Comprehensive AI-run Dental Office System
Budget / Salary$1,000–10,000
TypeFreelance project
LocationRemote
Posted1 hour ago
WHAT WE ARE BUILDING
We are a dental practice group in Los Angeles. We want a fully AI-run office and we want to own the platform that runs it.
One office is the pilot, more offices follow, and it has to be buildable into a product we can license to other practices. That is a design constraint, not a someday idea. If your architecture only works for one location with hardcoded settings, it is the wrong architecture.
Today nothing talks to anything else, every report is built by hand, and we pay monthly for platforms that do a fraction of what AI can now do directly. We want to replace that with our own. This is a long-term engagement, not a one-off gig.
WHAT WE RUN TODAY
Open Dental, Slack, Trello, Google Workspace, Google Ads, Google Analytics, Google Business Profile, Facebook and Instagram, a marketing website, and a GoHighLevel-based platform for ads, pipelines and patient follow-up.
The GoHighLevel-based platform is what we intend to replace with what you build. We run no project management tool and do not want one bolted on. Boards live inside the platform you build. We are not hosting our own infrastructure, so propose hosting we can own and pay for directly.
ARCHITECTURE
Not a pile of Zapier scenarios. An orchestrated agent organization, structured like a real business.
- An apex orchestrator that routes work
- Department agents: operations, marketing, finance, front office
- Task sub-agents under each, each with a narrow job
- A shared skill and tool library, so capability is added once and inherited everywhere
- A vetting gate before anything third-party gets connected
- Per-agent logging, cost tracking and observability
If you want to replace that with one chatbot and some webhooks, we are not a fit.
1. OUR OWN AI-NATIVE CRM
Built AI-first, not automation-first. Matching what we pay for now is the floor, not the goal.
- Lead capture from every source: ads, forms, calls, Google Business Profile, social DMs, walk-ins
- Pipeline stages that advance on what actually happened, not on someone updating a record
- AI follow-up over SMS, email and chat that answers real questions, handles objections and books
- Speed to lead measured in seconds
- Recall and reactivation generated from the patient base, not from static uploaded lists
- Attribution from ad spend through to booked and completed production
- Every outbound message logged and reviewable, with approval gates where we want them
If you think we should buy this instead of building it, say so and make the case. A straight answer counts in your favor.
2. AI-POWERED MARKETING, ALL OF IT
Continuously optimizing, not a human running a monthly checklist.
- SEO: local search, service and location pages, content, schema, technical health, monitored and adjusted continuously
- Google Maps and local pack: profile optimization, posts, photos, categories, Q and A, tracked by map position
- GEO and AEO: we want to be surfaced when someone asks ChatGPT, Claude, Gemini or Perplexity for a provider in our area. If those acronyms mean nothing to you, skip this project.
- Google Ads and Meta Ads: creative generation and testing, budget shifts driven by booked production, not clicks
- Reviews: generation, monitoring, response drafting held for approval
- Content engine: blog, social and email on a schedule, on brand, approval before anything publishes
- Analytics: every source unified and attributed down to booked appointments and revenue
3. THE DASHBOARD
One screen that shows the true state of the business and lets us act on it.
- Pulls from every system above plus the phone system
- New patients, production, schedule gaps and open chair time, unconverted leads, missed and unreturned calls, treatment presented versus accepted, recall status, spend against booked revenue
- Alerts when something is off, not just charts we have to interpret
- Role-based views: front office, hygiene, manager, owner
- Per-location views and a roll-up from day one, even with one location
- Readable on a phone
4. INTRA-OFFICE OPERATIONS
The part that changes daily life in the office, and the part most bidders underestimate.
- Kanban boards inside the platform that move because the work actually happened in a connected system, not because someone dragged a card
- Manager issues a directive, it becomes a structured task on the right board, assigned to the right person or agent, with a due time and a definition of done
- Agents complete what they can and escalate what they cannot, with the reason
- Slack as the human surface: agents post to the right channel, staff reply there, and the reply advances the task
- Auto-generated morning huddle brief and end-of-day recap
- Nothing sends, posts, charges or writes to a system of record without a human approval gate
- Accountability reporting: what was assigned, what got done, what stalled and where
5. AGENT FRAMEWORK
So we can add capability without hiring you again every time.
- An agent registry plus a scaffolding or template mechanism
- Tool and permission scoping per agent, least privilege by default
- Approval gates, kill switches, rollback
- Logging and cost tracking per agent
- An interface where a non-programmer can create, pause and retire agents
Agents we want over time: recall and reactivation, backfilling unfilled appointments, insurance verification chasing, new patient follow up, review requests, content production, competitor monitoring, monthly reporting, staff task follow up.
MULTI-LOCATION AND PRODUCT
Build against these from the start. Retrofitting them later is a rewrite.
