ClockCatcher — clock-in catcher for Medicaid home-care agencies

1. One-liner

Catches missed caregiver clock-ins in real time and captures audit-proof EVV verification before a clawback hits.

2. Trend signal — why now?

Electronic Visit Verification (EVV) went from “have a system” to “stay under a number” in 2026. States flipped from soft edits (minor errors paid with a warning) to hard edits (automatic denial, no grace period). The new audit hammer is the manual-edit rate — every retroactive edit a coordinator makes because a caregiver didn’t clock in/out in real time.

The thresholds are now explicit and punishing:

• Pennsylvania: no more than 15% of EVV records may be manual edits; exceeding it for two consecutive quarters triggers a formal non-compliance notice and a required Corrective Action Plan.
• Minnesota: compliance floor jumped to 80% in early 2026.
• Multiple states: exceeding a 15–20% manual-edit rate triggers a compliance review, a corrective-action letter, or direct payment holds.
• Maryland: caps mistakes at four per caregiver per month — Medicaid simply does not pay the agency for the fifth.

And the root cause is documented, not behavioral: “the fix isn’t disciplining caregivers… high manual entry rates trace back to EVV software that creates friction in the field — apps that load slowly, don’t handle poor connectivity, or require more steps than caregivers can reliably complete at the start of a shift.” Agencies are told to “review EVV exceptions daily: missed clock-ins, GPS mismatches, time discrepancies, before they stack up.” That daily scramble is the workflow we automate.

Provenance:

• Signal 1 (demand): 2026 hard-edit enforcement + manual-edit thresholds (PA 15% over 2 quarters → Corrective Action Plan; payment holds for excessive manual corrections) — https://www.myezcare.com/state-by-state-evv-requirements-for-2026-what-every-home-care-agency-must-know/ — 2026-06-10
• Signal 2 (feasibility): Cheap real-time AI voice/SMS now production-grade — OpenAI gpt-realtime-mini, audio ~$0.06/min, sub-second latency, no-code builders (Retell/Synthflow) — https://www.forasoft.com/blog/article/openai-realtime-api-voice-agent-production-guide-2026 — 2026-06-10
• Signal 3 (economic): ~9,961 US home-health agencies, 5M+ Medicaid HCBS recipients across 300+ programs; documented payment withholds for excessive manual edits; Maryland non-payment past 4 mistakes/caregiver/month — https://homecentris.com/mistakes/ + https://www.kff.org/medicaid/medicaid-home-care-hcbs-in-2025/ — 2026-06-10 Category: Regulatory arbitrage

3. The opportunity

The EVV systems of record — HHAeXchange, Sandata, AuthentiCare, CareBridge — are paid by states to record visits, not to chase caregivers and prevent edits. Their incentive ends at “data captured.” Most small Medicaid-only agencies are mandated onto the free aggregator portal (HHAeXchange’s free Mobile App + IVR), which records visits but does nothing active when a caregiver no-shows the clock.

Full-suite platforms (CareSmartz360, AxisCare, Alora, myEZcare) do bolt on “missed clock-in reminders” — but a reminder that a caregiver ignores still ends as a manual edit. None of them watch the live manual-edit ratio against the specific state threshold and run an automated capture-and-document loop that produces audit-ready reason-code evidence.

The gap is a thin compliance overlay that sits on top of whatever EVV the agency already runs (free aggregator or paid suite), monitors the manual-edit rate in real time, and intercepts at-risk visits before they tip the agency over the line. We don’t rip and replace the system of record — we defend it. That’s a focused wedge incumbents are structurally disincentivized to build.

4. Target market

• Primary customer: Owner / scheduling coordinator at a small US Medicaid home-care agency — 5–75 caregivers, personal-care / 1915(c)-waiver services, operating in open-model EVV states (PA, TX, IL, MN, OH, NC, MD, etc.).
• Why they buy: “Every billing cycle I’m manually fixing 20–30% of visits because caregivers forgot to clock in, and now the state is threatening to hold my Medicaid payments over that exact number. One bad quarter and I’m on a corrective action plan.” Margins are thin (personal-care rates $15–30/hr) and a payment hold is existential.
• Rough TAM reasoning: ~9,961 US home-health agencies plus tens of thousands of non-Medicare personal-care/HCBS agencies. Conservatively 15,000–25,000 agencies live in open-model EVV states with a manual-edit exposure. Capturing 1,000 at ~$300/mo = $3.6M ARR.
• Why now for them: 2026 is the first year manual-edit rate carries automatic financial penalties. The clock-in problem they’ve tolerated for years just became a balance-sheet risk.

