The AI StudioSample engagement · what we build
Sample · Client details anonymized
The AI StudioSample engagement · what we build
Sample · Client details anonymized
Your own private AI platform - built for you, owned by you, running on your own server. A team of AI agents for every employee, the ability to build more, and the first eleven builds chosen straight from your evaluation.
Today your numbers live across many separate systems - the ERP, the spend platform, two ticketing platforms, the counts platform, the operations database, shared drives, and more - and answering a question that spans them means a person stitching exports together by hand.
The true economics of a show, including the ancillary revenue, are not visible in any single tool. At the same time, AI is already entering daily work with no central control, no audit trail, and no shared memory of the business.
The solution - a private AI platform that you own. It runs on your own server, your team logs in with their own Studio accounts, and every person gets their own team of AI agents plus the ability to build more. One place, company-wide, for everyone doing work on company time. You have seen it work. This document is what we build, what it costs, and when we start. The Statement of Work has the full technical detail.
This proposal is custom-built from the six-week evaluation of your business: fifteen people interviewed, six workflows mapped, 57 opportunities identified and force-ranked. The Platform delivers the evaluation’s entire recommended starting set - the top five, the full quick-win cluster, and all three deployments of the categorization core - eleven builds covering 14 of the 57, touching every operating area. The Full Studio delivers all 57.
Not a concept, and not a pitch deck. The program described in these pages is in delivery today: the evaluation is complete, the platform build is under way, and it wraps in Q4 2026.
The evaluation told you where the value is. This proposal is the vehicle that goes and gets it - in the order the evidence says it compounds fastest.
Your evaluation’s own sequencing begins with governance tiering, read-only connectors, verifiable reporting, and earned autonomy. That is not a feature list added after the fact - it is what the Platform is.
Not scoped as “your team can build it.” Built, working, and delivered by us: the top five, the quick-win cluster, and the categorization core deployed three ways - eleven builds, 14 of the 57 opportunities, inside the Platform price.
Your draft AI policy asks for augmentation over replacement, a human in the loop on financial matters, clear ownership, and autonomy that is earned. Those are how the Platform is built: read-only defaults, confirm-or-override interfaces, per-group permissions, and a complete audit trail. The Platform is the enforcement layer for your own policy.
You own the platform, the agents, and everything your team builds. It is portable, so your IT can move it anytime. You are never locked in.
For everyone, in plain English. No code, no dashboards to learn. Ask a question, get an answer from your real numbers.
Your people create new agents themselves, just by describing what they want. No ticket, no wait.
Your team builds their own automations: draft a proposal from last year’s plus live data, handle a recurring email task, assemble a report on a schedule. Anything that does not write into your core databases.
During the build we identify every database, site, and third-party tool worth connecting, and we connect everything we can, with permission-based access so each team sees only what it should.
Per-user keys, a complete audit log of every request, a live cost dashboard by user, and hard spend caps. This is the enforcement layer for your company-wide AI policy.
Ask your books questions in plain English. Take repetitive coding work off the team.
Put your ticketing and counts data to work answering questions you cannot easily ask today.
Check every offer against your real history, and surface the true economics of a show, including the ancillary money the standard tools do not show.
Surface the content and search opportunities that pull organic traffic, draft and organize content, and capture what worked tour over tour.
Automate the admin around advancing and settlement - surveys, packs, parsing - while the relationship layer stays human.
Two agents every employee gets on day one: one that answers from your connected systems, one that builds new agents and automations from a plain-English description.
Two ways to buy. Both are custom to your evaluation. The difference is how much of the 57 we build for you, and how fast.
You run more than 1,000 shows a year. If the platform saves a few hundred dollars of staff time per show, it pays for itself inside the first year - and after that, you own it outright and the rest is gain. Simply having every system’s data in one place is likely to clear that bar before a single automation or agent is built on top.
The investment scales to the size of the prize - each step scoped and priced from the one before it.
| Step | What it is | Time | Investment |
|---|---|---|---|
| The Evaluation | Six weeks of discovery across your business, the full opportunity inventory, and the roadmap that comes out of it. That is Document I. | 6 weeks | A fixed fee, quoted up front |
| The Platform | The platform plus the eleven builds - 14 of the 57, delivered working and yours outright. | 120 days | Priced from the evaluation |
| The Full Studio | All 57 delivered, with the four domain agents built turnkey. | 12 months | Priced from the evaluation |
| An enterprise program | For larger organizations: multiple divisions, deeper integration work, more builds running at once. Scoped as four-month engagement programs, run back to back or in parallel. | 4 months per program | Scoped per program |
What a program is worth is set by the size of the business it changes, so the same method is scoped up or down: the same platform, the same governance, the same way of working, sized to what is on the table.
