P01
In-app extraction
Agents open the app, navigate to the data, and extract it the same way a person would: screen by screen, on a real device.
01
// What breaks today
Manual workflows cost more than the agent does.
Three pains every sales and BD team hits weekly. Each one is what your reps actually complain about, not what a feature page would call them.
- 01
The data you need lives inside mobile apps that have no API, and web scraping does not reach it.
02
// Pipelines you can build
Agent workflows: compose, approve, replay.
Every pipeline below is a shape you wire on the canvas using the crew and tools further down. Not a feature we ship for you, a pattern you configure.
P02
Scheduled, scoped runs
Scoped to the apps you allow, on a schedule you set, with every run logged and reviewable.
P03
From screens to datasets
Extracted screens become validated, structured records your pipeline can consume.
03
// The multi-agent crew
Data extraction crew
Real personas from the research_team crew. Each ships with a tuned system prompt and a default tool allowlist. Swap models per persona on the canvas.
Data Scientist
DataScientistTurns screen-by-screen extraction into clean, validated datasets.
Market Analyst
MarketAnalystDecides which in-app sources matter and what to pull from them.
Literature Specialist
LiteratureSpecialistDocuments each source and keeps extractions reproducible.
04
// Scoped tools
Tool allowlists: only the actions you grant.
Every tool below is a real shared tool from the Melaya bundle. Allowlist per agent; HITL-gate the writes; revoke any of them in one click.
01shared/tools/phone/
Navigates the app and reads the data from the screen on a real device.
phone_open_appphone_get_screen_treephone_scrollphone_screenshot
02shared/tools/data_utils/
Validates and profiles what came off the screens.
csv_lintdf_describejson_validate
03shared/tools/database/
Lands the results where your pipeline expects them.
sql_querysql_export_csv
04shared/tools/knowledge/
Builds a searchable record of every extraction run.
build_knowledge_from_textbuild_knowledge_from_file
05
// Three knowledge layers
The crew reads what you give it.
Every pipeline ships with three layers of knowledge access. Mix and match per agent on the canvas. No shared vector space with another tenant, no surprise reads, no opaque retrieval.
L1
Static context
includeContextPer-pipeline documents appended to specific agents' input on every run. The ICP brief, playbook, pricing sheet, or won-deal email corpus. Whatever needs to be there before the agent thinks. You pick which personas get which docs.
L2
RAG retrieval tool
rag_retrieveA scoped tool granted per-agent. When the agent decides it needs more depth, it queries the workflow's vector store on demand. Same knowledge base as Static context, accessed only when the model asks for it.
L3
Cross-run memory
pipeline_memoryPipeline-level state that carries from one run to the next. Yesterday's research is in scope for today's follow-up. The crew remembers what it already prospected, what got approved, what was sent. The audit log is the second-order knowledge base.
07
// FAQ
AI agent questions we get every week.
What about apps with no API?
That is the point. The agent reads the data from the app screen on a real device, so no API is required.
Is every run auditable?
Yes. Runs are scoped to the apps you allow, scheduled by you, and every step is logged and reviewable.
Why not just write a scraping script?
Scripts stop working when the data lives inside a mobile app with no API and no stable web endpoint. Melaya's Device Control operates a real Android phone, so agents read the same screens a person sees. The Data Scientist persona then validates the output with the data_utils bundle, and full run traces show what every step read.
Can n8n or Zapier extract data from a mobile app?
No. n8n, Zapier, and Make execute predefined trigger-action steps against APIs, so an app without an API stays out of reach. Melaya's agents plan across steps on a canvas of agents, tools, triggers, and approval gates, and the phone bundle reads app screens directly on a real device. Zapier and Make do win on connector breadth for simple API automations.
How do screen reads become a clean dataset?
Three tool bundles carry the pipeline. The phone bundle navigates the app and reads each screen on a real device, data_utils validates and profiles the extracted fields, and the database bundle lands results where your pipeline expects them. The knowledge bundle keeps a searchable record of every extraction run.
Can extraction runs be scheduled and scoped to specific apps?
Yes. Runs fire on the schedule you set, and each agent is scoped to an allowlist, so the phone bundle can open only the apps you approved. The Market Analyst persona decides which in-app sources matter, and every scheduled run produces a full run trace you can review.
How do data teams govern what extraction agents can access?
Governance in Melaya rests on scoped tool allowlists, human-in-the-loop approval on every write, and full run traces. Agents open only allowed apps, writes through the database bundle can pause for approval before landing, and deterministic-first evaluation runs rule-based checks on extracted data before any model-graded ones.
Are extraction runs reproducible?
Yes. The Literature Specialist persona documents each source and keeps extractions reproducible, and any run can be saved as a reusable template. Full run traces record each step with timestamps, so you can compare runs, catch layout drift inside an app, and rerun the same pipeline unchanged.
Does extraction run on a real phone or an emulator?
A real phone. Melaya's Device Control operates a real Android phone: agents open allowed apps, read the screen, and tap and type the way a person does. Anything that would publish or submit pauses for on-device approval, and a read-only extraction never posts on your behalf.
Can extracted data stay on our own infrastructure?
Yes. Melaya runs in the cloud or on your local runner, so sensitive extraction steps and the resulting datasets can stay on hardware you control. Per-step model routing across 49 providers lets you keep parsing on local models, and cost per accepted result is visible for every run.
Build data teams pipelines on Melaya.
Sandbox tier is free with no card. Join the waitlist and we will email you the moment a slot opens.
← Back to every use case
// Cookies
Melaya uses a small set of first-party cookies that are strictly necessary to authenticate you, maintain your session, and protect the platform from abuse. We do not use advertising cookies, cross-site trackers, or third-party analytics by default. The full cookie list is in our Privacy Policy.