5. Product sketch (MVP)

• Live manual-edit meter: pulls the agency’s visit + exception data (via aggregator export/API or, where needed, a thin portal sync) and shows the rolling manual-edit % against that state’s threshold, with a red line.
• Real-time miss interception: the moment a scheduled visit window opens with no clock-in, ClockCatcher texts the caregiver a one-tap verify link; if ignored, it escalates to an AI voice call that confirms presence, captures start/end time, service, and location by voice.
• Audit-proof reason-code documentation: every capture generates the state-valid reason code + a timestamped note and (optional) client-confirmation record — the exact evidence auditors demand — instead of a bare “Other.”
• Daily exception worklist: replaces the manual “review EVV exceptions daily” scramble with a ranked queue of only the visits that still need a human.
• Threshold early-warning: alerts the owner before the rolling rate crosses the state line, not after the corrective-action letter arrives.
• Caregiver friction report: flags which caregivers / which times-of-day generate the most misses, so owners fix the source.
• Multi-state rule pack: state-specific thresholds, reason-code lists, and consecutive-quarter logic baked in.

6. AI angle — what’s load-bearing

Two places. First, the voice capture: when a caregiver ignores SMS, an AI voice agent calls and conducts a compliant verification conversation — confirming the six required data points (service type, recipient, caregiver, date, start/end, location) and writing a clean reason-code note. That’s only viable now because real-time voice dropped to ~$0.06/min with sub-second latency. Second, note generation: turning a messy voice/SMS exchange into audit-grade, reason-code-mapped documentation that survives a state review. Remove the AI and you’re back to a coordinator phoning each caregiver and hand-typing notes — i.e., the exact manual labor that creates the problem. The AI is the product, not a garnish.

7. Localization angle (if any)

N/A — this is a US-only play by design. The wedge is the regulatory localization: per-state EVV thresholds, reason-code tables, and aggregator quirks (HHAeXchange vs. Sandata vs. AuthentiCare). The moat is encoding 15+ state rule packs correctly — there is no global version because EVV is a US Medicaid construct.

8. Business model — path to $1M–$5M ARR

• Pricing: tiered by caregiver count — $199/mo (up to 15 caregivers), $399/mo (16–50), $699/mo (51+). Anchored well below a single avoided clawback and below the $24–36K/yr verification suites quoted for mid-size practices.
• ACV: ~$4,000–$5,000.
• Math to $1M ARR: ~220 agencies × $375/mo × 12 ≈ $1.0M.
• Math to $5M ARR: ~1,100 agencies at the same blended ACV, or fewer agencies with a per-AI-call usage add-on and a “managed exception desk” upsell.
• Expansion path: per-state expansion (each new state = new addressable agencies), usage-based AI-call overage, and an upsell where ClockCatcher fully works the daily exception queue as a done-for-you service at a higher tier.

9. Go-to-market wedge — first 100 customers

• State Medicaid provider directories are public lists. Most open-model states publish enrolled HCBS/personal-care provider rosters with addresses. Scrape PA + IL + MD (high-threshold-pressure states), filter to 5–75-caregiver agencies, send a personalized one-pager: “Your state now holds payment if manual edits exceed 15%. Here’s a free 60-second read of your last quarter’s rate.” — lead with the threshold, not the feature.
• Free manual-edit audit as the hook. Offer a no-cost “where do you stand vs. your state’s line” report from a single exported visit file. It’s a falsifiable, scary number that converts to a paid pilot.
• Ride the aggregator transition moments. When a state announces a new threshold or hard-edit date (frequent in 2026), run targeted outreach + a webinar with a state home-care association. Associations (e.g., PHA, state chapters of HCAOA) have member lists and actively warn about compliance.
• Billing-company channel. Outsourced home-care billers ($3–10/visit) feel the manual-edit pain on every claim. Partner: they resell ClockCatcher to their book of agencies; revenue share.

