SaaS gave you tools. AaaS gives you results.
Software as a Service sells you a login and leaves the work to your staff. Algorixtec agents do the back-office work, every day, and hand you the result to approve by reply. They run on the systems you already pay for, inside your own boundary, with a person signing off on anything that sends, pays or files. Priced by agent, month to month.
Four things every agent comes with
An agent is a system we build and run that does one defined piece of back-office work the way a careful employee would, and then shows its work. These four parts are not optional, because together they are what makes AI safe to put inside a regulated office.
An agent that owns one task, end to end
Not a chatbot you ask things. The agent takes the work in, does it, files the result where it belongs and reports back, on a schedule or when something arrives. One job, defined in writing, with what “done” means.
The human check
Agents propose; people decide anything that sends, pays, files or deletes. Anything uncertain, and anything that matters, waits for a named person, so an agent never signs a payroll, pays an invoice or files a record on its own.
A morning brief, reply to approve
Each run ends in a short report in your inbox or Teams: what came in, what was done, what needs a person. Exceptions come with a one-line approve or reject. No new app to open.
The ledger
Every run is written to an append-only ledger: who asked, when, what the inputs were, what it found and who approved. If it cannot be reconstructed afterward, it was not under control, and a surveyor or an auditor can be shown it.
What our agents do today at a California home health agency
At one California home health agency, our agents run more than a dozen daily jobs. They track every caregiver credential, screen staff against federal exclusion lists, audit vendor invoices, close payroll the morning after, answer the phone and keep the survey binder current. Every run is logged. Every exception waits for a person.
| Agent (role) | What it does today | Status | Case study |
|---|---|---|---|
| Compliance | Tracks every caregiver credential, screens staff against the federal exclusion list monthly, watches CMS, federal and California rule sources weekly | Live | Compliance, AI assistants |
| Back office | Audits vendor invoices, reconciles route sheets against visits, closes payroll the morning after and chases what is missing | Live | Finance and billing, Payroll and operations |
| Intake | Reads scanned start-of-care packets and physician orders and files them | Built and verified, switched off on purpose | Intake, Orders and signatures |
| Front desk | Answers the phone line, routes calls to queues, turns voicemail into tasks; answers staff handbook questions | Live | Communications, AI assistants |
| Readiness | Keeps the survey binder and Medicare pre-claim review packets current | Live | Not yet written up |
| Reputation | Watches Google reviews and tells the owner the same day | Live | Not yet written up |
| Training | Assigns in-service modules, tracks hours per aide, issues certificates, in five languages | Built as Faravix; not yet running for an agency’s staff | Not yet written up |
Statuses are exact. Names, vendors and identifying details are removed from every case study. We publish hours or dollars saved only when we have measured them.
The case studies behind the table
Every caregiver credential in one trustworthy picture
A weekly audit and a color-coded board replace three sources that disagreed, and a monthly federal exclusion screen runs alongside.
Audit every Monday, board always current, exclusion screen monthly.
Running in productionRead the case study Finance and billingPay therapy invoices only for visits that are documented
Vendor invoices are matched to visit notes in the EMR automatically, and the office gets one report: ready to pay, wait, already paid, or needs a person.
Hand-ticked paper invoices became one weekly report.
Running in productionRead the case study Payroll and operationsRoute sheets to payroll, with exceptions caught before checks are cut
Paper and photographed route sheets are read, matched to EMR visits and filed, and payroll gets an exceptions-only list instead of a pile.
94.4% of handwritten rows read correctly by a small local model.
Running in productionRead the case study IntakeAdmission paperwork scanned once and filed to the right chart
A start-of-care packet scanned at the office printer is read, matched to the right patient and split into one document per form, with completeness checked against the agency’s own admission policy.
42 pages in, 7 forms recognized, 18 blank backs dropped, patient pinned unanimously.
Built and verified, switched off on purposeRead the case study Orders and signaturesPhysician orders filed to the right doctor’s folder, never the wrong one
Loose therapy orders awaiting signature are identified by patient and filed under the physician of record from the EMR, not the one printed on the PDF.
21 of 21 loose orders identified with no OCR and no AI.
Built and verified, switched off on purposeRead the case study CommunicationsEvery call answered or captured, with voicemails that become tasks
A Teams Phone menu routes callers by type, and unanswered calls become tracked tasks with a transcript instead of a blinking light.
A voicemail becomes a task in 8 seconds.
Running in productionRead the case studyHow Agents as a Service replaces Software as a Service
Every app you add is another login, another password, another tool someone has to remember to open. Agents work the other way around. They open the tools for you, do the job, and send you the result. You pay for work done, not for seats nobody uses.
Look at what Software as a Service sells: a seat. A login for each person, and a tool that person must learn, open, feed and drive. The work is still yours. If the office gets busier you buy more seats, and the pile of work is the same size the morning after.
