Outworx
Products

Everything we ship, one team.

We dogfood the same playbook we sell. Four flagship platforms, eleven more products across web, desktop and mobile, and the open-source tooling around them — every one built and run by the same engineers we place with our outsourcing clients.

Underlayer logo

Underlayer

Your product already has users. Now it can teach them.

Training you embed — an API, an iframe, and a white-label player that looks like it was always part of your product. No portal, no seat licences, and nobody signs up for anything. Generate courses from a doc or a prompt, theme every block, track every answer, and export SCORM when a customer insists.

Free sandbox · Build $149/mo · Scale $599/mo · Partner $1,800/mo — billed per course view, not per seat

  • 01

    No learner accounts

    Pass the id you already have for that person on the iframe URL. No sign-up, no password, no invitation email — the course remembers where they stopped and what they answered.

  • 02

    39 blocks. None look like a form.

    Six question types, flashcards, drag-to-match, hotspots, timelines, decks, scenarios, code, polls, and reflections — every one takes your theme.

  • 03

    Generate a course in one call

    Turn a doc, a URL, or a prompt into a structured course with a quiz. Treat it as a first draft — you edit what it got wrong.

  • 04

    It should look like you built it

    Four colours by job, two fonts, a corner radius, and a density — applied to every block and every state, light and dark.

  • 05

    Right-to-left, done properly

    Arabic and Hebrew mirror the whole layout — spine, arrows, drag direction — not just the text. Add a translation and pass ?lang=.

  • 06

    Results land on your server

    course.completed, certificate.issued, and screen.viewed arrive as signed webhooks — or query every view, answer, and completion over the REST API.

  • 07

    SCORM, when they insist

    Export SCORM 1.2 or 2004 for customers who won't leave their LMS. It still reports back to you. Courses as JSON and learner records as CSV, on every plan.

  • Built and operated by the same team that ships your outsourced engagements.
Intilq logo

Intilq

Turn one idea into a full, scheduled, multi-channel campaign.

Describe a launch in a sentence and Intilq generates a structured 7–90 day campaign — captions, hooks, hashtags, and on-brand images or video — then schedules and auto-posts it across Instagram, TikTok, LinkedIn, Facebook, and X. Tracked-link analytics and white-label reports included. From $0.

Free · Solo $24/mo · Pro $49/mo · Agency $199/mo — 3-day Pro trial, no card

  • 01

    Describe it once, get a full plan

    One prompt becomes a structured 7–90 day campaign — hooks, captions, hashtags, and creative prompts, tailored to each platform's voice.

  • 02

    Multi-channel auto-posting

    Connect Instagram, TikTok, LinkedIn, Facebook, and X. Intilq schedules, posts, retries on failure, and logs every status.

  • 03

    On-brand images & video

    Every post ships with a generated image or AI video — in your colors and voice. No designer required.

  • 04

    Analytics that matter

    Reach, engagement, and click-through from tracked links that work on every platform — per post, channel, and account, with exportable white-label reports.

  • 05

    Multi-brand & agency white-label

    Run multiple brands or clients from one dashboard, with team seats and white-label reporting on Agency.

  • 06

    Drag-drop calendar & scheduler

    Batch-schedule an entire campaign, then override timings and fine-tune the calendar with drag-and-drop.

  • 07

    Bilingual, RTL-ready

    Ships in English and Arabic out of the box — full right-to-left support for Gulf and MENA audiences.

  • Built and operated by the same team that ships your outsourced engagements.
Marsad logo

Marsad

Errors, traces & analytics in one SDK.

One SDK carries errors, network calls, performance traces, Web Vitals, and cookie-free analytics out of your Next.js and Expo apps — including native crashes that happen before JavaScript is running. Source maps resolve automatically, and an MCP server lets your AI assistant triage production issues. Install, call init(), done.

Free · Solo $9/mo · Team $29/mo — flat, no per-seat charges

  • 01

    Errors grouped into issues

    Exceptions and rejections are deduplicated into issues with full stack traces, runtime context, and the breadcrumbs that led there.

  • 02

    Source-mapped stack traces

    Debug IDs match each build to its source maps automatically, so minified web frames and Hermes frames resolve back to your original files.

  • 03

    Native crash capture

    iOS signals, NSException, Swift traps, Android JVM crashes, ANRs, and C++ aborts — queued offline and sent on the next launch.

  • 04

    Traces, Web Vitals & analytics

    Network calls, performance traces, LCP / CLS / INP, and cookie-free analytics from the same SDK. No cookie banner required.

  • 05

    MCP server for your AI tools

    Connect Claude or Cursor and ask what broke since Friday — 14 tools to find, explain, assign, and comment on production issues.

  • 06

    Release health

    Releases are detected from your CI commit SHA and tracked with their events, issues, and first/last-seen.

  • 07

    Triage, not just a feed

    Resolve, ignore, or reopen, assign to a teammate, discuss in threaded comments — or push to Linear in one click.

  • Built and operated by the same team that ships your outsourced engagements.
Harfi logo

Harfi

The AI does exactly what you said — not what you meant.

A prompt-engineering game: forty trials, each a spec the model's answer has to hit — exact bullet counts, hard word caps, forbidden letters, valid JSON with no fence. You write the instruction, a parser scores the result, and you get it there in as few strokes as you can.

Free — Set 1, all 8 trials · Plus $29 once — all 40 trials & 160 medals, no subscription

  • 01

    Scored by a parser, not an opinion

    Code counts the bullets, counts the words, parses the JSON, and scans for banned letters. No model judges the answer, so the same attempt always scores the same.

  • 02

    Told exactly which rule broke

    Every miss names the rule that failed and by how much — one bullet too many, a word over the cap, a banned letter — then you go again.

  • 03

    Forty trials, five sets

    Shape, Schema, Forbidden, Transform, and Control — from word caps and clean JSON to lipograms and messy input with one exact right answer.

  • 04

    Golf scoring

    You never write the answer, only the prompt — 120 words at most. Fewer strokes and shorter prompts score better.

  • 05

    160 medals to chase

    Four per trial: clear it, beat par, ace it on the first stroke, or clear it with a prompt under 25 words.

  • 06

    Does it hold on another model?

    The Control set runs the same prompt against a second model, testing whether a prompt that works on one survives on another.

  • 07

    A deliberately fallible opponent

    Most trials run on Claude Haiku 4.5 — on purpose. A model that followed every instruction perfectly would make the trials trivial.

  • Built and operated by the same team that ships your outsourced engagements.
More from Outworx

Eleven more products, same foundation.

Every product shares one multi-tenant core — auth, organisations and roles, billing, usage limits, custom domains — which is how one team runs this many of them.

Developer tooling

The packages behind the products.

SDKs, CLIs, and CI integrations — published on npm and GitHub, most with zero runtime dependencies.

Free 15-min onboarding call

Tell us what you're building. We'll bring the team.

Outsourced engineering, design, consulting — or kicking the tires on one of our products. Book a 15-minute call, or send us a message and we reply within one business day.

No credit card · No upfront commitment · NDA on request

Chat with Outworx on WhatsApp