Public health AI assistant
PH360™ Chat: ask in plain language, get an answer that cites a source.
Chat is where your staff talk to PH360™, and there are two ways in. Maple is the chat itself: ask a question the way you would ask a colleague and get an answer that points back to the section, subsection, or document it came from. Oak is the builder: your administrators turn your own code and SOPs into new assistants, which then show up in Maple for everyone. Birch sits under both, holding the users, the roles, and the audit trail.
Chat is the part of PH360 you can try free. Create your account and you are in, no credit card and no waitlist.
The two ways in
Maple answers. Oak is how you teach it.
Two applications, one knowledge base. Below them, what makes an answer different from a generic chatbot.
Maple
Ask it
The chat. Three assistants are ready on day one: case investigation and surveillance, PHAB accreditation, and the 2022 FDA Food Code.
Oak
Build it
The builder. Turn your protocols, local code, and SOPs into new assistants in about thirty minutes, with no code.
Birch
Govern it
Underneath both. Users, roles, licenses, and an exportable log of every prompt and response.
Grounded answers
Every answer points back to where it came from: a section, subsection, or document name your staff can go check themselves. No unsourced free text.
Preloaded from day one
Three assistants ship out of the box: case investigation & surveillance, PHAB accreditation, and the FDA 2022 Food Code. No setup, no configuration before your team can start asking questions.
Plain language in, cited answer out
Your staff describe a case, ask a compliance question, or look up a code section the way they would ask a colleague. Chat always shows which assistant is active.
The whole thing, uncut
Watch it draft an outbreak situation report.
Eight and a half minutes, recorded live at our NACCHO webinar on August 18, 2026. Jefferson works a foodborne outbreak in a fictional health department and asks for an incident situation report. He opens the draft, corrects the case counts himself, then asks for new information to be worked in, and it does not overwrite the numbers he just fixed. The last two and a half minutes are the questions health departments actually ask us.
Read the transcript
Machine transcript of the recording, corrected for product names only. Timestamps link into the video.
0:00 And so Maple is, it works similar to how any AI tool that is a chat-based tool that you might work with today. So you put in a question or you ask it to do a task and gives you a response. That's sort of straight out of the box, but straight out of the box with those public health card variables, by citing sources that you can trust and consistent answers every time. Second is Oak, which is a no-code agent builder, which can help you execute on that five-step
0:34 process that I showed you, right? So what it does, it allows you to identify a workflow, identify, give it instructions, and give you, have it draw from specific bodies of knowledge that you reference on your daily basis so that you can make those steps and your workflow faster. I want to always like to start with like, what can this thing not do? And this is a really important point. So I'm going to ask Maple to tell me to create a retirement party for our financial director
1:05 and that I want a funny toast for it. And what you're going to notice is that Maple responds and says that falls outside of my request, my abilities. And it's because we have locked PhD 60 down to be public health only. So every general rate, if you think about it, every other general purpose assistant, AI assistant out on the market today that you could potentially look at, it's gonna be able to instantly and carefully respond
1:36 to that, what ours doesn't. One of the other things that the public health reasoning engine can do and really any part of Maple can do is it can draft word documents for us, which is a really common thing for us in public health. So all I'm gonna do is I'm just gonna ask it to draft an incident situation report for the outbreak. I gave it a date, I gave it an incident number, I gave it those six sections, and then I gave it some facts that I want it to use. And it's gonna go through
2:06 and it's gonna draft those for me. And it's gonna output it as a Word document. Now, what is unique about the way maple works is one of the things that we've heard is that you really want to be able to work with the document with the AI, right? It's not that you want the AI to produce a document, they have to download it, they have to edit it, they have to go back to the AI and like have it do all of these different things. We really want to be able to work, use the AI in our workflow so that you're never leaving it. And so what Maple does is it's going to generate
2:38 that first draft for you. And once that's generated, I'll be able to open it and review it. So two things are really important about that. One is we don't just allow straight downloading of word documents without somebody reviewing it. Remember, our whole goal and design decision here is to help to enforce good AI practices. And good AI practices means we review everything that the AI produces for us before we accept it.
3:08 The second piece about this is I can change this and then go back to ask the AI to do something different. So if we just scroll through here, we look at the epi summary, and let's just say that, you know, for whatever reason, I wanted, you know, as this was coming in, I got new case reports. So I want to confirm cases to go to 23. I want 31 to go to 38.
3:40 38, and then let's see, my total case is to go up to 61, and then we'll do 3, 2, 6. So I'm going to just mark that as reviewed, and notice that I couldn't download it until I do that. So I reviewed it, now I can download it. But the more powerful thing here is that, okay, so I just made those changes. Now I want to go back and update that and I'm going to give it some more information.