What you pay for the evaluation comes straight off the price of the build if you proceed within 90 days of the evaluation being delivered. You do not pay twice to learn what to build. And if you decide not to build, the evaluation is still yours to keep and to act on with anyone you choose.
The evaluation’s recommended starting set - built by us, delivered working, accepted on your instance with your data. Eleven builds, covering 14 of the 57, touching every operating area in the business.
| # | Build | From the evaluation | Notes | Area |
|---|---|---|---|---|
| 1 | Natural-language finance reporting | F3 · top 5 #1 | Read-only by design. Finance leadership’s confirmed first target - the trust-building opener. | Finance |
| 2 | The categorization core, deployed three ways | F1 + WA7 + Mk5 · top 5 #2 | One engine with human confirm/override, deployed to AP pre-coding, ad invoices, and venue-expense mapping. | Finance · Marketing · Event Ops |
| 3 | Subsidiary settlement automation | iEO1 · top 5 #3 | The best-quantified number in the evaluation (5–7 hours every Monday). Scoped with the PCI review. | Event Ops |
| 4 | SQL warehouse connector | T3 · top 5 #4 | The only current path to the two-year ticketing record. Unlocks pricing, segmentation, pacing, and demand work downstream. | Ticketing |
| 5 | Venue-expense auto-populate + offer flagging | B1 + B6 · top 5 #5 | The revenue-side anchor, serving the 400-to-550 offers-per-year goal. Current-data scope; the legacy backfill is a scoped add. | Booking |
| 6 | ZBA daily transfer automation | F12 · quick win | Deterministic, fast, visible. | Finance |
| 7 | New-vendor completeness check | F13 · quick win | Deterministic, fast, visible. | Finance |
| 8 | Income-recognition file generation | F6 · quick win | File-generation-for-review; nothing posts on its own, per your governance principles. | Finance |
| 9 | End-of-tour marketing rollup | Mk6 · quick win | The zero-baseline learning loop across roughly 1,000 shows a year. | Marketing |
| 10 | Offer-sheet templating | B3 · quick win | The lowest-barrier booking win. | Booking |
| 11 | Post-event survey automation | EO1 slice · quick win | The event team’s own pick. The full advancing repository lives in the Full Studio. | Event Ops |
Deployment 1 puts pre-coded invoices in front of the accounts-payable workflow. Deployment 2 applies the same engine to ad-invoice coding. Deployment 3 applies it to venue-expense mapping, alongside build 5. Each is confirm-or-override; nothing posts into your accounting system on its own.
Finance five builds · Booking two · Event Operations two · Ticketing one (the one that unlocks the most) · Marketing one, plus a share of the categorization core. The point is not spread for its own sake: it is that every division’s power user has something real, in their own workflow, inside the first 120 days.
If you sign, what does the engagement actually fulfill inside the evaluation? Here is the answer, line by line: where every one of the 57 opportunities lands. Nothing is left without a home.
Built by us, inside the Platform engagement: the fourteen items above, delivered as the eleven builds. Every remaining item stays fully buildable by your own team on the foundation we hand over.
Delivered inside the four domain agents - Finance, Ticketing, Booking, Marketing - in the evaluation’s sequence, plus the deeper data work. All 57. Two PCI-scoped subsidiary items are delivered with a qualified compliance specialist we coordinate; one builds on the in-house venue hub already in flight rather than replacing it.
