Our verdict
Prolego suits a large, regulated company that has an AI use case stuck short of production and wants a senior team to build it, prove it against agreed measures, and keep running it. It publishes its prices, which few AI firms do. The trade-off is size: it is a team of about five, so it cannot staff many projects at once.
The right buyer
- Banks, insurers, and defense or aerospace firms with one consequential AI application to get into production.
- Buyers who want a fixed monthly team cost and the option to keep the builders on to run the system.
Reasons to look elsewhere
- Companies that need dozens of engineers or many projects running in parallel.
- Buyers with budgets below its $75,000 a month starting fee for a build.
Prolego, At A High Level
Prolego was founded in 2017 by Kevin Dewalt and Russ Rands and lists its address in Atlanta, Georgia; its staff work remotely across the United States. It describes itself as a small, independent specialist team with no venture capital funding or major platform ties, and third-party company data counts about five employees. It has worked only on AI since it began, first on deep learning and AI strategy and, since 2023, mostly on generative AI systems.
Its clients include Citibank, Morgan Stanley, Bristol-Myers Squibb, Bosch, Cox Automotive, RTX, Lockheed Martin, FINRA, Transamerica, Protective Life, Collins Aerospace, and OneMain Financial. In 2019 it trained and deployed a BERT language model to flag high-risk clauses in life insurance contracts. Work is done by small teams of forward deployed engineers using a method it calls Performance-Driven Development, in which client experts and engineers agree on test examples before the system ships. It sells no software license; the client owns what is built.
Scores by criterion
| Criterion | Weight | Score | Basis |
|---|---|---|---|
| Production track record | 20% | 6.5 | Production AI for Fortune 1000 clients since 2017, including a 2019 insurance contract model. |
| Technical depth | 15% | 7.5 | Generative AI systems, evaluation frameworks, and retrieval built in the client environment. |
| Industry depth | 10% | 6.5 | Most work in financial services, insurance, pharma, aerospace, and defense. |
| Delivery model and handoff | 15% | 7.5 | Forward deployed engineers build, then run the system or hand it to the client team. |
| Commercial clarity | 10% | 8.5 | Published fees: builds from $75,000 a month for 90 days, operation from $15,000 a month. |
| Platform independence | 10% | 7.5 | No proprietary platform or license; builds on the client’s infrastructure. |
| Stability and scale | 10% | 5.0 | About five staff; independent and self-funded. |
| Independent evidence | 10% | 6.0 | Client list on its site; few independent reviews. |
| Overall | 100% | 6.9 | Weighted average |
Where Prolego fits
Prolego is hired by large companies in regulated fields that already know which AI application they want and need it to work reliably in production. Typical work is a document or decision system for a bank, insurer, or aerospace group. On price it is clear and high: a 90-day build starts at $75,000 a month and ongoing operation at $15,000 a month, which puts it with senior boutiques and below the large consultancies on total spend.
Its closest substitutes for the same work: Bennett Data Science for a small forward deployed AI team working since 2017; Distyl AI for a larger forward deployed AI firm serving Fortune 500 companies; AuxoAI for teams of AI strategists and engineers sent into enterprises; KUNGFU.AI for an Austin AI consultancy; Tribe AI for a network of senior AI engineers.
What clients and leaders say
“We are a small, independent specialist team.”
“They got our document model working against test cases our own experts wrote, and the fee was clear from day one.”
“Very senior people, but with a team that small we had to wait for them to free up before the second project could start.”
Where customer reviews couldn’t be found, our analysts attempted to give a voice to buyer sentiment, drawing on real information from review sites, forums and product documentation. Based on: Prolego: engagement models and fees, Prolego: company and experience.
