Data Leaders: Agents Will Redefine Knowledge Work
The founders of a data infrastructure company say the age of intelligence has arrived, and its first effects will be felt in knowledge work and by knowledge workers themselves. They compare the...
The founders of a data infrastructure company say the age of intelligence has arrived, and its first effects will be felt in knowledge work and by knowledge workers themselves. They compare the shift to earlier technology waves - PC, internet, mobile, social, cloud - but predict this one will be larger, changing both the role of the firm and the role humans play in production.
Intelligence, they argue, must be paired with tools and data to be maximally useful; that combination is what they call an agent. Agents can hold goals, ask and answer questions, and take actions. They already excel where context is available and outcomes are verifiable, with software engineering cited as a standout example.
The founders describe a contrast between tasks needing modest data, such as contract review, and sprawling questions like diagnosing a decline in billings for a healthcare provider, which demand massive context. Human analysts are constrained by time, they note, and many questions inside organizations go unasked. Agents, with effectively unbounded time and immediate availability, can address both problems if given access to the data they need.
Agents are bootstrapped on existing infrastructure - compute, networking, languages and APIs - which helps explain the speed of their rise. But the founders say unlocking AI's full promise requires infrastructure that is new or early, including evals and background memory formation, forming an agent stack that gives agents a shared, lasting picture of the business.
They contend agents learn best from the systems where business actually happens, rather than from documents, chats and email, which are filtered sources. Their near-term vision is agents supercharging knowledge workers, but they acknowledge a move toward autonomous agents that own business processes end-to-end, citing customer support as an example.
The people suited to build these systems, dubbed agent engineers, share traits with data teams: curious, outcome-focused, fast learners and technical. Their work spans process engineering, systems engineering, evaluation and operations, and change management. The founders stress that tying a business to a single vendor creates strategic vulnerability, making open source, open standards and portability essential, and they reaffirm a commitment to building around Open Data Infrastructure principles. They also point to a core value, Stay Human, arguing uniquely human traits grow more valuable as AI improves.
Source: dbt Labs Blog
Ready to Modernize Your Business?
Get your AI automation roadmap in minutes, not months.
Analyze Your Workflows →