is growing fivefold and has placed engineers inside FIS to work on an anti-money-laundering agent for BMO and Amalgamated Bank.

He pointed to similar moves by large cloud and enterprise software providers. AWS, he wrote, committed $1 billion to a new forward deployed engineering organisation with “thousands” of engineers. Microsoft launched what he described as the “Frontier Company” with $2.5 billion and 6,000 people, while Google Cloud opened 59 forward deployed engineering roles across eight countries in one week. Salesforce, ServiceNow and Accenture were also cited as expanding or creating programmes tied to embedded AI engineering.

Kostyuchenok argued that these investments would be less likely if AI systems could replace SaaS applications on their own. In his view, vendors are spending on deployment capacity because enterprise customers need AI features to operate inside the software they already use, rather than as separate demonstrations.

SaaS vendors use AI as expansion channel

The post also cited examples of SaaS companies using AI features to increase product use. Kostyuchenok said Atlassian added AI credits to its premium tier, with more than 1,000 customers upgrading and purchasing more than one million seats. He wrote that its Service Collection passed $1 billion in annual recurring revenue and had more than 65,000 customers.

Monday.com was described as using AI to extend into adjacent products including CRM, service desk and marketing tools, with new products accounting for more than 10% of revenue, according to Kostyuchenok. He also cited HappyFox, saying its AI autopilot reviewed closed support tickets, identified expansion opportunities and helped close $1 million in expansion revenue over three months at an agent cost below $20.

Oracle was cited as telling investors that AI could support large, integrated SaaS systems because customers want AI embedded in applications where their data already resides. Kostyuchenok’s broader conclusion is that SaaS platforms with embedded AI gain new pricing and expansion options, while products that do not add AI capabilities face greater competitive pressure.

This story draws on original reporting from Finextra Research.