CSATalk to us
Services & pricing

Priced so each step pays for the next

Published numbers, written scope, no retainers you don't understand. Every engagement starts with the proof of concept.

Step one, always

Proof of Concept at cost — $950 flat

About 12 hours at a $75/hr cost basis plus up to $50 of API and hosting. No margin. 2 weeks, one workflow, a working demo you keep whether or not you continue. If you do continue, the full $950 is credited against the build.

How the POC works
Starter Automation
$2,500–$6,000
one-time, plus $75–$150/month to run

One AI-driven workflow, in production.

  • One workflow end to end (e.g. invoice triage, lead intake, appointment follow-up)
  • Runs on a $6–12/month server you own, or ours
  • Hosting, API usage, and light monitoring in the monthly run cost
  • Plain-language runbook so you can operate it yourself
  • Teach & Trust: staff session on your data, two-week shadow period, operating guide
Talk about Starter Automation
Growth Build
$7,500–$20,000
one-time, plus $150–$400/month to run

Connected workflows and a dashboard, wired to the tools you already use.

  • Three or more workflows that hand off to each other
  • Integrations: QuickBooks, Square, Google Workspace, Microsoft 365, spreadsheets, your POS
  • A dashboard showing what the automation did today and what needs a human
  • Audit trail on every AI decision
  • Teach & Trust for every workflow, plus a configuration record you own
Talk about Growth Build
Managed AI
$400–$1,500
per month, no large upfront

We run it, tune it, and review it with you monthly.

  • Monitoring and alerting on every workflow
  • Prompt and model tuning as your data changes
  • Monthly review call with a written summary
  • Small enhancements included; larger ones quoted
  • Staff refresher sessions included when your team changes
Talk about Managed AI
How we implement

Five phases, and the fourth is the one most people skip

Twenty-seven years of ERP go-lives taught us that automation fails at the front desk, not in the code. So training the people and proving the AI on your data is a phase with deliverables, not a courtesy at the end.

  1. Phase 1

    Discover

    One call, no charge

    Find the one workflow where a person is doing something a machine should, and put a number on what it costs today.

    • Written POC scope: the trigger, the inputs, what 'done' looks like
    • Your current cost of the manual process, in hours and dollars

    Exit: You approve a one-page scope.

  2. Phase 2

    Prove

    2 weeks · $950 at cost

    Build the workflow on your real data and show it running before you spend anything meaningful.

    • A working demo you keep
    • A short list of what the AI got wrong on your data and how it was corrected
    • A fixed-price quote for the build

    Exit: You decide to continue; the $950 is credited in full.

  3. Phase 3

    Build

    2–8 weeks by tier

    Make it production-grade: integrations, the audit trail, the human review queue, and the rules for what the AI may decide on its own.

    • The automation, deployed on a server you own or ours
    • Configuration record: every rule, prompt, threshold and example used to tune it, versioned
    • Test run against a month of your history

    Exit: The output matches what your best person would have done, on last month's data.

  4. Phase 4

    Teach & Trust

    2 weeks, included

    Train the people who will live with it, then run the AI in shadow mode so nothing acts until you have watched it be right.

    • One-page 'what the AI decides, what a person decides' sheet per workflow
    • 45-minute hands-on session for your staff, on your data
    • Two-week shadow period: every action reviewed by a person before it goes out
    • Operating guide: how to correct it, pause it, and who to call

    Exit: You sign off and the automation goes live.

  5. Phase 5

    Run

    Ongoing

    Operate it yourself from the runbook, or let us watch it under Managed AI as your data and your business change.

    • 30-day check-in with a written summary of what it did and what it escalated
    • Refresher session whenever staff change (included under Managed AI, quoted otherwise)

    Exit: Measured against the cost you wrote down in Discover.

Two documents leave every engagement: a customer operating guide per workflow (what it does, what it will never decide alone, how to correct or pause it) and an internal configuration record (rules, prompts, thresholds and the examples used to tune them, versioned) so the next person can repeat what we did.

What moves the number

How estimates are built

A build is priced from its tier base, then adjusted for how many distinct workflows it covers, how many outside systems it must talk to (about $1,200 each), your team size, and whether you need it faster than our standard schedule (+15%).

Monthly run cost is real infrastructure: a small server, database, the AI API usage your volume generates, and monitoring. It is billed at cost plus a small margin, and we show you the invoice from the provider if you ask.

Managed AI is optional. Plenty of clients run their automation themselves after handoff using the runbook we leave behind.

Get a ballpark in sixty seconds

The estimator uses the exact model above. Enter what the manual process costs you today and it will show when the build pays for itself.

Open the estimator