IT giant Cognizant has teamed up with Workfabric AI to create a new job category — context engineers — who will convert corporate knowledge into structured inputs for AI agents. The move has developers buzzing about a whole new career path.
A LinkedIn post went viral this morning: “Forget prompt engineers — the hottest AI job is now context engineer.” Thousands of likes and comments later, the underlying story broke: Cognizant will deploy 1,000 context engineers over the next year, working with startup Workfabric AI to industrialise agentic AI applications. These engineers won’t write code in the traditional sense. Instead, they’ll capture organisational DNA — roles, rules, processes and data sources and feed it into AI systems that can act autonomously but responsibly. This echoes the rise of ChatGPT’s new AI agent, which highlighted how task-driven agents are becoming a cornerstone of the AI ecosystem.
What is context engineering?
In the language of Cognizant and Workfabric, context is an enterprise’s entire knowledge pool: operating models, goals, metrics, workflows, data governance policies and compliance rules. Large language models are powerful but general; without the right context they hallucinate, misinterpret instructions or violate corporate policies. Context engineers work as translators. They:
Catalogue an organisation’s roles and responsibilities.
Define rules, procedures and edge cases for each business process.
Identify data sources and map which fields can be shared with AI agents.
Build guardrails to ensure models operate within compliance frameworks (e.g., privacy laws, industry regulations).
Create continuous feedback loops to refine context as policies or data change.
By treating context as a “runtime layer,” Workfabric’s ContextFabric platform acts as the foundation on which AI agents run. Instead of injecting instructions at the prompt level, context engineers embed them at the system level, allowing agents to reason and adapt in real time across tasks.
Why is Cognizant investing now?
Cognizant CEO Ravi Kumar S. said in the announcement that “the lever of AI in the LLM era is context.” In other words, the value of generative AI no longer comes from the model itself but from how well it’s grounded in the client’s business. The company believes context engineering will address several pain points:
Risk reduction: By encoding policies and rules, companies can minimise hallucinations and ensure outputs adhere to legal standards.
Higher ROI: Trusted agents mean fewer mistakes, leading to higher adoption and better returns.
Efficiency: Proper context reduces errors and rework, freeing human workers for high‑value tasks.
Cost optimisation: Streamlined processes cut operational costs.
Accelerated time to value: With the right context, AI projects deliver results faster.
Differentiation: Companies that invest in context can build unique capabilities that competitors can’t easily replicate.
Workfabric AI’s secret sauce
Founded by Rohan N. Murty (son of Infosys co‑founder Narayana Murthy), Workfabric AI positions itself as the middleware for enterprise AI. Its ContextFabric platform compiles context instructions and feeds them into models at runtime. According to the company, pilot deployments improved accuracy by up to 3× and reduced hallucinations by 70 percent. Murty likens it to an operating system: “You wouldn’t run an app without Windows or Linux; why run an AI without context?”
ContextFabric’s architecture includes connectors to enterprise data sources, libraries of regulatory rules and an orchestration layer that assigns tasks to agents. Early customers include banks, insurers and healthcare providers — industries where compliance is critical. In one case study, a bank used ContextFabric to summarise loan applications. With context encoding credit policies and risk thresholds, the AI agent delivered accurate decisions with minimal human oversight.
Developer and business reactions
On GitHub, discussions about context engineering shot to the front page. Developers debated how to represent policies in machine‑readable formats and whether context engineering tools would remain proprietary or spawn open standards. Some praised Cognizant’s plan as a sign that enterprises understand the limitations of prompt engineering. Others criticised it as “consulting jargon” designed to sell expensive contracts.
Business executives, meanwhile, are intrigued by the prospect of a talent pipeline. Universities quickly pitched “context engineering” courses, and bootcamps are already advertising training programmes. Recruiters on LinkedIn reported a spike in searches for people with policy analysis and process mapping experience.
Challenges ahead
While context engineering promises to tame unpredictable models, it also raises new questions:
Who owns the context? Companies may be reluctant to share proprietary processes with a third‑party platform. Data security and intellectual property agreements need clarity.
Skill diversity: A good context engineer must understand law, business operations and AI — a rare combination.
Dynamic environments: Business rules change frequently. Tools must support continuous updating or risk stale context leading to outdated AI outputs.
Cost: Deploying 1,000 context engineers is expensive. Cognizant plans to train a workforce but has not disclosed salary ranges or billing rates.
FAQs
What is a context engineer?
A context engineer analyses a company’s business rules, data and processes, then translates that information into instructions that AI agents can interpret. This ensures models operate within legal and operational boundaries.
Why does Cognizant need 1,000 context engineers?
The company plans to scale AI deployment across multiple clients and industries. Having dedicated teams to build and maintain context ensures consistent quality and reduces risk, making AI more commercially viable.
What is Workfabric AI’s ContextFabric?
ContextFabric is a platform that stores context instructions and feeds them to AI models at runtime. It acts like an operating system for AI agents, improving accuracy and reducing hallucinations.
Will context engineering replace prompt engineering?
Not necessarily. Prompt engineering focuses on crafting input text to elicit desired outputs, while context engineering structures the environment in which AI operates. In complex enterprises, both are likely to coexist.
How can developers prepare for context engineering roles?
Study business process management, data governance, compliance frameworks and AI fundamentals. Experience in policy writing or operations analysis could be more valuable than traditional coding skills.
