How a custom AI build runs at ColabContent, written out in full.

An AmLaw 100 firm, a Big Four accounting firm, a top-100 national agency, a Fortune 500 manufacturer: each of these has an in-house innovation team, a CIO with a discretionary budget, and the seat count to make a horizontal SaaS (software rented as a monthly subscription) platform make sense. For example, a law firm doing roughly $20M in revenue, a specialty manufacturer around $40M, a regional insurance agency near $25M, or a PE-backed (owned by a private equity firm) home services platform around $30M: each of these illustrative profiles has the budget to commission a custom system but not the in-house engineering bench to build one, and not the user count to make a rented software subscription pencil. Mid-market AI is a separate buying motion with separate failure modes and separate winners.

The failure modes look like this. Off-the-shelf AI is calibrated against the average customer in the category, which by definition is the larger firm. The mid-market operator pays per user for features the firm does not use and loses a meaningful share of the platform's value to misfit on the workflows that matter. The internal AI hire generally has to stand up infrastructure before the first production workflow can ship, and that work comes out of the first year. The Big Four consulting engagement is typically priced well above this band and typically ends at a strategy document, leaving the build to a third party that the firm then has to manage. None of those motions are wrong; they are wrong for this revenue band.

A fixed-fee build addresses the gap directly. A boutique principal scopes the constraint, builds a working prototype on the firm's real data inside seven to ten days, and ships the production system in four to seven weeks. The code, prompts, and models transfer to the client at handoff. The system runs inside the firm's own cloud tenant (a private cloud account) under NDA (a signed non-disclosure agreement). There is no per-seat pricing, no proprietary runtime, no vendor lock-in. The full fee structure for each stage is published on our pricing page.

Every engagement starts with the $499 AI-Ready Audit. It names the costly workflow, ranks the fixes, and shows whether a custom build is justified at all. On the call that follows, the principal writes the constraint down in a sentence. Both sides leave with that sentence. Either party can stop here.

If both sides decide to proceed, we sign a one-page NDA (a confidentiality agreement) and the firm provides a representative slice of real data: ten anonymized matters, a hundred recent RFQs (requests for quote), a quarter of COI (certificate of insurance) requests (certificates of insurance), a season of PBC (the prepared-by-client document list) packages (the documents a client prepares for its accountant), a month of dispatch logs. Inside seven to ten days we ship back a working prototype that performs the constraint task on that real data. The firm sees the system actually work before any payment changes hands. If the prototype does not perform to the target written down after the audit, the firm owes us nothing and keeps the work product.

If the prototype performs, the fixed-fee production build begins. The fee is quoted after the audit and the prototype, based on the problem being solved, the integrations required, and how much of the handoff the firm wants to manage itself, and it is agreed in writing before the build starts. Build runs four to seven weeks. The system ships inside the firm's own Azure, AWS, or Google tenant (a private cloud account) under NDA. The firm receives the code, the prompts, the models, the datasets, and the runbook (the written operating instructions). The firm owns the system. There is no proprietary runtime to license and no per-seat fee to renew.

Ongoing care after handoff is optional: $997 a month, cancel on 30 days notice. We do not disguise a retainer as a subscription. The full fee structure for every stage, from the audit through ongoing care, is published on our pricing page.

We build in five verticals because the buying band is right, the workflows are bespoke enough to defeat horizontal SaaS, and the dollar volume of the leakage justifies a fixed-fee engagement. Each vertical has a different constraint, and we are further along in some than others. Where we have shipped, we say what we shipped and what it has handled. Where we have not shipped yet, we say that instead of borrowing a number.