An agent changes what is being bought. You do not buy access to a tool. You buy a piece of work, finished: the route sheets read and matched, the invoices checked against the visit notes, the credential board current on Monday morning, the voicemail turned into a task with a transcript. The agent does that work. A person on your staff approves it.
The honest version, which is also the stronger pitch. AaaS does not throw your software away. It operates the software you already pay for. The EMR, payroll and Microsoft 365 stay; the agent becomes the staff member who never forgets to log in. That is why it replaces the next software purchase and most of the clicking inside the current ones.
Side by side
| Software as a Service | Agents as a Service | |
|---|---|---|
| What you buy | Access to a tool | Work done |
| Who does the work | Your staff, inside the tool | The agent; your staff approves |
| Priced by | Seat, per user, whether used or not | By agent, for the work it runs: from $495 per agent per month |
| Where it lives | Another login, another password | Your inbox: a morning brief, reply to approve |
| Integration | Your problem; “we have an API” | Ours. If your system has no API, the agent works it the way your staff does |
| Rules | Generic, configured by you | Built around your rules, your state, your payers |
| Proof it ran | Usage dashboards | A ledger of every run, what it found, who approved |
| When it fails | You find out at audit | It tells you that morning |
| Shelfware risk | High: paid, unused | None. A month in which an agent misses its scheduled runs for reasons on our side is not billed for that agent. The run ledger is the evidence. |
Human approval on anything that sends, pays or files
Patient data never leaves your control. A signed BAA when an agent touches protected health information. An off switch on every agent.
We learned this in HIPAA-regulated home health, where the rules are strict and a mistake has a name attached. The same rules apply to every agent we run, in any organization. How we handle HIPAA and the BAA.
The five things every owner asks about
- A human approves. Agents propose; people decide anything that sends, pays, files or deletes.
- Your data stays home. Patient information never goes to an AI provider without a signed BAA; where none exists, the agent runs on dedicated hardware we control.
- We sign a BAA for any agent that touches protected health information.
- Switch-off is a feature. Every agent has an off switch, and we say which ones are off and why.
- Nothing is locked in. Runbooks, configuration and the ledger are yours.
Assess, deploy, run, prove, teach
- You and us
Assess
A read-only review of where staff time goes, a ranked list of the jobs an agent can take, and a fixed scope and price in writing before anything starts. Nothing changes while we look.
- Us
Deploy
One agent at a time, two to four weeks per agent, built on your own systems and around your rules, your state and your payers. Tested on your real work with the output held back, and documented in a runbook before it goes live.
- Us and you
Run
The agent works daily. You get a short brief; exceptions come with a one-line approve or reject. We watch the first weeks closely and fix what the first real run shows.
- Us
Prove
A monthly ledger report: runs, findings, approvals and misses. When we have measured a result, it goes on the case study, misses included.
- Us
Teach
Training for the people who approve, read the briefs and own the ledger, so the capability stays with you and not with us.
Agents as a Service FAQ
Is this AI making decisions for us?
No. Agents propose; people decide anything that sends, pays, files or deletes. The agent does the work and hands a named person at your organization the result, with a one-line approve or reject. Where a step is certain and has no judgment in it, it runs on its own, and even then the ledger shows what it did.
What about HIPAA?
Patient data never leaves your control. Patient information never goes to an AI provider without a signed BAA; where none exists, the agent runs on dedicated hardware we control. We sign a Business Associate Agreement for any agent that touches PHI and work with the minimum necessary access. Whether your organization is compliant is a determination for your compliance officer and counsel, and no one is “HIPAA certified”, because no such certification exists.
What if the agent is wrong?
It will be, sometimes, and the design assumes it. Anything the agent is not sure about is routed to a person instead of acted on. Every run is written to an append-only ledger: what came in, what it did, what it found and who approved, so a mistake can be seen, traced and corrected. When we measure an agent’s accuracy we publish the number on the case study, and we publish the misses with it.
What does it cost, and how long does it take?
Agents are priced by agent, month to month, from $495 per agent per month, and you get a fixed scope and price in writing before anything starts. Deploying an agent takes two to four weeks, one agent at a time. A month in which an agent misses its scheduled runs for reasons on our side is not billed for that agent. The run ledger is the evidence. The first agent is scoped small on purpose, so the price and the result are both easy to judge.
Do we have to replace our software?
No. AaaS does not throw your software away. It operates the software you already pay for. The EMR, payroll and Microsoft 365 stay; the agent becomes the staff member who never forgets to log in. Where a plain script does the job without AI, we build that instead, and we say so.
What do we keep if we stop?
Everything that was built in your accounts: the configuration, the runbooks, the ledger and the training. We write the runbook so another provider could pick it up. Nothing is locked in.