How an employer's request becomes a worker on a worksite — and a worker safely home — through a transparent flow that an AI orchestrates and a human supervises. Every step. Every signature. Every cost. In one platform.
Whether you send us a job order PDF, a list of names and phone numbers, or last year's contract, our platform converts it into a live, tracked, evidence-grade process — and gives every actor in the chain the right view at the right moment.
Real life is messy. We meet the employer where they are — from a clean job order to a half-finished spreadsheet to last year's contract — and the platform routes the work accordingly.
The employer sends a job order (PDF, email, RFP). They need workers — they don't have anyone in mind yet.
The employer already chose workers — but only sends names and phone numbers. We must gather everything else.
The employer sends last year's contract or worker list — they want the same crew back. We have to know which ones came through us.
Whatever the input, the platform produces a single, structured contract record with: workers identified, services scoped, fees estimated, consulate assigned, AM owner, recruiter, kickoff date. From there, the seven phases run the same.
Every action in the H-2 process belongs to someone: the worker, our team, our AI agents, an external vendor, or the employer. The platform tells each one when it's their move, and keeps the others in sync.
Each phase has a dedicated AI agent, a clear human owner, an auditable output, and a measurable SLA. Nothing is left to email threads or memory.
The employer creates the contract themselves in the H2 Organizer portal — uploading the documents they have. The platform parses, prefills and structures. The AM only steps in when something needs a human.
From "we need 60 strawberry pickers" to "here are 80 pre-validated candidates who said yes" — in hours, not days.
A US-side filer (the employer's law firm or US agency) submits ETA-9142 and I-129. When approved, the certified data and I-797 land in H2 Organizer — and the platform activates the MX-side workflow.
Labormex does not file ETA-9142A/B with DOL, nor I-129 with USCIS. Those are handled by the US-side filer (the employer's law firm or the partnering US agency). LM receives the certified case data, validates it, and runs everything downstream from the consulate.
Petition approved. Now: DS-160 per worker, MRV payment, group appointment booking, document prep — without 200 spreadsheets.
The 4-day reality at the consulate, transformed: kiosks for check-in, a PWA for reps, smart lockers for documents, geofencing for arrival.
From the consulate to the worksite, then through the season — every leg scheduled, every paycheck attested, every complaint heard.
The trip ends the way it started: in MiChamba. Return travel, expense reconciliation, debrief, and a pre-loaded offer for next season.
Every successful return adds signal to the matching engine. Each year, the system gets sharper, workers more reliable, and the employer's experience smoother.
Each contract can mix any of these — billed to the agency, the employer, or split. The platform tracks every quote, every vendor selection, every actual cost against estimate.
Note: DOL ETA-9142 and USCIS I-129 filings are done by the employer's US-side filer. Labormex ingests the certified data.
Beyond the traditional H-2 services, the platform unlocks a set of premium offerings that were too expensive to deliver manually but are economical at scale with AI.
Job orders · ETAs · employer messages · worker complaints. Translated, validated, attached to the audit trail.
CEFR A1–C2 scoring in 8 minutes through MiChamba, with audio sample & written test. Per-worker pricing.
OCR + passport NFC chip read for government-issued data. Works from MiChamba on the worker's phone (registration & login) and from kiosks on consular day. Cross-checks photo, MRZ, biographic data — zero room for fake profiles.
Trained on H-2 rules + contract specifics. Replaces the late-night AM call. Escalates real cases to humans.
Voice interview → structured CV pdf, employer-ready. Faster than worker typing it themselves.
Predict which workers won't show, which will abscond, which contracts will run late. Acted on, not just reported.
AM team replies 3× faster. Bilingual, tone-matched. Trained on AM voice · in pilot.
One-click DOL / WHD audit response. Worker file, signatures, payments, communications. Premium tier.
Bland AI calls 50 candidates at once. SLA-priced per successful contact, not per call.
Every Mexican passport since 2008 carries a small NFC chip (ICAO 9303 eMRTD). It stores the holder's biographic data and photo, signed cryptographically by the issuing government. Any modern phone can read it — once it has the key.
Same technology used by gov.uk, Onfido, Jumio, EU eIDAS apps. Already field-proven at scale.
A worker can't sign up with a fake name. The first time they open MiChamba they tap their passport — the app reads the chip, validates the signature, and creates their account with identity confirmed by the Mexican government. Login afterwards uses face match against the chip photo.
Workers arrive at the consulate, tap their phone to the kiosk (or scan their passport at the kiosk directly). The kiosk reads the NFC, confirms identity in seconds, prints the labels for their documents, opens their locker. Three minutes per worker, no representative needed for routine cases.
Every checkpoint of the trip — hotel check-in, CAS biometrics queue, border crossing — the worker taps the phone on the rep's PWA reader. Status updates land in the dashboard in real time. The employer sees "checked into hotel · 8:14pm" without anyone typing anything.
Beyond the passport, an H-2 file needs a moving target of documents depending on the job, the employer, the consulate, and the worker's history. MiChamba is where the worker takes a photo of each one — and where AI confirms, in seconds, that it's the right document, complete, current, and matches the person on file.
Driver placements (poultry, trucking, agricultural transport) require materially more documentation than a typical H-2 ticket — and the costs of a mistake are higher. Our pipeline detects "this is a driver contract" from the job order in Phase 1, automatically requests the right set of documents through MiChamba, runs CDL-specific OCR (front + back of license, endorsements, restrictions), and pre-validates the driving record before the employer ever sees a CV. What used to take a week of back-and-forth happens in 48 hours, end-to-end.