4:15 So I'm going to say that one of the confirmed cases is actually a food employee who reports working on show August 15th and the 16th law symptomatic, the person in charge attempted employee attributed employee GI symptoms to the heat and didn't exclude. So what it's going to do is it's not going to take that information and it's going to update my sections for that instant report with that new information and what you'll notice is it won't override the changes that I made in the epi section. So that's what we mean by when we say
4:51 like being able to work with our with our AI assistants instead of having to do an in our workflow rather than having to like go through all of these different things. It's going to tell me what it added so I have a very quick idea here and then it will at the end and then I'll be able to open it up and I can re-review it and notice that my counts are still the same just like I updated 2338 661 all as well but it did get it added those additional pieces for me. So that's a
5:26 power of AI that allows you to work right output to a document work within the document go back give it more information. And so you're constantly able to have PH360 open for like as your as you're always ready assistant. The department, you, you own it. We don't own it. I can't like even as an admin, I can't access your database. I think that's just a thing that we keep security
5:56 on. That's one of the big parts about PH360 is that it you can put PHI in it, you can put PII And it is fully secured. We are audited to SOC 2 standards, as well as we support everyone's HIPAA compliance. We sign a BAA. Any of our clients will, we sign a BAA with to ensure that we can do that. And that's also how we can support things like Willow. So that's one of the key benefits is your staff doesn't have to think about, like, can I put this into an AI?
6:27 Because it's secured. Our platform is secured so you can pretty much put anything into it. So we use AWS as our hosting partner. It does live within our virtual private cloud. So what that means is that everyone's tenant sort of lives within our inside of our virtual private cloud, but your database is your own database with your own key and everything that passes back and forth
6:58 between that database to the agents are their own thing. Every agent you build is its own runtime in our fabric. So it's not like you're sharing that with anyone else. And then your data is your data. So when you decide you're done, we don't keep that data. Yes, so the way that the public health reasoning engine works is it looks for a local requirement first or a local document first. So something that is localized to you through an agent or through a document you upload.
7:31 If it finds it, it will use that information. If it doesn't find it, then it goes out to the national corpus. If it finds something in the national system that are public health reasoning engine, the national piece, it will cite that and use it. If it doesn't have something in either one of those, it will say it doesn't know and stop. So I had to think about it as these concentric circles that the reasoning engine is working from. So in exported Word documents, which is what you saw when you click generate document and it shows on the side, there is a footer in every exported document that says co-created
8:05 with PH360 AI assistance, whether or not you all want to leave that in there is up to you, but it comes with it there as a footer. There is not a limited number of users. Our pricing is on that website. We have population-based pricing. What that means is that this is priced to be accessible to smaller health departments.
- 0:05Maple and Oak, the two ways in
- 0:52What it will not do
- 1:40Drafting a Word document
- 4:03The review gate, and your edits preserved
- 5:46Who owns what we upload
- 5:57PHI and HIPAA
- 6:33Whose servers it lives on
- 7:16Local guidance vs national standards
- 7:54The AI-assistance footer
- 8:13How it is priced
Two design decisions
A person reads it before anyone accepts it.
Two of the choices in that demo matter as much as the drafting does, and both of them are the product refusing to let the AI be the last word.
- You cannot download the document until you mark it reviewed. Not a warning you can dismiss, and not a checkbox in a settings page. Good practice means a human reads what the AI produced before anyone accepts it, so the product enforces it.
- Every exported document carries a footer saying it was co-created with AI assistance. It travels with the file, so the next person to open it knows how it was made without having to ask.
- Your corrections survive the next request. Fix the case counts by hand, then ask for new information to be worked in, and the numbers you fixed stay fixed.
What a question looks like
A tularemia case, at 4:38 in the morning.
This is a real exchange, trimmed for length. Someone opened a case investigation before dawn and needed the current case definition. Chat had it, and named where it came from.
Every response on this page comes from a chat that actually ran in PH360. We trim for length. We don’t write Maple’s answers for it.
Why departments trust the answers
Grounded reasoning is how PH360 works, everywhere, not just here.
This isn’t a Maple-only trick. It’s the reasoning approach every PH360 app is built on.
- Every answer carries a citation a section, subsection, or document name your staff can verify, not a paraphrase you have to take on faith.
- Maple abstains rather than guesses. If it can’t find a grounded source for an answer, it says so instead of inventing one.
- The active agent is always visible. Your staff can see which assistant is answering: case investigation, PHAB, Food Code, or a custom one your department built.
- Compliant with the HIPAA Privacy and Security Rules. Maple runs under F&T Labs’ HIPAA Compliance Program and is part of a SOC 2 audited platform. Full detail lives on the PH360 security page.
- Three assistants preloaded, zero setup. Case investigation & surveillance, PHAB accreditation, and the FDA 2022 Food Code are ready the day you turn Maple on, with no setup.
