A $10,000 invoice is an unlikely place to look for a theory of the modern consulting firm. Yet bld.ai puts one on its pricing page. The client pays. The platform keeps $1,000. The remaining $9,000 goes to service providers and solution vendors. A complicated arrangement of people, software, and obligations becomes a small piece of arithmetic.
- Custom AI products, built by a global network of specialists.
- A platform coordinates proposals, work records, contracts, and payments.
- The published client platform fee is 10% of paid invoices.
That little example contains the company’s larger ambition. Professional services involve two kinds of work: solving the client’s problem and organizing everyone who solves it. bld.ai sells the first while trying to turn more of the second into software. The interesting question is how far that arrangement can travel.
Start with the paperwork
The name sounds like an instruction, and the offering largely follows it. bld.ai supplies strategy, designers, engineers, data scientists, and product management to build enterprise AI and digital products. It also operates a marketplace connecting clients, service providers, and vendors with existing solutions. Buyers can commission something specific while drawing on more than one specialist.
Its website describes an AI agent that drafts proposals, matches providers, tracks work, and handles onboarding and contracts. These are familiar consulting chores, assembled into a common operating system. The proposed advantage is continuity: the scope that wins approval should remain connected to the people, deliverables, and payments that follow.
For the buyer, that matters because assembling a team is only the beginning. Someone must translate a business objective into a workable brief, decide who owns each piece, and keep the handoffs legible. A marketplace listing alone does little of that. bld.ai’s proposition includes the coordination.
An illustrative paid invoice from bld.ai’s pricing page
The freight truck is the point
To understand the work, consider a less glamorous phrase than artificial intelligence: AS400 integration. In its freight-dispatch project, bld.ai describes connecting an existing enterprise system to a platform used by dispatchers, drivers, and shippers. Dispatchers assign loads. Drivers accept jobs, update their location, and upload paperwork. Shippers see tracking information and arrival estimates.
The importance lies in those different perspectives converging. A driver’s update becomes useful to someone waiting for a shipment. Paperwork moves with the load. The older system remains part of the workflow. This is the sort of engineering detail that can determine whether a new application belongs in daily operations.
A separate tender-estimation project collects cost inputs through a web interface or CSV uploads, calculates totals, and preserves assumptions. Its stack includes React, Python, and AWS. The practical benefit is a shared calculation process: finance managers can see what has arrived, what remains outstanding, and what went into the estimate.
Then there is mining. The portfolio describes an economic digital twin that combines production and financial data, tests assumptions, and exposes results to analysts and engineers. The company’s client directory names BHP and describes exploration data work. BP, DataRobot, Cyient Australia, and Four Star Transportation also appear. The range explains why the offering stretches across data engineering, design, and integration.
A firm that became its own customer
Danny Castonguay founded bld.ai in 2017 after working at McKinsey. In a 2022 account, he described clients paying hourly for products ranging from frontline-safety digital twins to a remittance wallet and pediatric-research software. The connecting idea was human-centered development: useful software requires attention to the person expected to operate it.
The firm needed its own machinery, too. Castonguay called the internal platform Flow. His May 2023 retrospective reported $14 million in revenue and $3 million in profit for 2022, alongside a contractor-heavy network. Those dated figures describe a business already selling substantial work while experimenting with its operating structure.
His diagnosis was unusually plain. Sales depended on very few people. Internal platform users were captive. Growth made personal familiarity an inadequate quality-control system. These were emerging constraints, rather than evidence that every client project had failed. They explain why simply recruiting more people would not settle the company’s next questions.
The bottleneck moves upstairs
By a 2025 LinkedIn update, Castonguay was discussing AI-assisted sales, SaaS resale, and dedicated integration teams. He reported more than one million work records and about $50 million in historical revenue. That second number is cumulative. It should not be mistaken for annual sales, or for proof that the proposed sales model had already worked.
The direction makes sense as a response to the earlier diagnosis. Automating administration leaves the problem of finding and winning the next engagement. Supporting another software company’s customers could also create longer relationships through implementation and continuing development. DataRobot and Samsara were his examples of the resale direction.
“We bill you hourly and are always agile.”bld.ai company video, LinkedIn
The commercial model has several layers. Alongside hourly development, current platform pricing offers free client access and a $10 monthly Pro membership for offering services or solutions. The 10% invoice fee pays for the platform layer. It describes a payment mechanism; it is no substitute for agreeing on the actual project budget.
Copy the discipline, then size the team
There is something useful here for anyone buying software. Write down the deliverables, budget, timeline, payment terms, and IP arrangements before the team starts. bld.ai explicitly brings those decisions into its project setup and says clients retain ownership of commissioned products. That makes the agreement part of the product experience.
Its public product-management advice favors early AI prototypes, one accountable developer, and short daily feedback sessions. The transferable lesson is fast contact with the problem. For a narrowly scoped prototype, fewer handoffs can help. For sensitive data, regulated decisions, or complex infrastructure, buyers should size review, security, and testing to the consequences of error.
bld.ai sits between a development consultancy and a services marketplace, competing for work a buyer might give to an internal team or another engineering firm. Its distinctive wager concerns the connective tissue: can software make a distributed group easier to commission, coordinate, and pay? The invoice is tidy. The work underneath it is where the answer has to be earned.