Skip to content

AI agents and models for research

Four products that work alone or together, from a single analysis to a complete study.

Research Agents

Agents built for discovery

Agents for every stage of the research cycle, from open questions to submitted papers. Use one for a single task, or chain them.

Design

3 agents

  • Idea assessment

    Novelty scan and a go or no-go memo

  • Study design

    Design, variables and power analysis

  • Statistical analysis plan

    Estimands, models and sensitivity analyses

Evidence

2 agents

  • Systematic review

    Search strategy, screening and PRISMA flow

  • Gene annotation

    Gene lists turned into annotated tables

Analysis

4 agents

  • Discovery analysis

    Every pattern and anomaly, ranked as hypotheses

  • Formal methods

    Theorems, proofs and algorithms in LaTeX

  • Figures and tables

    Publication-grade, colorblind-safe figures

  • Multi-omics figures

    Volcano plots, heatmaps and pathway figures

Verification

2 agents

  • Citation audit

    Every claim matched to its source

  • Reproducibility check

    Results regenerated from code and data

Writing and review

9 agents

  • Manuscript drafting

    A first draft from your results

  • Title and abstract

    Titles and abstracts tuned to the journal

  • Language polishing

    Clear, natural academic prose

  • Pre-submission review

    A tough reviewer before the real one

  • Response to reviewers

    Point-by-point responses and revisions

  • Journal selection

    Ranked targets, predatory venues screened out

  • Cover letter

    A submission letter editors read

  • Grant writing

    Proposals checked against review criteria

  • Preprint and promotion

    Preprint posting and research summaries

Autonomous Runs

Start with a question, come back to findings

An agent plans the exploration, writes and runs the code, tests what it finds and checks the results on its own.

  • Plans the work and waits for your approval
  • Writes and runs its own code to test hypotheses
  • Returns findings, figures and a report
Autonomous runSample

Question

What predicts treatment response in cohort_v2?

  1. Plan, done

    Exploration plan drafted and approved

  2. Explore, done

    38 variables profiled, 3 anomalies found

  3. Hypothesize, done

    12 patterns ranked as hypotheses

  4. Test, done

    Top 3 tested, 2 supported by the data

  5. Write, done

    Findings and figures drafted

  6. Verify, needs review

    Every number rerun; 1 finding for your review

6 of 6 steps complete
Findings and report ready

Model API

MindCODE models through one key

An OpenAI-compatible API for chat and embeddings. Point the OpenAI SDK at api.mindcode.cc and use your MindCODE key.

MindCODE models
ModelRuns
mc-code-1Flagship model for research code and analysisHosted
mc-code-1-liteLower cost for everyday tasksHosted
mc-code-1-fastLow latency, runs on local hardwareHosted or private
mc-embed-1Embeddings for search over papers and dataHosted or private

Priced per million tokens. Volume pricing is negotiable.

Get an API keySee pricing
Chat completion
from openai import OpenAI

client = OpenAI(
    base_url="https://api.mindcode.cc/v1",
    api_key="sk-mc-...",
)

reply = client.chat.completions.create(
    model="mc-code-1",
    messages=[
        {"role": "user", "content": "Summarize cohort_v2.csv"},
    ],
)
print(reply.choices[0].message.content)

Verify

Every claim, every number

Checks each citation against its source and reruns each result from the code and data.

  • Citations matched to their sources
  • Results regenerated from code and data
  • Every finding located in your files

Verification report

manuscript_v3.pdf and analysis repository

Sample

  • Citations, passed

    48 of 48 in-text citations match the reference list

  • Claim support, needs review

    2 claims cite papers whose abstracts do not support them (sections 2.3, 4.1)

  • Results, passed

    Table 2 regenerated from the deposited data; every value matches

  • Data, needs review

    Figure 3 legend reports n = 212; the dataset has 209 participants

  • Environment, passed

    Dependencies pinned; the analysis runs end to end with fixed seeds

2 findings need your decisionEvery finding links to its evidence

Bring us your next study

Start with a pilot on a real project: a manuscript, a dataset or a model.

Request accessView products