Queue Tonight. Pull Requests by Morning.
We turn planned jobs into overnight Claude Code runs, with safety rules attached and pull requests ready for human review. Every job gets a receipt with its verdict.
488 of 580 queued jobs delivered
Elevated Labs · 118labs.ai
Build Companies. Put AI to Work.
Our AI queue delivered 488 jobs while we slept, night after night. We build companies and the operating system that runs them.
elevated-labs ~ % drain --tonight → 580 jobs in the queue ✓ 488 delivered, pull requests open for review ✓ 38 sources routed, 0 lost ✓ CI on Spot containers: 0 billed minutes receipt grep -l '^estado: entregue' logging/dispatch/queue/*.md | wc -l
38 sources, 0 lostVoice notes to specs
5 plugins, 29 skills, one sourceTeam skills, always current
0 billed minutesOn-demand CI
36 checkouts foundAd-level reporting
This is the AI era. Everyone is producing something. We produce companies and the operating system that runs them. Elevated minds. Working systems.
Ventures
The companies
We build the company and the operating system together. WeCover is the first company we built and run. Next: a company we are building.
WeCover
AI-native travel-insurance broker selling travel protection to Brazilians traveling abroad.
wecover.travelSolutions
What we build
Wake Up to Work Delivered
Your team rests. The work keeps moving.
We build AI that moves your team's work overnight, draining a written queue into reviewable pull requests and documents, each with its receipt. Every person gets a daily architect session and a morning page. Voice notes become tasks and drafts.
488 of 580 queued jobs delivered
Build With UsAnswer Customers. Keep People in Control.
AI on the line. Your team in command.
We build inbound and outbound AI voice agents plus an AI-first WhatsApp line that hands the thread to a person the moment the customer asks. Every prompt is versioned code with a contract test. Every outbound call is approved before dialing. We keep disclosure configurable per jurisdiction and on by default.
4 versioned, contract-checked agent configurations
Build With UsTurn Ideas Into Scheduled Content
Your ideas. Your brand. A working production line.
We turn your ideas and a weekly plan into scripts and video rendered as code. An AI judge checks every piece against your brand before a person does. Approved posts are scheduled ahead.
10 packaged skills
Build With UsCases
Receipts, Not Promises
Fifteen things we built to run WeCover. Each has a number and its receipt: the command that reproduces it in our repository.
Platform and operations
Say It Once. Tasks and Specs. Nothing Lost.
We turn voice notes and meetings into tasks, delegated work, and draft specs. Commitments land where they belong. Irreversible actions stay in draft, and nightly checks verify the full trail.
38 sources, 0 lost, in one week
Your AI Architect. Your Morning Game Plan.
We give each teammate a scheduled AI session that builds a morning page of meetings, open work, and deadline-ranked priorities. Our authenticated team server gives every Claude session the same tools.
31 morning pages since 15 September; 20 tools on the team server
All Company Knowledge. One Private Address.
We bring our specs, reports, references, and daily team state into a private, searchable HQ. We rebuild it every morning so the team opens a fresh view of the company behind authenticated access.
New Skills Reach Every Teammate. Automatically.
We package reusable AI instructions into audience-specific plugins from a shared git repository. Merged, versioned updates reach every teammate's Claude automatically, with checks enforcing unique skills and versioned changes.
5 plugins, 29 skills, one source
AI-Ready Specs. Every Claim Has Its Command.
We start work with structured specs and finish with ready-to-run AI prompts. Automated checks tie acceptance to what a person can do, and every verified claim includes the command that reproduces it.
169 ops specs; 270 documents end in a prompt an AI can run
Heavy CI. Spot Containers. Only on Demand.
We run heavy CI on short-lived AWS Fargate Spot containers that start with the job and stop at completion. Change-aware rules select relevant suites and reserve heavy runs for review-ready pull requests.
0 billed minutes on the first job
AWS Alarms In. Owned Tasks Out.
We route AWS alarms, health events, and budget alerts into owned tasks with severity and deadlines. Repeat events update the thread, recoveries close it, and our daily architect checks open alarms throughout the day.
50 tests on the relay; live since 11 October 2026
Customer-facing
AI-First WhatsApp. A Person One Message Away.
We built Sol to answer first on WhatsApp and hand the full conversation to a person on request. We keep e-mail available and run without per-seat software.
1,000 free replies a month per number
AI Voice Agents. No Calls Without Approval.
We build agents that book, confirm, follow up, verify policies, and explain coverage. We approve outbound calls, authenticate inbound callers, route unresolved questions to callbacks, and track results daily, with disclosure configurable per jurisdiction.
4 versioned, contract-checked agent configurations
WhatsApp Voice Notes Become Team Tasks. In Pilot.
We're now in pilot with a WhatsApp workflow that returns summaries and open questions, then files action items into the team's task board. We verify signatures and accept messages only from approved numbers.
Ideas Become Scheduled Content. AI Plays Judge.
We turn plans into OpenAI-drafted briefs and scripts, Remotion videos, and generator-built graphics. We run every piece through an AI brand review, then human approval, and schedule approved content directly in Metricool.
10 packaged skills
Campaign Views Hide It. Ad-Level Reporting Finds It.
We report daily at the individual ad level, using direct account data and deterministic checks to surface delivery signals beyond campaign summaries. We keep every change human-approved and read back the result into a receipt.
36 checkouts found
WeCover's Portal. Storefronts Built to Sell.
We built and run WeCover, where travel sellers quote for clients, share branded storefronts or referral links, and track sales and commissions. We keep the canonical commission table directly in the portal.
9 partners approved in the first 60 days
Nightly Cache. WeCover Quotes in Seconds.
We built and run WeCover. We serve US travel insurance quotes from a nightly refreshed rate cache, mapping age-based pricing curves to trip length through a shared carrier integration.
3 carriers, one integration, a nightly cache
Approach
Define It. Build It. Show the Receipt.
- Design
We write the spec and define "done" for the person who will use it.
Deliverable: Written spec with clear acceptance criteria.
- Architect
We design the system, the data, and the gates that refuse a bad result.
Deliverable: System architecture, data flows, and quality gates.
- Implement
We ship the system, measure its work, and keep the receipt.
Deliverable: Working system, measured results, reproducible receipt.
Builds take 2 to 4 weeks, including handover. Optional retainer to run and improve. Few clients at a time.
About
Who is behind it
Franco Bontorin builds companies with AI in the open. With more than twenty years building cloud platforms, the founder of Elevated Labs and WeCover puts elevated minds to work: AI that extends human intelligence and runs real companies.
Contact
Put Your Next Build in Motion
Describe the problem and what done looks like for your team. Write to franco@118labs.ai.