Q&A: Why boutique consultancies might be better for AI rollouts than the bigwigs

2026-07-20 · Source: Computerworld · Field: Business & Management — Consulting & Professional Services, Corporate Strategy & Leadership, Project & Product Management · Depth: Intermediate, long

Summary

New York-based 28Stone Consulting, a 230-person technology consultancy for capital markets, is competing against larger firms in AI rollouts by emphasizing deep domain expertise and human involvement. Founders Thomas Dolan and Frank Erickson argue that agentic AI is not a one-size-fits-all solution, particularly in vertical markets, and requires discipline and human oversight to mitigate risks. They highlight that "AI-first" means "AI done intelligently," integrating human expertise throughout the software development lifecycle (SDLC), especially in requirements discovery and code ownership. The firm pushes back against the hype of "vibe coding" and the idea that AI democratizes enterprise software delivery, stressing that human engineers remain critical. AI redefines scale, allowing smaller firms to compete on output velocity rather than headcount, though evolving governance and token costs present new challenges.

Key takeaway

For Directors of AI/ML evaluating AI rollout strategies, recognize that deep domain expertise and human-in-the-loop processes are crucial for enterprise agentic AI success. Avoid the "do-it-yourself" trap and "vibe coding" mentality, which can lead to costly failures and erode trust. Instead, prioritize disciplined delivery, integrating expert human oversight from requirements to code ownership. This approach allows smaller, specialized firms to compete effectively and enables new projects previously deemed too expensive.

Key insights

Successful agentic AI rollouts require deep domain expertise and human oversight throughout the development lifecycle, especially in vertical markets.

Principles

• AI will become synonymous with software development.
• Human expertise must wrap the AI SDLC.
• Enterprise AI demands deep domain knowledge.

Method

Start with requirements discovery by expert business analysts, feed into BA agents, then developers own the output code, ensuring human accountability and domain-specific quality.

In practice

• Integrate expert BAs in AI requirements discovery.
• Assign developers ownership of agent-generated code.
• Prioritize domain interest in hiring AI engineers.

Topics

• Agentic AI
• AI Rollouts
• Consulting Strategy
• Capital Markets
• Human-in-the-Loop AI
• AI Governance

Best for: Director of AI/ML, Consultant, CTO

Related on AIssential

• They Fixed Agentic AI's Biggest Problem - 70% Failure Rate Solved | Ft. Ui Path & Omega Healthcare — Enterprise AI success requires clear use cases, robust orchestration, and human oversight, not just technology.
• Alibaba And Bytedance Quash Human Like Bots — AI development is balancing regulatory compliance, advanced model capabilities, and user adoption with persistent skepticism.
• AI Assistants and Preparedness Among Enterprises - with Chris Caldwell of Concentrix — Integrating machine-speed agentic AI with human-paced enterprise processes creates significant operational risk.
• Your Company Doesn't Need an AI Strategy — The future of firms in an AI economy hinges on building proprietary learning systems that compound human and AI capital.
• Anthropic Doubles Down On Agentic For The Enterprise — Enterprise agentic AI success depends on governance, integration, and trust, not just speed or task completion.
• The Big Con of Agentic AI — Over-reliance on agentic AI structurally erodes human agency and critical capabilities, mirroring the "con" of management consulting.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computerworld.