“Most leaders aren’t asking whether their teams should use AI. They’re asking how to do it safely, efficiently, and without runaway costs. After almost a decade in tech, and building training and coaching programs as a team lead at Datadog, I’ve learned that the answer starts with giving everyone the same skills, the same guardrails, speaking the same language, using the same vocabulary, and having a clear sense of which tool fits which task. That’s how firms save time, keep spending under control, and can measure a real return on their investment in AI.”
Tommy GudmundssonCEO, Bifrost AI
Meet the founder of Bifrost
Bifrost is headquartered in Boston and run by Tommy Gudmundsson, an AI training consultant. Tommy has worked in the technology and AI space for almost a decade, including as a team lead at Datadog, where he built onboarding playbooks, training sessions, and rep coaching. A business education at Boston College and a Boston-based career give him a grounded understanding of the Massachusetts market, and Bifrost brings that enablement experience to the construction, architecture, and engineering firms that build it.
The approach is practical. Training is built around the documents your teams already work with, from RFPs and job-cost reports to meeting notes, and it covers governance and cost controls from the first session, so adoption is consistent, safe, and sustainable.
The name comes from Norse mythology: Bifröst, the burning rainbow bridge connecting the realm of gods to the realm of people. It's a fitting image for the gap most organizations are trying to close: between what AI can do in theory and what a team actually does with it on a Tuesday afternoon.
Training is available on Claude, ChatGPT, and other leading AI assistants.
What published studies say about AI and productivity.
Independent research on how AI affects knowledge workers. Each figure links to its source so you can read it yourself.
of work hours saved by U.S. workers who use generative AI, about 2.2 hours in a 40-hour week.
Based on a nationally representative survey in November 2024. Self-reported, and limited to people who use AI.
Federal Reserve Bank of St. Louis, Bick, Blandin and Deming, Feb. 2025for consultants using GPT-4 on tasks within AI's capabilities, who also completed 12.2% more tasks at over 40% higher quality.
A field experiment with 758 consultants at Boston Consulting Group.
Harvard Business School, Wharton, MIT and BCG researchers, Sept. 2023on professional writing tasks with ChatGPT (27 minutes down to 17), with higher-rated quality.
An experiment with 444 college-educated professionals, including marketers, consultants, and managers.
Noy and Zhang, MIT working paper, March 2023in productivity for novice and low-skilled workers, compared with 14% across all workers.
A study of customer support agents. It suggests newer team members can gain the most.
Brynjolfsson, Li and Raymond, NBER, 2023less likely to get a correct answer when consultants used AI on a task outside its capabilities.
The same BCG study. It is why we teach people to check AI output before relying on it.
Harvard Business School, Wharton, MIT and BCG researchers, Sept. 2023of architecture, engineering, and construction professionals say they use AI in their operations.
A Bluebeam survey of 1,000 AEC professionals, reported by ASCE. Most of the industry is still early.
ASCE Civil Engineering Source, Dec. 2025These studies come from other settings such as consulting, customer support, professional writing, and the general workforce, not from construction firms using Bifrost. Results vary by task, tool, and team, so we share them as context and not as a promise. Use the estimator above to test your own assumptions.
Frequently asked questions
Straight answers to what most firms ask before they get started.
Who is the course for?
It is built for business development, marketing, finance and accounting, and project management professionals at construction, architecture, and engineering firms, and for any office-based professional who wants foundational AI skills. No prior AI experience is needed.
What will it cost after the grant?
The Express Program can reimburse up to 100% of eligible course costs for qualifying Massachusetts employers, within per-person, per-course, and yearly caps. Awards are decided by Commonwealth Corporation, so the exact amount depends on your application. We will give you a clear price and walk through what a grant could cover on your intro call.
Who qualifies for Workforce Training Fund funding?
The Express Program serves eligible Massachusetts businesses with 100 or fewer Massachusetts employees. Employers with more than 100 employees can apply to the General Program, which is a competitive application where the employer names its training vendor. Training that is legally required, such as OSHA-mandated courses, is not eligible.
How does the grant process work?
Your company applies through Commonwealth Corporation at least 21 days before training begins, with a Certificate of Good Standing, a DUA Certificate of Compliance, and a signed W-9. If approved, you sign the service agreement before training starts. After the training is completed and paid for, you submit a reimbursement request with our invoice, proof of payment, and the list of trainees. Your company submits the application, and Bifrost does not charge for information about the program.
Do our employees need AI tool accounts?
Yes. Each participant needs a laptop and an account with the AI assistant your firm chooses. Your company sets up the accounts before the first session, and they are not part of the course price. We can advise on which plan fits your team.
How is our company data kept safe?
The course uses de-identified sample documents that we supply, so participants never need to bring bid-sensitive, client, or employee information to class. We also teach a clear list of information that should never be entered into any AI tool, along with acceptable use, confidentiality, and cost controls.
When and where do the sessions happen?
Sessions are live and instructor-led, scheduled during your team's normal work hours, and delivered in person or virtually. The course runs as three 4-hour sessions over three weeks, with up to 15 people per cohort.
We already use ChatGPT or another AI tool. Is this still useful?
Yes. The course builds skills that carry over to any AI tool: writing clear prompts, giving the tool the right context, checking its output, and handling company data responsibly. Teams that already experiment on their own often benefit most from a shared vocabulary and shared standards.
Which AI tools do you train on?
Our training covers the major AI assistants, including Claude, ChatGPT, and Gemini. We choose the examples and exercises around the tool your firm uses, and the core skills transfer between them: clear prompting, giving the tool the right context, checking its output, and handling company data responsibly.
Have a different question? Send us a message and we will get back to you.
Have more questions?
Send a message or book a short call and we will get back to you.