The Narrow Path of AI Enabled Estimation

There is no doubt that AI will make companies more efficient, but embracing inefficiency may be a competitive advantage in the long term. AI is not magic, neither should it be underestimated. With increasingly automated bid pipelines, business owners and estimators need to be wise and deft in the use of AI tools. Many risks accompany adoption. If used recklessly, companies may face high costs and have less insight. If not used at all, companies will miss out on the benefits, making them less competitive. The companies who benefit will be the ones who leverage AI to empower workers, add efficiency while prioritizing people and infuse the process with accountability at every step.

Organizations who race to adopt AI systems, replacing workers and automating workflows, may find themselves abdicating responsibility and dependent on wobbly technology. AI systems are non-deterministic. They hallucinate, make mistakes and mess up in unexpected ways. The axiom “trust but verify” is essential for responsible adoption. Verification is a discipline that must not be allowed to atrophy. It is an inefficiency that should be embraced and practiced. Bid volume should never exceed what estimators are capable of checking manually. AI use upstream may reduce conflicts and errors typical of bid sets or it may make errors less interpretable, leaving downstream estimators with the responsibility of interpreting AI generated slop. If too much responsibility is handed to AI systems, these errors could compound. Responsible adopters remember where accountability lies, not with the tools, but with the ones using them.

As more tools arrive, pressure will build to update tech stacks. Many new software products will be built with AI or incorporate AI features. Only a fraction will add value. Adopters need to evaluate publishers as well as software in this dynamic environment. Will this publisher be around in 6 months to support the tool? Do they understand how it works? Are there real people at the Publisher that I can talk to? Does the tool produce verifiable and consistent output at each step of the workflow? Does it add accountability or obfuscate it? In a crowded field of software, the inefficiency of building trust will illuminate the path forward.

In the short term, the largest benefits of AI adoption will come from smoothing out verifiable workflows. AI tools can be leveraged to write scripts to handle data import/export, excel macros and handle translations reliably. The leading edge of LLM capability is in text and coding and many excellent tools will emerge from that toolbox. Those who navigate the landscape successfully will be those who retain their agency and own their responsibility in an increasingly agentic world. On the other hand, the most effective way to demonstrate a lack of care to a customer, employee or colleague is to use AI as a substitute for effort. Person-to-person communication face to face, through emails or phone calls is another inefficiency that needs to be embedded. AI enabled sales and customer support is the death knell of customer relationships. In the long run, the people and businesses who succeed will be the ones who generally do, those who care about their customers, employees and coworkers and who approach their work with curiosity, purpose and integrity.