| ID | Opportunity | Lane | Notes |
|---|---|---|---|
| F1 | AP pre-coding agent | Platform | Categorization core, deployment 1 |
| F2 | Ad-spend allocation agent | Full Studio | Finance agent · fed by show-level ad data |
| F3 | Natural-language finance reporting | Platform | Deepens across the engagement |
| F4 | Forecast-variance flagging | Full Studio | Finance agent |
| F5 | Accounting-settlement automation | Full Studio | File-generation-for-review; after settlement assembly |
| F6 | Income-recognition + accruals | Platform | File-generation-for-review |
| F7 | Cash-forecast auto-population | Full Studio | Finance agent |
| F8 | Working-capital forecasting by artist | Full Studio | Finance agent |
| F9 | Audit-support assembly | Full Studio | Finance agent |
| F10 | Settlement three-way variance compare | Full Studio | Needs structured settlement data first |
| F11 | Margin-anomaly explainer | Full Studio | Finance agent |
| F12 | ZBA daily transfer automation | Platform | Quick win |
| F13 | New-vendor completeness check | Platform | Quick win |
| ID | Opportunity | Lane | Notes |
|---|---|---|---|
| T1 | Feed watchdog agent | Full Studio | Paired with the connector |
| T2 | Daily-count platform historical backfill | Full Studio | Ticketing agent |
| T3 | SQL warehouse connector | Platform | The prerequisite |
| T4 | Dynamic price-recommendation engine | Full Studio | Gated on T3, which the Platform delivers |
| T5 | Daily-count dashboard + pacing flags | Full Studio | Ticketing agent |
| T6 | Kill manual daily-count entry | Full Studio | Ticketing agent |
| T7 | Email management assistant | Full Studio | Ticketing agent |
| T8 | Internal profitability dashboard | Full Studio | With the access-control layer |
| T9 | Sold-map visibility, non-vendor venues | Full Studio | Ticketing agent |
| T10 | Historical feedback loop | Full Studio | Ticketing agent |
| ID | Opportunity | Lane | Notes |
|---|---|---|---|
| Mk1 | Single source of truth for conversion data | Full Studio | Marketing agent |
| Mk2 | AI budget management + dayparting | Full Studio | Marketing agent |
| Mk3 | Actionable-flag + visibility dashboard | Full Studio | Marketing agent |
| Mk4 | Audience building from ticket-buyer history | Full Studio | Marketing agent · unlocked by T3 |
| Mk5 | Ad-invoice auto-coding | Platform | Categorization core, deployment 2 |
| Mk6 | End-of-tour reconciliation rollup | Platform | Quick win |
| Mk7 | Search ads + SEO | Full Studio | Service engagement inside the Studio |
| M1 | Customer segmentation + lookalikes | Full Studio | Unlocked by T3 |
| M2 | Ticket velocity / efficient-frontier | Full Studio | Unlocked by T3 |
| ID | Opportunity | Lane | Notes |
|---|---|---|---|
| B1 | Venue-expense auto-populate + historicals | Platform | Current-data scope; legacy backfill in the Full Studio |
| B2 | Centralized venue / deal-knowledge base | Full Studio | One shared build with EO1 / EO5, seeded from the existing hub |
| B3 | Offer-sheet templating | Platform | Quick win |
| B4 | Auto-populate scaling tiers | Full Studio | Compounds directly on B1 |
| B5 | Routing base + availability / conflict integration | Full Studio | Booking agent |
| B6 | Anomaly / risk flagging on offers | Platform | Rides behind B1 |
| B7 | Venue rebate / ancillary optimizer | Full Studio | Validated first, as the evaluation recommends |
| B8 | Competitive-analysis + routing bolt-on | Full Studio | Phase 2, after B5 |
| B9 | Artist evaluation / demand model | Full Studio | With the evaluation’s honest data caution |
| B10 | Financial vs demand model + Monte Carlo | Full Studio | Booking agent |
| B11 | Offer-to-settlement reconciliation loop | Full Studio | Scoped once, cross-division |
| B12 | Intuition-to-data bridge | Full Studio | Piggybacks B11 |
| ID | Opportunity | Lane | Notes |
|---|---|---|---|
| EO1 | Show-advancing repository + pack assembly | Platform | Survey slice in the Platform; full repository in the Full Studio |
| EO2 | Show-settlement assembly agent | Full Studio | Populate-never-bypass, per the two-eyes culture |
| EO3 | Production template-reuse | Full Studio | Assist-only |
| EO4 | Contractor agreement + 1099 automation | Full Studio | Coordinated with the in-house build; light compliance check |
| EO5 | Venue-data hub + mileage grid | Full Studio | Built on the in-house hub already in flight |
| iEO1 | Subsidiary settlement automation | Platform | PCI scope review |
| iEO2 | Subsidiary AI-assisted customer service | Full Studio | Deflect-and-escalate |
| iEO3 | Subsidiary PCI phone card capture | Full Studio | With a PCI-qualified specialist, coordinated inside the Studio |
| ID | Opportunity | Lane | Notes |
|---|---|---|---|
| WA1 | Ad-pack auto-assembly | Full Studio | Generalizes the existing 2-hour-to-15-minute win |
| WA2 | Email / attachment intake agent | Full Studio | Cross-cutting |
| WA3 | Cross-system re-keying elimination | Full Studio | The subsidiary instance is PCI-scoped |
| WA4 | Subsidiary event-build automation | Full Studio | With a PCI-qualified specialist, coordinated inside the Studio |
| WA7 | Expense / document categorization core | Platform | One build, three deployments |
| Lane | Items |
|---|---|
| Platform | 14 - the eleven builds |
| Full Studio | 43 more - everything else, turnkey |
| Total | 57 - every one with a lane |
Every one of the 57 has a lane, and none requires you to hire developers, stand up infrastructure, or project-manage AI work. Your team’s job is to use the platform and tell us where it should go next.