Labormex coordinates every vendor in the chain — transport, lodging, medical, photo, translation, fianza, courier — and pays each one directly. The client receives a single invoice with a transparent markup, and a single point of accountability.
Workers never pay anything. Vendors invoice us, not the client. The client never has to chase down a hotel in Monterrey or a doctor in Hermosillo. This is the operational difference that lets the platform — and the AI — see everything end-to-end.
Labormex bills the US-side agency partner who fronts the worker package; common in association / cooperative-style employers.
Labormex bills the employer directly. Used for direct-hire contracts and most mid-size operations.
Some services billed to the agency, others to the employer. The platform routes each line item to the right invoice automatically.
Small, well-bounded components that talk through clear contracts, run on a mix of local and cloud compute, and stay easy for the team to evolve.
Where workers, employers and operators meet the platform.
A small team of specialized agents, each with a clear job, supervised by a human owner.
The canonical record of every worker, every employer, every event.
Tech that lives where the work happens: at the consulate, on the bus, at the locker.
Each tier handles what it does best. Sensitive personal data stays close to home; frontier reasoning happens in the cloud only when it must.
Beyond the AI agents, a set of internal cockpit tools keeps the team sharp: live dashboards, real-time tickets, natural-language analytics, and audit-ready reports.
The single pane of glass for the operations team. Every contract, every appointment, every worker — live, with WebSocket updates the moment something changes.
Ask the database anything, in Spanish, in plain words. The agent understands every status code, worker type, crossing flag and contract enum used internally — no SQL needed.
Tickets, tasks & hand-offs between AMs, recruiters, reps, logistics and onboarding. Drag & drop kanban with real-time notifications.
One agent, trained on Labormex tone and data, serving the client in the two channels they actually use: an interactive chat inside H2 Organizer, and a quiet email drafter inside the AM team inbox. Same context, same escalation rules, same audit trail.
Shared escalation: billing $, legal questions, complaints & sensitive cases always go to a human.
Health metrics on the AM team's Gmail & Drive workspace — response times, thread aging, document handling. Feeds H2 Buddy's behavior model.
At every place the platform interacts with a worker, an employer or a vendor, an AI sensor watches for frustration, confusion or risk. When the signal crosses a threshold, a human is paged with full context. No silent dropouts. No quiet escalations.
Tone, pace, interruptions, raised voice, frustration phrases ("ya les dije", "no entienden") detected in real time.
Sentiment in messages, repeated failed actions, abandoned flows, explicit complaint button presses. Anti-fee guard also lives here.
Frustration markers ("URGENT", "ASAP", all-caps, "I asked already", "this is the third time"), tone escalation across thread turns.
Repeated scan failures, timeouts, abandonment mid-session, prolonged inactivity. The kiosk literally calls a representative over to help.
Repeated identical questions, agent-can't-answer streak, escalation requests, drop-off in active session. Tells us the product itself isn't enough.
Payroll complaints, attendance drops, missed check-ins, late-night chatbot anxiety, mentions of mistreatment or unpaid wages.
Mild sentiment dip, slow response, single repeated question.
Clear frustration phrase, multiple failed actions, urgency markers.
Anger, abandonment, "I want to cancel", mentions of fees / mistreatment.
Trafficking signals, illegal fee, threat, injury, worker missing.
Every escalation is reviewed weekly. Patterns that should have been caught earlier are fed back into the sensors. The platform gets calmer because we get sharper.
If a worker drops out, abandons MiChamba, or stops responding for 48+ hours, that itself triggers a sensor — silence is a signal.
Sentiment models run on transcripts and signals — not on biometrics or surveillance. Workers know what's measured. Audit trail preserved.
Every action in our platform produces evidence. Every signature is time-stamped. Every payment is reconciled. Every call is recorded. Every form is versioned. The platform doesn't just help us pass audits — it helps employers pass theirs.
Offers come from the system, recorded calls, MiChamba complaint button, recruiter commission tied to confirmed placements only.
Auto-generated recruitment report. SWA job order tracked. U.S. applicants logged with dispositions, ready for any DOL inquiry.
AEWR / prevailing wage validated against SOC code at form-fill time, before the package is ever submitted.
Cross-check across CURP, passport MRZ, photo, voice, prior seasons. Reduces fraud and protects the worker's own identity.
Per-worker file is one click away: ETA-9142, I-129 evidence, DS-160 receipts, signatures, payments, communications.
Tight cadence, board reviews quarterly, shipping starts week one. Everything below is sized to a team of 6–12 engineers shipping in parallel.
2026 Q3 → Q4
2027 Q1 → Q2
2027 Q3 → Q4
Every person on the team moves from doing repetitive work to supervising AI that does it for them. Capture becomes review. Outreach becomes closing. Field representation becomes case management. The job gets more interesting, more human — and the same team handles four times the volume.
Spreadsheets, paperwork, repeated phone calls.
Reviewing what the AI flagged, deciding hard cases.
Hours dialing, chasing leads, retyping forms.
Talking to confirmed candidates, building relationships.
Counting passports, chasing missing docs, manual lists.
Supervising the hardest cases, the ones AI can't solve.
The employer creates the contract themselves. The platform handles the rest. The worker is identified, contacted, prepared, screened, scheduled, transported, supported and brought home — and every step is visible to whoever needs to see it.
"The network. The data. The team. The consulate footprint.
Now: the platform that turns all of it into a product."