Build it yourself · Oak
Your own code and SOPs become new assistants.
Three jobs are covered on day one: case investigation and surveillance, PHAB accreditation, and the FDA 2022 Food Code. Everything else your department does, your administrators add themselves in Oak. No code, and no engineering team.
Roughly 30 minutes, start to finish, for one new assistant.
Staff turnover
When someone leaves, what they built stays.
The knowledge your team layers in lives in the platform, not in one person’s head or a spreadsheet only they knew about.
- Any administrator can see and edit any assistant. Nothing is locked to the person who built it.
- Knowledge lives in the platform, not in a personal file nobody else ever saw.
- The workflow outlives the person who built it. When your longtime inspector or accreditation lead retires, what they taught the assistant stays behind for whoever is next.
Govern it · Birch
What your admins can do without calling support.
Birch is the admin layer, included for the people your department designates rather than sold as a seat.
- Provision and activate a new user the day they start, and deactivate one the day they leave. No ticket required.
- Pause or restrict use inside your organization at any time, for any reason, without waiting on F&T Labs.
- Audit chat logs, token usage, and user activity on your own schedule, whether that is a monthly review or a spot check.
- Pull chat history for any licensed user the moment a public-records or Sunshine-law request lands, and export it in JSON, CSV, or XML.
Department leadership answers to a board, a council, or the public, and answering for what an AI tool did on your watch is a real question rather than a hypothetical one. Because the logging and export live in your own admin app instead of a support queue, that answer does not depend on us. More on the PH360 security & compliance page.
The rest of the ecosystem
Five applications, one ecosystem.
PH360 is an environment your department works inside. Maple, Oak, and Birch are the three applications behind Chat, and they are the door most departments come through first. Two more carry the same intelligence into operational work, over the same knowledge base, so what your team teaches it here it knows there.
Environmental health
Application to permit to inspection, in the same system that already knows the code you adopted. Field work runs on a tablet, and a drafted citation waits on the inspector.
Explore environmental healthCase management
Immunization clinics, STI and TB, home visiting, and screening events. The note, the codes, the claim, and the registry report are drafted from the encounter for a clinician to confirm.
Explore case managementCommon questions
Chat FAQ.
What is Chat, exactly?
Chat is the part of PH360 your staff talk to directly. Ask a question in plain language and it answers with a citation attached, pointing to the section, subsection, or document it came from. It is also where the assistants your administrators build show up, and where the audit trail of every prompt and response is kept.
What comes loaded, and what do we add?
PH360 comes preloaded with the FDA Food Code, the CDC’s communicable disease guidance, and the PHAB Standards & Measures, plus three ready assistants: case investigation and surveillance, PHAB accreditation, and the 2022 FDA Food Code. Your state code, local ordinances, SOPs, standing orders, and program binders are what your administrators add themselves. Nothing state-specific ships with the product.
Can it make things up?
It is built to ground every answer in a real source, and when it cannot find one it says so rather than filling the gap. That is the same grounded-reasoning approach every part of PH360 is built on.
Do we need to know how to code?
No. Building an assistant is five steps in plain language: name it, tell it what job it is doing, upload the knowledge, test it yourself, then activate it. Roughly thirty minutes for one assistant, and no engineering team.
Who can build or edit an assistant, and what happens when they leave?
Any administrator your department designates can see and edit any assistant, so nothing is locked to the person who built it. When the person who set up a workflow retires or moves on, the workflow and everything they taught it stay behind.
How does this hold up to a public-records request?
Every prompt and response is logged automatically: user input, response, timestamp, attached documents, model used, and token metadata. Your own administrators pull and export it in JSON, CSV, or XML on their own timeline, without opening a ticket with us.
Is there a free trial, and how is it priced?
Chat is the part of PH360 you can try free. No credit card and no waitlist. After that it is priced by the population you serve and the number of staff on the platform, as Basic (chat and the ready assistants) and Pro (chat plus the builder) seats. Administrator access is included. See pricing for an estimate.
Is our data safe, and does it train a model?
Your knowledge base never trains a model and there is no third-party sharing. Each department runs in its own isolated tenant on AWS Bedrock. PH360 complies with the HIPAA Privacy and Security Rules, with a Business Associate Agreement signed at contract, and holds SOC 2 Type II. Full detail on the security and compliance page.
Is that a real chat in the screenshot?
Yes. Every PH360 response shown on this page comes from a chat that actually ran in the product. We trim for length. We do not write the answers.
Ask it a real question from your own department.
Bring a real case, a real code question, or a real accreditation gap. Start the trial and try it yourself, or bring it to a walkthrough and we will run it with your own documents in front of us.
info@fandtlabs.com · (630) 358-9348 · Back to the PH360 overview
