Ask every firm you are considering the same questions: who will do the work, what they have running in production and how they measured the results, what the software will cost to run, and how your team will run it after they leave. Compare the answers side by side. Then start with a short project with a fixed scope before you commit to a large build.
Questions to ask every firm
Ask each firm these questions, and ask for evidence behind each answer:
- Who will do the work, and how much of their time will they give it?
- What have you built that is in production today, and how did you measure the result?
- What will this cost to operate after launch, including model usage, review, and maintenance?
- When would you recommend a rule, a report, or no project at all instead of AI?
- How will you handle our data, and who will be able to see it?
- What will our team need to know to run and change what you build?
Here are our answers.
Measured results
Ask what the work looked like before, what it looks like now, how the firm measured the change, and who checked the numbers.
For an insurance carrier, the StrataEdge Reconciliation Platform reduced quarterly auditing and reconciliation from five people working for a week to one person working for a day. Manual checks found that 99 percent of outputs met audit requirements, compared with 90 percent under the prior process. For Boston Consulting Group (BCG), StrataEdge and two engineers built a working AI agent in four weeks, with structured outputs and human review built in, and it was in production three months after we started. Read the BCG case.
As GEICO’s CTO, Linda managed a $1.2B technology budget and saved $150M a year. Our team built LoanFlow, mortgage software that cut the time to clear a loan for closing from 30 days to under 24 hours and cut back-office costs by 75 percent.
Who does the work
Ask for the names of the people who will do the work and their time commitments, and put both in the statement of work. At StrataEdge, Linda and Allan lead every engagement themselves. Over their careers they have worked with thousands of engineers, and we draw on that network to staff each project. Meet the founders.
Operating cost and your team
Building the software is only part of the cost. Ask for an estimate of model usage, infrastructure, licenses, exception review, monitoring, and maintenance for each year you expect to run the software. Ask the firm to report staff time released separately from budget saved.
Ask what your team will need to run it. For DigiKey, StrataEdge designed the organization for technology teams in the United States and India and delivered leadership and generative AI training. DigiKey implemented the roles, jobs, and development plans.
Starting with a short project
A short project shows you how a firm works before you commit to more. Our AI investment review usually takes two to four weeks and ends with a ranked list of what to build, what to buy, what to fix without AI, and what to stop. A proof of concept usually takes four to eight weeks, and we measure it against a result we agree on before work starts. See our typical engagements.