How much will my AI Solution Cost?
Hi, you're probably here to try to figure out on your own how much an artificial intelligence-based project will cost your company, without having to talk to an annoyingly pushy salesman.
We get it!
Overall Project Cost
Small projects typically start around $10,000, however, our sweet spot starts around $50,000 for initial development, and an average monthly retainer of 20 + hours for ongoing training, iteration, and maintenance. Average project duration is typically 4 - 12 weeks.
What does a two-week sprint cost?
What does a two-week sprint cost?
Sprint is a misnomer as AI +Strategy does not normally follow a Scrum methodology, we are not fans for multiple reasons. Two-week “sprints” are what most of our clients are familiar with, so we package development this way, so that our clients have an apple to apple comparison when comparing AI +Strategy to other vendors.
Depending on the skill level, a small full-time developer team (3) costs approximately $15,000 - $25,000 for a two-week sprint on a nominal project.
Part-time project lead
Platform specialist or Data Scientist
Part-time teams are possible, but as you guessed, they tend to take twice as long to get your project accomplished.
A complete project team (8-11) costs approximately $70,000 - $100,000 per two-week sprint.
Data Scientist or Platform Specialist
Data curator (pt.)
DevOps (as required)
API developer (as required)
Solutions architect (as required)
Typically require five development teams separated by function with a dedicated project manager, project coordinator and leads for each team. Complex project teams typically have 25 – 40 developers in total and have a monthly cost of approximately $500,000 - $800,000 per month.
Curation and data wrangling
AI training (may be more than one team)
Backend development and API connectivity
DevOps and security
Complex projects frequently explore multiple prototype areas for maximum competitive advantage with a small sub-team whose primary tasks are to break things and do the impossible. The results of their work run parallel to the primary development effort and are typically at the forefront of AI solutions.
More expensive than what you were thinking?
The typical AI project contains all the elements of a standard application development project then adds several layers on top of that (validation, data wrangling, data sourcing, development, AI training, algorithm development, and extensive testing). People with both skills and experience are in extremely high demand and even a simple AI project can utilize a stack of 20 applications without trying hard. It is an expensive undertaking and requires a skilled team, or else you will be throwing your money out the window.
Regardless of your project scope we know that AI solutions are a big investment.
How to get project pricing?
Iterative projects have no clearly defined finish date, flexible objectives, and functions, with goals, may change at any point in time. These projects are typically billed on 2-week sprint cycles and usually involve three separate teams:
Back End Development
Front End Development
Cross-functional teams are not generally recommended for most B2B AI projects unless it is for a simple application such as a bot, virtual agent or a small-scale solution such as video analytics or decision support in a clearly defined area.
Iterative projects can burn a large amount of money chasing down ideas or in the worse-case, the changing whims of the client’s project champion. Iteration gives the client the most flexibility in meeting end-user demands, but it comes at a price, as the finish line keeps moving.
Caveat, training complex AI systems come with a time investment of 6 – 12 weeks per Corpus or DataSet. It is difficult, if not impossible, to iterate AI training once it has begun. This stage is not flexible.
Traditional projects have clearly defined goals, with a start and finish date, with an ongoing support and development element. Pando Innovations works in a modified Agile Kanban environment, so it is possible to rearrange most priorities with little notice. The above caveat about AI training time still holds true.
How long to get a price?
Preparing a proposal generally takes one day to two weeks depending on the complexity of your project, and our current workload. Quick and efficient communication with your teams lead and key players is critical to delivering a proposal promptly. If it takes days to get answers, it is going to impact how long it takes to prepare a proposal.
If we have not done business with your company before, or do not know you personally, we generally decline unsolicited RFP’s.
The reason is simple, you need three quotes to comply with internal regulations at your company, but already have a project supplier chosen. Our weeks of effort in developing your RFP is a complete waste of time so that you may tick off a box. Worse, any IP or unique solutions that we are proposing created for your project is now going to be shared with our competitor.
Factors that impact Pricing
What problem or pain points are we solving?
Is this a standard and well documented process, or are we about to embark on something completely novel that has never before been done?
What type of environment is this?
Is this a prototype where breaking things is highly encouraged while pushing the boundaries of technology, or are we integrating into a mission-critical environment, where a failure of any sort cannot be tolerated.
What is at stake?
What kind of questions will the AI be answering? Low-value where an inaccurate response has limited repercussions or the other end of the spectrum where the answer could be the difference in a go/no-go decision on a multi-million dollar investment?
Who is your team?
How large and experienced is your internal IT team in managing complex projects and product life cycle maintenance? Do they have the bandwidth and skill to learn to support AI systems in addition to their existing workload? Have you recently managed a complex development effort?
What data is available, and how clean is it?
Is this an international application? What are the repercussions in language, data storage, privacy, security, payment processing, governance and individual country laws?
Iterative development – two-week “sprints” are payable in advance with a minimum of one month’s prepayment at any given time.
Defined projects are Milestone-based typically with 50% payment in advance for the first milestone.
Financed projects, payments are released on a schedule (typical) or up front for small efforts.