Mid-market law firms. Twenty to one hundred fifty attorneys. The leakage is unbilled partner time, mis-routed intake, and timesheet reconstruction loss. This is where our shipped work runs deepest. For Jim Glaser Law we built five channel-specific voice agents (PPC, organic, TV, Meta, and LSA) that have handled 3,787 calls and 5,514 minutes (measured in the client's call platform, August 2026), giving the firm per-channel attribution on every answered call. A law firm runs its matter, invoice, and IOLTA (the client trust account a law firm must keep separate) trust accounting on a platform we commissioned: 13,296 matters, 4,396 clients, 5,684 invoices, and a trust ledger (an IOLTA account, the lawyer's trust account that holds client funds) that reconciled byte-identical against the system it replaced. Build cycle five to seven weeks. Compared in depth on our best AI consultants for mid-market law firms guide.

Specialty manufacturers. Our target profile is roughly $15M to $150M in revenue, our own estimate of the fit, not a published threshold. The leakage is quote turnaround time, an RFQ (request for quote) that waits while an estimator hunts for pricing history. We have not shipped a build inside a specialty manufacturer yet, and we would rather tell you that than show you someone else's number. The closest system we run is for a regional third-party logistics operator, where the AI intake layer has handled 211 calls (measured in the client's call platform, August 2026). In manufacturing we scope against the shop's own RFQ log: measured quote turnaround before the build, re-measured after, on the same sample. Build cycle seven weeks. Compared in depth on our best AI consultants for specialty manufacturers guide.

Regional P&C insurance agencies. Our target profile is roughly $10M to $50M in commission revenue, our own estimate of the fit, not a published threshold. The leakage is COI turnaround and submission processing. We have not shipped a build inside an insurance agency yet, so we have no agency numbers of our own to show you. What we would put in the scope is the agency's own clock: time from COI request received to certificate issued, and the share of submissions that still need a human touch, baselined before the build and re-measured after. Compared in depth on our best AI consultants for insurance agencies guide.

Mid-market CPA firms. Thirty to one hundred fifty professionals. The leakage is partner time on PBC reconciliation, season-driven workflow chaos, and CCH Axcess or UltraTax data plumbing. We have not shipped a build inside a CPA firm yet. The measurement we would agree to before the build starts is partner and manager hours spent chasing PBC items across one season, counted from the firm's own time entries, before and after. Compared in depth on our best AI consultants for CPA firms guide.

PE-backed (owned by a private equity firm) home services platforms. Our target profile is roughly $20M to $100M in revenue, our own estimate of the fit, not a published threshold. The leakage is call abandonment and dispatch friction across multiple brands and field service systems. The AI intake layer we run for a multi-location home services operator has handled 1,486 calls and 2,203 minutes of live conversation (measured in the client's call platform, August 2026). What we scope against is the platform's own call log: abandoned calls and after-hours misses counted before the build, counted again after, on the same brands. Build cycle four to six weeks. Compared in depth on our best AI consultants for PE-backed home services guide.

Is the system replacing staff? No. The systems we commission are scoped against a workflow constraint, not a headcount target. The work they absorb is the queue nobody wanted: after-hours calls, intake routing, reconciliation, document plumbing. What the firm does with the reclaimed senior capacity is the firm's decision, and we do not scope a build around removing seats.

What if our data is messy? Messy is the baseline assumption. The prototype is built on real, unsanitized data from the firm. Cleaning is part of the build, not a prerequisite for it.

What if we already have an internal AI hire? A commission works alongside internal AI capability rather than in place of it. The internal hire owns adoption, governance, and the next twelve months of evolution. The commission ships the first system, on schedule.

What about model risk and confidentiality? The build runs inside the firm's own cloud tenant under NDA. Client data does not leave that environment.

What if the AI category changes again in six months? The build is owned by the firm. When a better model or technique appears, the firm replaces the relevant component without renegotiating a vendor agreement. That is the structural advantage of a commission over a SaaS subscription.

More questions like these are collected in the questions section below.

The diagnostic questions covered earlier on this page, on whether AI is the right intervention, the right buying motion, and the budget, are asked of the senior operator and answered in writing on the 20-minute call that follows the audit. The call is the work of deciding which answer carries the leverage and writing down, in one sentence, the exact constraint that the build will address. That sentence becomes the scope, and the scope becomes the fixed fee.