10. Build complexity — justification

Medium. Off-the-shelf: AI voice (Retell/Synthflow/OpenAI realtime), SMS (Twilio), standard web app, state rule packs as config. Custom work: reliable ingestion of visit/exception data from aggregators — some have APIs (HHAeXchange EDI, Sandata), others need scheduled CSV export sync, which is fiddly per state/aggregator. The compliance-grade note + reason-code mapping must be exactly right per state. A pair can ship a single-state v1 in 10–14 weeks; multi-state and deeper aggregator integrations push to ~4–5 months.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	A compliance-assist overlay; reads the agency’s own data with consent. No regulated entity status needed.
Ethical — no harm / dark patterns	✅	Helps agencies capture true visits accurately; reduces fraud risk, doesn’t manufacture records. Must guardrail against fabricating un-worked visits.
Market exists (evidence above)	✅	2026 thresholds, documented payment holds, 5M+ recipients.
1–5 person team can build this	✅	Pair ships single-state v1 in ~3 months.
Launchable with <$50K / ₹40L	✅	API + dev time; no capex.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Hair-on-fire in 2026: payment holds and corrective-action plans tied directly to a number the agency can’t currently control daily.
Demand evidence	15	12/15	Strong regulatory + vendor signals; documented penalties. Light on direct verbatim owner quotes (forums sparse) — verify in interviews.
Build feasibility	15	11/15	Voice/SMS off-the-shelf; aggregator data ingestion per-state is the gnarly part.
Distribution clarity	15	11/15	Public provider directories + association channel + billing-company resellers; conversion math realistic but unproven.
Revenue mechanics	15	12/15	Clear $/caregiver pricing, ACV well below avoided clawback, benchmarked vs. $24–36K suites.
Time to first revenue	10	8/10	Free-audit hook → paid pilot is a weeks-not-months funnel.
Defensibility	10	5/10	Moat = multi-state rule packs + workflow lock-in + aggregator integrations; suites could bolt on a meter, so speed matters.
Total	100	76/100

13. Qualitative modifiers

Founder-fit tags

operations-heavy · domain-expertise-required — you need someone who understands EVV mechanics state-by-state and can talk credibly to agency owners. Not a pure-tech play.

Key assumptions to validate (3–5)

1. Assumption: Small agencies will pay $199–699/mo to stay under their state’s manual-edit threshold. How to test: 20 owner interviews + 5 paid pilots from the free-audit hook in PA/IL.
2. Assumption: We can reliably ingest visit/exception data from at least HHAeXchange and Sandata without an official partnership. How to test: Build the ingestion for one state, confirm export/API access with 3 real agencies.
3. Assumption: An AI voice call recovers enough missed clock-ins to materially drop the manual-edit rate (e.g., a 25% rate → under 15%). How to test: Run the loop for 3 agencies for one billing cycle, measure before/after.
4. Assumption: State auditors accept ClockCatcher-generated reason-code documentation. How to test: Have a compliance consultant / state association review sample records.

Risk flags

1. Platform dependency: Aggregators (HHAeXchange et al.) could restrict data export or ship their own meter. Mitigate by being aggregator-agnostic and moving fast on the rule-pack moat.
2. Regulatory drift: State thresholds and reason-code lists change often — maintenance burden is permanent (also a barrier to entry, so double-edged).
3. Ethics / fraud line: The product must only capture real visits. A tool that nudges toward fabricating verification is a compliance and legal landmine — hard guardrails required.

14. Structured verdict

Score:                  76/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Operations-minded founder with EVV/home-care domain advisor
Time to revenue:        6–10 weeks (free-audit hook → paid pilot)
Capital to launch:      $8–15K ($ for voice/SMS APIs + dev time)
Top 3 assumptions to validate first:
  1. Owners pay $199–699/mo to stay under state threshold — 20 interviews + 5 paid pilots
  2. Reliable HHAeXchange/Sandata data ingestion without partnership — build for 1 state, test on 3 agencies
  3. AI voice loop drops a 25%+ manual-edit rate under 15% — one billing cycle, 3 agencies, before/after
Kill criteria:
  - Abandon if <3 of 20 interviewed owners convert to a paid pilot
  - Abandon if the AI capture loop can't move the manual-edit rate below the state line in pilot data
  - Abandon if a major aggregator ships an equivalent live-meter + capture loop before your multi-state v1

15. Next step — 1-week validation sprint

• Day 1–2: Scrape the PA + IL Medicaid HCBS provider directories; filter to 5–75-caregiver agencies; build a list of 200 with owner contacts.
• Day 3–4: Cold-email/call 60 with the “free manual-edit audit” offer; for any that send an export, hand-compute their rate vs. the state line.
• Day 5: Decide go / no-go on a falsifiable bar: ≥5 agencies hand over a visit export AND ≥3 of those are already over (or within 5 points of) their state threshold AND verbally commit to a paid pilot if the loop drops their rate. Under that, the pain isn’t acute enough to bleed — go back to signals.