- Multi-tenant: a second or third office is configuration, not a code change
- Per-location data isolation
- Configurable branding, services, hours, providers and pricing, so it can run under another practice's name
- A repeatable onboarding path for standing up a new practice
- No hardcoded practice-specific values anywhere
DATA
This is a healthcare practice, so patient data is handled carefully. Phase 1 is non-patient-data only: marketing, website, SEO, ads, reviews, social, analytics, and task management with no patient identifiers. Anything touching patient data is a later phase on a separate agreement, on an architecture we approve first. Tell us in your proposal how you keep patient data out of Phase 1 entirely.
OWNERSHIP. NON-NEGOTIABLE.
- Every account, domain, API key and subscription under our ownership from day one
- All code in a repository we own, documented well enough for another developer to take over
- Full IP assignment, including the right to license and resell the platform
- You do not reuse or resell this build, or a near copy of it, to other dental practices. Expect that term in the contract.
- No black box that only you can maintain
WHO WE WANT
- You have shipped a real multi-agent or AI automation system that is running in production right now, and you can show it to us
- You have built a CRM or a marketing platform, not just connected other people's
- You have built multi-tenant software, or can explain clearly how you would here
- You have done local SEO and paid media for a service business and can prove it moved something real
- You can explain your thinking to a non-technical owner without hiding behind jargon
- You are available for months, not a one-week drop and run
BUDGET
No fixed number. Propose the scope and price it yourself, broken into milestones with a deliverable we can verify at each one. We will pay for a short discovery and architecture milestone before committing to the full build. We will pay properly for the right person and we are not choosing on lowest price.
TO BE CONSIDERED
Copy-paste proposals get ignored. Answer these in your bid, numbered.
1. Link us to one AI agent or automation system you built that is live in production today. What does it do, who uses it, how long has it been running?
2. Should we build our own AI CRM or keep buying one? Make the actual argument, either way.
3. How would you build an orchestrator with department agents and task sub-agents under it? What is the real mechanism, and what stops it producing a broken or runaway agent?
4. How do you make a Kanban card move because the work genuinely happened in a connected system, and not because someone dragged it?
5. How do you architect this so a second and third location is configuration rather than a rebuild, and so it can be white-labeled and sold to another practice?
6. Which tools in your proposed stack would need a BAA, and why?
7. In a medical office, where do you draw the line between an agent acting on its own and a human approving first? Give us your rule.
8. Estimate the monthly running cost of your proposed stack, split into LLM and API usage, hosting, and subscriptions.
9. Which parts of this project would you refuse to automate, and why?
We are a dental practice group in Los Angeles. We want a fully AI-run office and we want to own the platform that runs it.
One office is the pilot, more offices follow, and it has to be buildable into a product we can license to other practices. That is a design constraint, not a someday idea. If your architecture only works for one location with hardcoded settings, it is the wrong architecture.
Today nothing talks to anything else, every report is built by hand, and we pay monthly for platforms that do a fraction of what AI can now do directly. We want to replace that with our own. This is a long-term engagement, not a one-off gig.
WHAT WE RUN TODAY
Open Dental, Slack, Trello, Google Workspace, Google Ads, Google Analytics, Google Business Profile, Facebook and Instagram, a marketing website, and a GoHighLevel-based platform for ads, pipelines and patient follow-up.
The GoHighLevel-based platform is what we intend to replace with what you build. We run no project management tool and do not want one bolted on. Boards live inside the platform you build. We are not hosting our own infrastructure, so propose hosting we can own and pay for directly.
ARCHITECTURE
Not a pile of Zapier scenarios. An orchestrated agent organization, structured like a real business.
- An apex orchestrator that routes work
- Department agents: operations, marketing, finance, front office
- Task sub-agents under each, each with a narrow job
- A shared skill and tool library, so capability is added once and inherited everywhere
- A vetting gate before anything third-party gets connected
- Per-agent logging, cost tracking and observability
If you want to replace that with one chatbot and some webhooks, we are not a fit.
1. OUR OWN AI-NATIVE CRM
Built AI-first, not automation-first. Matching what we pay for now is the floor, not the goal.
- Lead capture from every source: ads, forms, calls, Google Business Profile, social DMs, walk-ins
- Pipeline stages that advance on what actually happened, not on someone updating a record
- AI follow-up over SMS, email and chat that answers real questions, handles objections and books
- Speed to lead measured in seconds
- Recall and reactivation generated from the patient base, not from static uploaded lists
- Attribution from ad spend through to booked and completed production
- Every outbound message logged and reviewable, with approval gates where we want them
If you think we should buy this instead of building it, say so and make the case. A straight answer counts in your favor.
2. AI-POWERED MARKETING, ALL OF IT
Continuously optimizing, not a human running a monthly checklist.
