"I see AI as playing a pivotal role in helping to drive quality and consistency across our teams," said Chad Rowland, Director of Operations at Bunnell-Lammons Engineering, Inc. (BLE), who uses HallianAI to produce custom secure workflows within his firm.

Chad Rowland, Director of Operations at BLE

As an early adopter of HallianAI, BLE began creating and implementing its custom AI workflow within the firm in the fall of 2024. Larry Simonson, Chief Engineer and Group Leader of the Regional Geotechnical Services Team at BLE, knew the company's cofounder, Cameron Duncan, when they were both civil engineering students at Clemson University. Both Hallian and BLE are based in Greenville, SC.

Simonson had been in talks with his firm's technical directors about leveraging AI and was excited to work with a company he could trust. "We have somebody local who I know is an upstanding person in a space that I knew was going to be full of snake oil salesmen," said Simonson. 

Since then, BLE has established a Technical Reference Library to facilitate knowledge management with technical documents. The firm also developed a workflow to improve accuracy and mitigate risks when drafting QA/QC reports.

PSMJ spoke to BLE to learn more about how they are using secure agentic AI to generate technical documentation and facilitate mentorship support for junior engineers.

Technical Reference Library

As a construction materials testing, geotechnical, and environmental firm, BLE has acquired thousands of documents across three engineering groups.

BLE initially worked with HallianAI to create chat assistants for each group. First, they collaborated to index a database of about 10,000 documents across the firm. This enabled AI chat assistants to use retrieval-augmented generation (RAG) to answer questions and find information, citing references within the database.

The chatbot helps engineers navigate their firm's critical information more efficiently and provides context on where the agentic tool gathered the information. Engineers are still expected to verify the information.

"All these tools involve a ‘trust but verify’ aspect to them," says Simonson, who finds that he's saving time on repetitive work, but that he spends more time reviewing for accuracy.

For instance, Simonson has a workflow to jump-start technical writing tasks like site visit reports, but still must read and edit documentation to reach the finish line. "It will still only get me about 80% of the way there with the best output on it, but it saves a lot of time," he says.

Risk Assessment Workflow

Bunnell Lammons Engineering has developed an innovative AI workflow designed to enhance the quality of every report produced by the firm.

As the Director of Operations, Rowland's goal is twofold: to advance the skills of junior engineers and harness the expertise of senior engineers who are nearing retirement. He's currently grappling with the question, "How do we capture the knowledge of all of our years of experience to guide our up-and-coming generations on what we're doing?"

At BLE, each document is assigned to a writer, typically a junior engineer, and is then reviewed by a senior engineer. This workflow not only aids junior engineers in refining their writing skills but also ensures compliance with AASHTO, ASTM, ASCE, and internal standards, while expediting the senior review process.

The workflow operates as follows:

1.    The engineer drafts the report.
2.    The draft is processed through the workflow.
3.    The workflow identifies critical areas of risk.
4.    The engineer addresses the highlighted risks.

A senior engineer will conduct a final review of the report. If any critical risks remain, they provide constructive feedback to the junior engineer. Additionally, if the workflow fails to flag a risk, the senior engineer can provide critiques to enhance the workflow for future reports.

Currently, BLE is advancing workflows to automate checks that compare standard reports, such as property condition assessments,  Phase I environmental assessments, geotechnical reports, and site visits against ASTM and internal standards, effectively flagging any missing details.

Insights on Engineering Workflows

What can BLE’s experience with HallianAI tell us about the future of agentic AI in AEC firms?

First, we should always keep the end goals in mind. Everybody has their own approach to AI, and adoption at your firm won't be even across all members, but you can still achieve improvements in efficiency and accuracy. "We have some super users and some slow adopters that aren't seeing the full benefit yet." However, through continuous training, "more individuals are stepping up and diving into it. Our main goal is to improve that quality," says Rowland.

This leads us to our next insight: training is key to adoption. "If your firm is not using AI tools and you decide to use them, don't just thrust them upon the employees," says Simonson. "There's smart implementation, and there's less than favorable implementation."

Lastly, especially in the context of quality, AI has the potential to advance the workforce, but is likely a long way away from replacing it. "Our industry is always going to require individuals with significant experience to put their eyes on and review deliverables that are going to promote the safety of the community," says Rowland, who thinks AI can accelerate the growth of junior engineers. One day, "maybe someone new coming out of school with the proper AI tools is the equivalent of someone now with three or four years of experience," he says.

"The team at BLE has been successful with their AI implementation because of a top-down leadership vision coupled with a strong bottom-up approach for practical use cases," said Duncan. "It has been a joy to work with such forward-thinking engineers. I expect BLE to continue to lead in innovation and drive outsized results because of it."

This is content from the PSMJ Newsletter, exclusive to PSMJ PRO Members. PSMJ PRO is the fastest-growing network of AEC firm leaders. Not a PRO Member? Try a 50-day trial (no credit card required). You can request a trial here: https://bit.ly/50dayLI