According to the MIT Technology Review Insights report, Redefining the Future of Software Engineering, enterprises intend to scale agentic AI to the point where AI agents manage product development and software development lifecycles end to end. More than 70% of respondents expect to be there by 2028.

Those are big ambitions. Reaching them requires service and technology partners working alongside enterprise teams to turn the ambition into software that actually delivers.

After long and fruitful collaborations, SoftServe and Anthropic have now formalized their partnership. SoftServe is a recognized partner within Anthropic’s ecosystem, a status that did not exist between the two companies before now. The agreement centers on building generative AI solutions on Claude, with a specific focus on agentic software engineering powered by Claude Code.

What the agreement does not do is start the work. It formalizes an engineering practice that has already delivered for clients.

Built before the announcement

For over 30 years, SoftServe has been at the front end of every major technology wave since the dawn of the World Wide Web. AI is no different. We have been deploying AI in production since 2015, and in 2023 we created a dedicated lab for generative and agentic AI. Inside that AI Lab, our agentic engineering practice has been running client work with Claude Code alongside our Agentic Engineering Suite for greenfield software development. We have also run a Generative AI Readiness Accelerator using Claude on Amazon Bedrock for clients operating within existing AWS environments.

Doing that work before the partnership existed gave us something more durable than early access. It gave us a tested view of where agentic engineering earns its keep, what it costs to run properly, and which parts of the lifecycle still belong to an experienced engineer.

Our practice centers on a proprietary agentic engineering methodology covering assessment, prioritization, proof of value, adoption support, governance, measurement, and scale-out, backed by reusable agentic SDLC accelerators, agents, skills, MCP servers, templates, and quality-control components. At the center of the approach is a spec-driven development philosophy connecting business intent through requirements, architecture, code, testing, and deployment, with human experts keeping explicit review and approval authority at every stage.

We call the people who configure, supervise, evaluate, and improve these agentic workflows Intelligence Engineers. A developer working this way still writes code, but the job now also includes directing and checking the work of AI agents doing the same. Our Intelligence Advisory practice complements the engineering work with assessment, strategy, governance redesign, use-case prioritization, and adoption support, along with role-based training for developers, architects, quality engineers, product leaders, and executives.

We connect Claude to the tools our clients already run, including Jira, Confluence, Figma, GitHub, CI/CD platforms, cloud environments, and internal knowledge sources, so agentic workflows fit into existing delivery processes rather than sitting apart from them. Our own intellectual property lies in how we configure and integrate Claude Code for each client’s environment, through reusable skills, MCP servers, engineering standards, workflow orchestration, quality gates, and adoption playbooks.