The 20-minute call that follows the audit costs nothing beyond the $499. The report is the firm's to keep regardless of whether a build proceeds. We will tell the firm honestly if the right next step is a different consulting house, an internal hire, an off-the-shelf product, or no action at all.

Booking that call is covered on our contact page.

AI-Ready Audit

A one-time review that ranks, in dollars, where a business loses money to missed calls, slow follow-up, unused software and weak visibility in Google and AI assistants. The deliverable is a report, not a sales call.

Commission

A fixed-scope, fixed-fee engagement to build one custom AI system. Not a retainer, not a subscription, not a strategy document. The deliverable is working software.

Custom AI system

Software built to one operator's data, stack, and workflow. Not a configured off-the-shelf tool. The system connects to the operator's existing platforms (practice management, CRM (customer relationship management software), ERP (enterprise resource planning software that runs back-office operations), accounting) and runs inside their own cloud tenant.

Prototype-before-pay

The operator sees the system work on their own data before any production fee is committed.

Mid-market operator

An established business too large for off-the-shelf tools to fit without customization and too lean to carry a full-time AI engineering team. That is where a fixed-fee custom system pays off.

See how these terms apply across each vertical on our industries page.

A prospect who has already run the sequence below gets more out of the audit and the call, because the conversation starts at the constraint instead of at the introductions. The four questions take five minutes, and a "no" on any of them is useful information, not a failure.

Once you have run the sequence, the pricing page lays out what each stage costs.

The sequence asks four questions in a specific order. First, is the leading constraint actually addressable with AI, or is it a process problem, a staffing problem, or a stack problem that AI would not solve. Second, if AI is the right intervention, is the right buying motion an owned build, an off-the-shelf product, or an internal hire. Third, if the right motion is a build, is the firm comfortable running the system inside their own cloud tenant under NDA and owning the code at handoff. Fourth, is the budget real this quarter: $499 for the audit, from $2,997 for the fixes, and a quoted fixed fee for a custom build.

If the answer to all four is yes, the audit is the right next step. If the answer to any one of them is no, the honest options are to change the question (the leading constraint is different, the budget moves, the cloud posture changes) or to take a different path entirely. We do not push a prospect who lands at a "no" on any of the four into an engagement they will not be served by.

The audit referenced in the fourth question is explained in full on the AI-Ready Audit page.

Twelve months post-handoff, three signals tell the firm whether the build performed against the target written down after the audit. First, the dollar or hour delta on the workflow the commission addressed, measured against the pre-engagement baseline. Second, the percentage of the workflow the AI layer now handles autonomously versus the percentage that still routes to a human reviewer. Third, the number of times the firm's team has modified the system's prompts, models, or integration code on their own without ColabContent involvement. All three should be improving over time. If they are not, the optional post-handoff care, $997 a month, cancel on 30 days notice, is the lever for diagnosing what changed. That care plan is detailed on our pricing page.

An owned build is not the right answer for every business. A firm whose workflow matches a horizontal SaaS product's calibration target is better served by the product. A firm with a five-to-ten-year horizon, a large AI investment runway, and the willingness to spend twelve months building infrastructure before shipping the first production workflow is better served by an internal hire. A firm large enough that a Big Four engagement makes sense is better served by that motion. We will tell the prospect which of those alternatives fits if a build does not.

The honest case for an owned build is narrow on purpose. Owner-run and mid-market firms carrying a named workflow constraint, running systems that the product market does not represent well, with the budget for the fee, and with the posture to run the system inside their own accounts. That band is where the math works.

Read more about which businesses fit which motion on our industries page.

Most consulting houses do not publish ranked comparisons against their competitors, do not publish the boundary of what they will not build, and do not publish their prices. We publish all three, and the price of every step, because a prospect should be able to work out whether we fit before they ever get on a call with us. We are not optimizing for top-of-funnel volume. We are optimizing for the right four firms each quarter. Publishing the comparisons, the rankings, and the boundaries selects for those firms. See one of these comparisons on our best AI consultants for mid-market law firms guide.