Every figure below comes from the evaluation - stated by your team on a dated call - and is deliberately conservative: low end of every range, roughly twenty opportunities carried at zero, revenue-side items counted only collectively.
| Already quantified | Figure from the evaluation | Conservative annual estimate |
|---|---|---|
| AP pre-coding (F1 → the categorization core) | Two AP temps at ~$40–45/hr loaded - a cost you are already paying, that grows with volume | $175–195K across the core’s three deployments |
| Subsidiary settlement (iEO1) | 5–7 hours every Monday - roughly 250–350 hours a year | $10–15K - bounded, recurring, provable in month one |
| Natural-language reporting (F3) | Reporting “laborious to build and run”; 3–5 power users | $15–30K direct - plus the foundation every later finance build reuses |
| Warehouse connector (T3) | 60–65M records unreachable; feed breaks fixed by hand | $10–20K direct - the real value is the nine revenue-side items it unlocks |
| Venue-expense auto-populate + flags (B1 + B6) | 400 → 550 offers/yr at ~20 min per offer; mis-keyed offers swing “tens of thousands” | $40–70K - capacity absorbed without proportional hiring + one materially caught error a year |
| The first five, combined | ≈ $250–330K per year at run-rate |
The hard savings de-risk the program; the return is made on the revenue side. The estimated hard savings more than cover the ongoing run cost on their own - and they recur and compound with volume. The revenue-side opportunities - offer accuracy against real history, ancillary-revenue visibility, routing, pricing, and segmentation on the two-year ticketing record - are therefore pursued with the downside already covered.
Dynamic pricing, the profitability view and its ancillary “secret money,” segmentation and velocity on the 60-million-row dataset, audience building, the rebate optimizer, and the untouched search channel are the largest value pools in the inventory - and the least responsible to dollarize item by item today, because their yields depend on data that becomes reachable only after the connector work. At your scale, a single point of operating margin is worth several million dollars a year.
The engagement’s own model pairs the full investment profile - the build fee plus first-year and ongoing run costs - against the deliberately slow benefit ramp (25% → 65% → 100% of run-rate), on an illustrative $300 million revenue base held flat. We have removed the investment figures from this sample, since every engagement is priced from its own evaluation; the shape of the result carries over: a roughly 9–10× five-year return in the conservative one-point case, with payback landing inside Year 1. A working session with your finance team substitutes your actuals; the arithmetic is mechanical.
The evaluation runs straight into the build, so everything we learned about your business goes directly into what you use. There is no wasted discovery phase - the work compounds directly into the finished product.
Phase 1 · months 1–4 - the Platform and the eleven builds, exactly as above. Phase 2 · months 3–6 - the data-spine wave: feed watchdog, historical backfill, categorization deployments hardened, the four domain agents stood up. Phase 3 · months 5–9 - the revenue tier: pricing recommendations, segmentation and pacing on the connected spine, the profitability view with its access-control layer, venue-expense history and scaling tiers, the venue-knowledge base. Phase 4 · months 9–12 - the strategic tier: the offer-to-settlement reconciliation loop, cash-forecast automation, the demand model, the remaining document-assembly family, and the PCI-scoped subsidiary builds with the specialist. Each phase has its own acceptance and its own payment.
Three ongoing costs, all billed directly to you, in your name, with no markup.
Runs the whole company on the best model for each job: one provider for reasoning and writing, others for images, creative, real-time, and high-volume work. Where you land in the range is your call. You set per-user caps and see every dollar by user.
We keep the system tuned and future-proof: monitoring and uptime checks, auditing it, reviewing spend against caps, adding connectors, and swapping in better and cheaper models as they ship, so your platform keeps getting sharper instead of frozen in time. When something breaks, we fix it. This is a real service line, not a token: it runs at roughly one and a half to two percent of the build price a month, which is the standard band for software of this size, and it is set at signature against the tier you choose.
