Is this HIPAA compliant / AI-safe by default?

No tool is compliant by default. We design operational workflows and will not claim HIPAA, security, or payer compliance as a property of a model or a Zap. Those requirements are scoped explicitly when they apply.

Will you put a chatbot on our admissions line?

Not as a substitute for a trained admissions conversation. We may use structured intake or after-hours capture with a clear handoff. Families in crisis get a person.

Do you build custom software or only automations?

Both. Some work is CRM automation and internal tools. Some is custom software. The audit decides. We do not bolt AI onto a broken process and call it transformation.

Automation & AI workflows

Automate the handoffs. Keep a human on the decision.

Solvhaus designs operational automations for behavioral health teams: routing, follow-up tasks, reporting, and internal process systems. We are explicit about which bottlenecks are removed, who receives the routed work, and what leadership can see. Tools support operational workflows. They are not a substitute for clinical, compliance, or payer judgment.

Discuss a systems sprint Read the AI documentation guide

Behavioral health workflow automation with a human on every decision.

Solvhaus designs behavioral health workflow automation for treatment operations: referral routing, follow-up tasks, authorization reminders, and reporting that refreshes from working records. We name the bottleneck removed, the person who receives the routed work, and what leadership can see, and we keep clinical, payer, and compliance judgment with people.

01The operational problem

Manual work is not the same thing as care.

Coordinators copy-paste between inboxes, EHR fields, and trackers. BD logs activity after the fact. Leadership waits on a Friday export. Automation can remove that drag, or it can silently mis-route a family. The difference is design and review, not a model name.

The same inquiry is typed into three systems.
Follow-up depends on a sticky note.
Reports are rebuilt by hand every week.
A chatbot was added with no ownership rules.
No one can explain what the automation is allowed to do.

02The system

Same inquiry, two operations.

The first swimlane is how a manual admissions day usually runs. The second is the same day with routing, reminders, and report refresh handled by the system, and every conversation and decision still owned by a person.

Manual admissions handoffsManual

Manual version: an inquiry arrives, the coordinator on shift retypes it, BD hears about the partner later, UR finds out at the end, and leadership waits for a Friday export.
Role	Inquiry arrives	Working it	Verification	Handoff	Reporting
Admissions	Checks the inbox when free	Retypes into EHR and a tracker	Emails VOB request	Texts the night shift	Rebuilds a spreadsheet
BD	Does not know it arrived	,	,	Hears from the partner first	Logs activity after the fact
UR / billing	,	,	Finds the request late	Auth date in a side sheet	,
Leadership	,	,	,	,	Friday export, already stale

Admissions

Inquiry arrives
Checks the inbox when free
Working it
Retypes into EHR and a tracker
Verification
Emails VOB request
Handoff
Texts the night shift
Reporting
Rebuilds a spreadsheet

BD

Inquiry arrives
Does not know it arrived
Handoff
Hears from the partner first
Reporting
Logs activity after the fact

UR / billing

Verification
Finds the request late
Handoff
Auth date in a side sheet

Leadership

Reporting
Friday export, already stale

Automated routing with human reviewWith system support

Automated version: the inquiry is routed to an owner with context, the record is created once, verification is a task with a due date, BD and UR see the same record, and reporting refreshes from it.
Role	Inquiry arrives	Working it	Verification	Handoff	Reporting
System	Creates the record and assigns an owner by coverage	Reminds when a stage sits too long	Opens a VOB task with a due date	Moves the record with its context	Refreshes pipeline and source views
Admissions	Makes the first human call	Works one record, not three tools	Reviews eligibility and fit	Confirms placement	Reads the board in the weekly rhythm
BD	Sees the partner’s referral land	,	,	Closes the loop with the partner	Source-to-outcome by partner
UR / leadership	,	,	Auth status on the same record	,	Same numbers, no data request

System

Inquiry arrives
Creates the record and assigns an owner by coverage
Working it
Reminds when a stage sits too long
Verification
Opens a VOB task with a due date
Handoff
Moves the record with its context
Reporting
Refreshes pipeline and source views

Admissions

Inquiry arrives
Makes the first human call
Working it
Works one record, not three tools
Verification
Reviews eligibility and fit
Handoff
Confirms placement
Reporting
Reads the board in the weekly rhythm

BD

Inquiry arrives
Sees the partner’s referral land
Handoff
Closes the loop with the partner
Reporting
Source-to-outcome by partner

UR / leadership

Verification
Auth status on the same record
Reporting
Same numbers, no data request

03Run it yourself

Map your coverage gap before you buy a chatbot.

Enter your coverage window. The grid shows when an inquiry waits for a human and for how long. After-hours routing to an on-call owner is a rule, not a model, and it closes the gap.

Coverage gap mapInteractive · your numbers

Coverage starts8a

Coverage ends6p

12a

12p

Mon

Tue

Wed

Thu

Fri

Sat

Sun

Seen nowWaits 2d 14h

Longest wait

2d 14h

Worst arrival time in the week

Average wait

14h

Across every hour of the week

Hours uncovered

70%

Share of the week an inquiry waits

Wait for a human owner by day
Day	Longest wait	Average wait	Hours seen immediately
Mon	14h	4h	10 of 24
Tue	14h	4h	10 of 24
Wed	14h	4h	10 of 24
Thu	14h	4h	10 of 24
Fri	2d 14h	16h	10 of 24
Sat	2d 8h	1d 21h	0 of 24
Sun	1d 8h	21h	0 of 24

This is arithmetic on your coverage rules, not a benchmark. It shows when the inbox is the only thing on duty, which is the gap routing and ownership rules are designed to close. Families in crisis still get a person.

04What we build

Automations we will implement

Lead and referral routing

Bottleneck removed: the inbox wait. Who receives it: the owner on coverage, with a fallback. Leadership sees: time to first human response.

Follow-up workflows

Bottleneck removed: the sticky note. Who receives it: the record owner when a stage sits too long. Leadership sees: stage aging by owner.

Reporting automation

Bottleneck removed: the Friday spreadsheet rebuild. Who receives it: nobody has to. Leadership sees: census, pipeline, and source views that refresh from working records.

Document and task workflows

Checklists, handoffs, and request tracking for operational work, not clinical note generation by default.

Internal knowledge and process systems

How we do admissions on Saturday, how we log a lost admit, how we notify UR, captured where the team already works.

Governance and human review

What is automated, what requires a person, who can change a rule, and how failures are caught.

05Fit

Where this belongs

Operators drowning in copy-paste between tools
Admissions teams with coverage gaps after hours
Leadership that wants operating reports without a data request
Teams exploring AI who need workflow design before a vendor demo

06Questions

Before you book a conversation.

07Continue

Related systems.

Admissions & referral operations

Referral management for behavioral health, the workflow most automations should serve.

Behavioral health CRM

Automations fail if the CRM is a mess.

BedFlow

Referral partner CRM where routed work and census updates land.

AI documentation guide for behavioral health

How we evaluate clinical-adjacent AI as operators, not vendors.

See the work Engagement paths Which system is breaking?Why Solvhaus

Next step

If this is the pressure point, start here.

A systems conversation is a working discussion with someone who owns admissions, operations, or growth. Bring the messy version.

Discuss a systems sprint Explore BedFlow