Melaya — Build AI agents for any job. Agentic platform for research, ops, outreach, reporting — and the only one where agents can actually trade.

// USE CASE · RESEARCH

Run market and macro research thatcites every source it stands on.

Most research desks lose half their week to chart pulling, filing skimming, and paper hunting before a single thesis gets written. Melaya gives you a four persona crew that runs the market structure read, the macro frame, the literature scan, and the feature proposal on a schedule. Every claim cites a FRED series, a 10 K paragraph, or an arXiv id, and an analyst signs off before the brief reaches the PM.

See the pipelines ↓
01
// What breaks today

The status quo costs 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.

  1. 01

    Analysts burn 12 hours a week pulling FRED charts, scraping 10 Ks, and copy pasting earnings transcripts before they write a single paragraph of thesis.

  2. 02

    One uncited number in a published note triggers a compliance review and a Monday morning rewrite, while the PM has already moved on to the next name.

  3. 03

    Last quarter's deep dive lives in someone's Notion, so the same name gets researched from scratch three times a year and nobody remembers the contradicting paper.

02
// Pipelines you can build

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.

P01

Run a single name deep dive

MarketAnalyst pulls EDGAR filings and insider Form 4, DataScientist proposes a feature on the read, LiteratureSpecialist scores the supporting papers. The HITL gate holds the brief until the analyst signs off on the regime call.

P02

Publish the weekly macro briefing

MacroEconomist queries FRED for rates, DXY, and inflation prints, then writes the regime call with the next 30 days of releases. Static context holds the house view so position sizing guidance stays consistent across weeks.

P03

Stand up an earnings reaction note

On 8 K hit, MarketAnalyst pulls the filing and the transcript, DataScientist diffs guidance against last quarter, the rag_retrieve tool fetches the prior note for tone match. Analyst approves before distribution.

P04

Run a quarterly literature survey

LiteratureSpecialist sweeps arXiv q fin and OpenAlex for the last 90 days, scores replication quality, and lists one contradicting paper per thesis. Cross run memory carries the read across quarters so duplicates drop out.

P05

Stress test a working thesis

DataScientist proposes a falsifiable feature with a statistical test, LiteratureSpecialist surfaces papers that argue against it, MarketAnalyst checks the on chain or funding rate data. The HITL gate forces sign off on the kill criteria before sizing.

P06

Extract structured data from filings

MarketAnalyst runs the EDGAR full text search across a coverage list, DataScientist normalizes the hits into a feature table with cited paragraph ids. Replay every cell back to its 10 K paragraph for the next compliance review.

03
// The crew

Research & analyst 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.

Market Analyst

MarketAnalyst

Builds the market structure brief with funding rates, open interest, on-chain flows, and a stated bullish, bearish, or neutral regime call with named risks.

Macro Economist

MacroEconomist

Sets the macro frame with Fed reaction function, DXY trend, global liquidity cycle, and a 30 day calendar of events that move position sizing.

Literature Specialist

LiteratureSpecialist

Reads recent quantitative finance papers, scores replication quality, and always surfaces one contradicting paper so the thesis is tested before it ships.

Data Scientist

DataScientist

Proposes new features with a formula, data source, decay half life, pseudo code, and a statistical test design that an engineer can implement the same day.

04
// Scoped tools

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.

shared/tools/core/

General purpose search, fetch, and local file reads for all four personas. Read only by default, so no persona can write outside its scoped bundle.

web_searchweb_fetchhttp_requestgrep_searchfile_read
shared/tools/sec_edgar_tools/

MarketAnalyst and DataScientist pull 10 K, 10 Q, 8 K, and Form 4 data straight from EDGAR. Read only surface, so no write gate is needed.

edgar_ticker_to_cikedgar_recent_filingsedgar_company_factsedgar_full_text_searchedgar_insider_form4
shared/tools/fred_tools/

MacroEconomist queries FRED for rates, inflation prints, and release calendars with the series id cited in every chart. Read only, no HITL needed.

fred_series_observationsfred_series_searchfred_series_infofred_release_dates
shared/tools/arxiv_tools/

LiteratureSpecialist scans recent q fin papers with replication scoring. Pairs with openalex_tools and semantic_scholar_tools for cross venue coverage.

arxiv_searcharxiv_get_paperarxiv_recentarxiv_by_author
shared/tools/openalex_tools/

Tracks citation graphs and follow on work so LiteratureSpecialist can stress test a thesis against later refutations. Read only.

openalex_search_worksopenalex_get_workopenalex_work_citationsopenalex_author_works
shared/tools/yahoo_finance_tools/

Lightweight price, options, and chart pulls for MarketAnalyst when a paid feed is not authorized. Read only.

yf_quoteyf_chartyf_summaryyf_options
shared/tools/knowledge/

Builds the per workflow vector store from prior notes, won thesis memos, and house coverage. Powers rag_retrieve and cross run memory. Indexing is HITL gated when new corpora are added.

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

includeContext

Per-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_retrieve

A 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_memory

Pipeline-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

Questions we get every week.

Will the research crew publish reports on its own?

No. Every report stops at a HITL gate so the analyst signs off on the regime call, the cited papers, and the feature proposals before anything goes to PMs or risk. Lift the gate per template once the workflow has earned the trust.

Can the agents reason over our internal research library?

Three ways. Static context attaches your house view, coverage list, and risk limits to specific personas on every run. The rag_retrieve tool lets MarketAnalyst and LiteratureSpecialist pull from filings, transcripts, and prior notes on demand. Cross-run memory carries last week's thesis into this week's update so follow ups do not start from zero.

Is this an AlphaSense alternative or a Sentieo alternative?

It sits next to them. AlphaSense and Sentieo are search and excerpt surfaces over filings and transcripts. Melaya is the reasoning layer that runs MarketAnalyst, MacroEconomist, and LiteratureSpecialist on a schedule and writes the brief, with your sources cited inline.

How does Melaya compare to Hebbia or BamSEC for filings work?

Hebbia and BamSEC focus on document Q and A over SEC filings. Melaya plugs SEC EDGAR in as a scoped tool that any persona can call, then composes a multi step workflow around it with the 3 layer knowledge model so the same answer is reproducible next quarter.

How do we keep the briefs from sounding like AI?

Drafts cite specific paragraphs from filings, FRED series IDs, and arXiv identifiers. The LiteratureSpecialist is required to surface one contradicting paper per thesis, so the tone reads as a working analyst note instead of a confident summary.

Which models can we run the research crew on?

Any. Claude on LiteratureSpecialist where long context and reasoning quality matter, GPT on DataScientist for code drafting, a local Ollama on MarketAnalyst when the on chain dataset cannot leave your network. Each persona picks its own.

How fast can a research team get the first pipeline running?

With FRED, SEC EDGAR, and arXiv tools enabled, the macro briefing workflow is a 4 node canvas: pull series, summarize, cite, approve. Most teams ship it in a working session and have the first reviewed brief in their inbox the same day.

Can I audit exactly what the agent did and why?

Every run logs every tool call, every retrieved chunk, every model invocation, and every approval. Replay any run to see which FRED series, which 10 K paragraph, and which arXiv paper drove the conclusion. The audit log is the methodology log.

Build research & analyst 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
Join the community