A dedicated server in your name, sized for roughly 140 metered users on the connector layer plus your power users on heavy models. The only hard cost beyond the AI itself.
Every dollar of it under your control. The AI number is a dial, not a fixed bill: lower caps on light users, higher caps on the people who use it most, adjusted as real usage shows where the value is. Power users get named, flat-rate model accounts (about $200/month each, around five to start); everyone else runs on the metered, capped, audited backend and never holds a personal AI subscription.
We do not hand you a manual and walk away. Your power users get hands-on time so adoption sticks and they become self-sufficient - informal working sessions during the power-user beta, company-wide enablement at the Day-60 rollout as each division comes on, and ongoing group and one-on-one time for power users through the rest of the engagement, aimed at the point where your builders create agents without us.
The default terms are half at signature and half on delivery. Delivery means a milestone you have accepted and signed off, not a date that has passed.
Both structures run against the same scope, the same delivery, and the same acceptance criteria. The difference is only where the money sits and what releases it.
Simple, standard, and the lowest total.
Each installment is released by an acceptance you sign, not by a date on a calendar. This option carries a modest premium for spreading the payments and carrying the training past delivery.
A payment schedule tied to dates transfers schedule risk to you. A payment schedule tied to acceptances keeps it with us. If a milestone is late, the payment attached to it is late. You do not pay for work you have not accepted.
Every user gets their own isolated AI workspace, reachable two ways - a plain-English Console for everyone, and a full Builder Workspace for power users - over a shared connector layer that safely brokers your real systems, wrapped in an audit, secrets, and cost-control spine.
Each system gets its own secure broker between the AI and the data. Sensitive fields walled off, every query capped, nothing destructive can run. Any write path is explicit, narrowly scoped, and audited.
Each user and the company as a whole get a persistent memory. The platform learns your business - the chart of accounts, the show taxonomy, each person’s conventions - and stops asking to be re-taught.
An encrypted vault for keys, a complete audit log, a live cost dashboard by user, and spend caps you set per group. The CFO’s instrument panel.
The companion Statement of Work carries the full architecture, the connector list, the security model, the delivery milestones, the data-accuracy verification loop, and the acceptance conditions for each of the eleven builds.
The platform is read-only by default, by design, to protect your data. Building agents that write into a high-value live database is next-level work: it requires a scoped build and a tested backup strategy so a bad write can never cost you real data. A later engagement, never a default.
On the Platform, new agents and workflows beyond the eleven builds are what your team builds inside the platform. On the Full Studio, all 57 are ours to deliver; anything beyond the 57 is scoped when it comes up.
Any parallel engagement - for example a paid-advertising platform - stays separately contracted and is not part of this build or price. It can connect to the platform through a read-only data feed so both sets of data answer questions together.
Each of the eleven carries an explicit boundary, stated in the Statement of Work. The boundary is both the ceiling and the floor: we deliver at least it, and we never quietly widen it.
That is what the included training and the plain-English Ask-Anything bar are for. There are no dashboards to learn, your power users are self-sufficient before we step back, and the first eleven builds put something real in every division’s own workflow inside 120 days.
We map and connect every system worth connecting during the build, and you see exactly what the platform can and cannot reach as we go - not after it is finished. Where a vendor API cannot reach the data, we build the connector that can, as we did for the ticketing warehouse.
You own the code from day one and the whole platform is reproducible from a script, so your IT can move it to any host, anytime. Portability is a build requirement, not a promise. At handoff we strip every credential of ours off the server.
You replace scattered, ungoverned AI subscriptions with one owned, audited, cost-capped asset your CFO controls - and you own it outright instead of renting it forever. The evaluation’s own numbers put the first five builds at $250–330K a year in hard savings alone.
No - it gives them a foundation. The evaluation reviewed the in-house work in flight and the roadmap coordinates with it rather than rebuilding it. Your builders stop rebuilding the same wheel division by division; the platform gives their work memory, connectors, and governance.
Accuracy is verified by your own people, system by system, from the first connector onward: we connect and self-check, hand your designated owner a verification pack, they check it against the source, we fix anything wrong at our cost. No connector counts as delivered until its owner has signed it off.
The evaluation tells you where the value is. The Studio is the vehicle that goes and gets it. A green light means signature and first payment start the clock: we schedule the PCI scope review and the warehouse access the same week, and your team is signing in to working software within the first 30 days. The program described here is in delivery now - evaluation complete, platform build under way, wrapping Q4 2026. We would rather show you results than projections.