- SEO: local search, service and location pages, content, schema, technical health, monitored and adjusted continuously
- Google Maps and local pack: profile optimization, posts, photos, categories, Q and A, tracked by map position
- GEO and AEO: we want to be surfaced when someone asks ChatGPT, Claude, Gemini or Perplexity for a provider in our area. If those acronyms mean nothing to you, skip this project.
- Google Ads and Meta Ads: creative generation and testing, budget shifts driven by booked production, not clicks
- Reviews: generation, monitoring, response drafting held for approval
- Content engine: blog, social and email on a schedule, on brand, approval before anything publishes
- Analytics: every source unified and attributed down to booked appointments and revenue
3. THE DASHBOARD
One screen that shows the true state of the business and lets us act on it.
- Pulls from every system above plus the phone system
- New patients, production, schedule gaps and open chair time, unconverted leads, missed and unreturned calls, treatment presented versus accepted, recall status, spend against booked revenue
- Alerts when something is off, not just charts we have to interpret
- Role-based views: front office, hygiene, manager, owner
- Per-location views and a roll-up from day one, even with one location
- Readable on a phone
4. INTRA-OFFICE OPERATIONS
The part that changes daily life in the office, and the part most bidders underestimate.
- Kanban boards inside the platform that move because the work actually happened in a connected system, not because someone dragged a card
- Manager issues a directive, it becomes a structured task on the right board, assigned to the right person or agent, with a due time and a definition of done
- Agents complete what they can and escalate what they cannot, with the reason
- Slack as the human surface: agents post to the right channel, staff reply there, and the reply advances the task
- Auto-generated morning huddle brief and end-of-day recap
- Nothing sends, posts, charges or writes to a system of record without a human approval gate
- Accountability reporting: what was assigned, what got done, what stalled and where
5. AGENT FRAMEWORK
So we can add capability without hiring you again every time.
- An agent registry plus a scaffolding or template mechanism
- Tool and permission scoping per agent, least privilege by default
- Approval gates, kill switches, rollback
- Logging and cost tracking per agent
- An interface where a non-programmer can create, pause and retire agents
Agents we want over time: recall and reactivation, backfilling unfilled appointments, insurance verification chasing, new patient follow up, review requests, content production, competitor monitoring, monthly reporting, staff task follow up.
MULTI-LOCATION AND PRODUCT
Build against these from the start. Retrofitting them later is a rewrite.
- Multi-tenant: a second or third office is configuration, not a code change
- Per-location data isolation
- Configurable branding, services, hours, providers and pricing, so it can run under another practice's name
- A repeatable onboarding path for standing up a new practice
- No hardcoded practice-specific values anywhere
DATA
This is a healthcare practice, so patient data is handled carefully. Phase 1 is non-patient-data only: marketing, website, SEO, ads, reviews, social, analytics, and task management with no patient identifiers. Anything touching patient data is a later phase on a separate agreement, on an architecture we approve first. Tell us in your proposal how you keep patient data out of Phase 1 entirely.
OWNERSHIP. NON-NEGOTIABLE.
- Every account, domain, API key and subscription under our ownership from day one
- All code in a repository we own, documented well enough for another developer to take over
- Full IP assignment, including the right to license and resell the platform
- You do not reuse or resell this build, or a near copy of it, to other dental practices. Expect that term in the contract.
- No black box that only you can maintain
WHO WE WANT
- You have shipped a real multi-agent or AI automation system that is running in production right now, and you can show it to us
- You have built a CRM or a marketing platform, not just connected other people's
- You have built multi-tenant software, or can explain clearly how you would here
- You have done local SEO and paid media for a service business and can prove it moved something real
- You can explain your thinking to a non-technical owner without hiding behind jargon
- You are available for months, not a one-week drop and run
BUDGET
No fixed number. Propose the scope and price it yourself, broken into milestones with a deliverable we can verify at each one. We will pay for a short discovery and architecture milestone before committing to the full build. We will pay properly for the right person and we are not choosing on lowest price.
TO BE CONSIDERED
Copy-paste proposals get ignored. Answer these in your bid, numbered.
1. Link us to one AI agent or automation system you built that is live in production today. What does it do, who uses it, how long has it been running?
2. Should we build our own AI CRM or keep buying one? Make the actual argument, either way.
3. How would you build an orchestrator with department agents and task sub-agents under it? What is the real mechanism, and what stops it producing a broken or runaway agent?
4. How do you make a Kanban card move because the work genuinely happened in a connected system, and not because someone dragged it?
5. How do you architect this so a second and third location is configuration rather than a rebuild, and so it can be white-labeled and sold to another practice?
6. Which tools in your proposed stack would need a BAA, and why?
7. In a medical office, where do you draw the line between an agent acting on its own and a human approving first? Give us your rule.
8. Estimate the monthly running cost of your proposed stack, split into LLM and API usage, hosting, and subscriptions.
9. Which parts of this project would you refuse to automate, and why?
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.