Graph Pipelines: LLM-assisted Qualitative Analysis

Evidano is an AI-powered qualitative data analysis platform that ingests transcripts, runs AI-assisted coding and cross-segment analyses, and exports reports. Fast-moving open-source tools matter for teams doing LLM-assisted qualitative analysis. On Jun 8, 2026 the soak project released version 0.13, a DAG-based pipeline toolkit for chaining LLM calls and text-processing steps (see source: PyPI). If you are a UX researcher, policy analyst, or product team looking to scale coding, this post shows a reproducible workflow and where Evidano (www.evidano.com) plugs in to speed transcription, theme extraction, cross-segment comparison, and secure reporting.

Key Takeaways

Evidano and soak 0.13 can be combined to build repeatable DAG pipelines for LLM-assisted qualitative coding, calibration, and reporting within two weeks. The soak 0.13 release (Jun 8, 2026) provides DAG-based orchestration for cleaning, chunking, LLM labeling, and aggregation, while Evidano handles ingestion, transcription, calibration, and visualization. Licensing and governance matter: soak is AGPL-3.0-or-later, and sensitive outputs should be treated as research signals.

soak 0.13 exposes DAG pipelines to chain text transforms and LLM calls for reproducible coding workflows (released Jun 8, 2026).
Evidano ingests transcripts, applies custom-dictionary transcription and PII redaction, runs calibration and cross-segment analyses, and imports soak outputs for visualization and reporting.
AGPL-3.0-or-later license requires attention for closed-source integrations, and model labels should be validated with human coders before acting on results.

Findings Snapshot (summary)

This table summarizes key package metadata and links for soak 0.13.

Findings Snapshot

Metric	Value	Source	Note
Project	soak (soaking) v0.13	PyPI	Released Jun 8, 2026
Packaging	wheel + sdist (45.0 MB / 44.4 MB)	PyPI files	Python >=3.12
License	AGPL-3.0-or-later	PyPI	Copyleft, consider compatibility for production
Primary purpose	DAG-based LLM-assisted qualitative text analysis	PyPI	Pipelines + tooling to chain transforms and LLM calls
Docs / Examples	Sample outputs and usage patterns	soak docs	Sample outputs and usage patterns

What soak 0.13 actually does

soak 0.13 exposes directed acyclic graph (DAG) pipelines you can compose to run LLM prompts, text transforms, and downstream aggregations in a repeatable way. Instead of ad-hoc scripts that call a model once per file, soak helps define multi-step flows: cleaning → chunking → LLM labeling → aggregation → export.

Designed for programmatic, reproducible LLM chains over text corpora.
Ships examples and a docs site with sample outputs (see project docs: soak docs).
Provides extras for calibration, local models, and PII scrubbers (optional).

So what for research teams?

For UX & Product Researchers

UX and product researchers can codify a coding pipeline (utterance split → sentiment/intent label → theme mapping) and re-run it as new transcripts arrive to improve reproducibility versus manual passes.

DAG pipelines make it easier to version steps (prompt text, chunk size, model) so experiments are auditable.

For Policy / Health Analysts

Policy and health analysts can chain scrubbers and calibration steps before models see data, which is useful when working with sensitive interviews.

License (AGPL) requires attention if you plan to integrate into closed-source systems.

For Data & Ops Teams

Data and ops teams can map a DAG approach cleanly to CI: test small sample outputs, run calibration jobs, and deploy pipelines that process large corpora reliably.

Built distributions and file hashes are provided on PyPI, which is useful for secure installs.

Turn this into an operational workflow (where Evidano helps)

Ingest & pre-process at scale

Evidano ingests and pre-processes transcripts, reports, and survey sheets and drops outputs into a soak pipeline as pre-cleaned text.

Problem: varied inputs, recorded interviews, PDFs, spreadsheets. Evidano: Native ingestion of transcripts, reports, and survey sheets, transcription with custom dictionaries and PII redaction, and export of cleaned text into soak.

Consistent LLM labeling & calibration

Evidano imports codebooks and runs AI-assisted coding with hierarchical codes and calibration tools to compare human versus model labels before committing changes to pipelines.

Problem: prompt drift and inconsistent codes across batches. Evidano: Import codebooks and run AI-assisted coding with hierarchical codes → subcodes, plus calibration tools to compare human vs. model labels.

Cross-segment analysis & visualizations

Evidano runs thematic, frequency, and cross-segment analyses, plus co-occurrence networks and word clouds that map directly to soak’s aggregated outputs for rapid reporting.

Problem: synthesizing themes across cohorts (region, persona, time). Evidano: Thematic, frequency, and cross-segment analyses, plus co-occurrence networks and word clouds that map directly to soak outputs.

Security & reproducibility

Evidano provides end-to-end encryption and a policy of not using customer data to train third-party models, which complements soak’s local-ai and scrub extras for secure pipelines.

Problem: data sensitivity and model training concerns. Evidano: End-to-end encryption and policy of not using customer data to train third-party models, complementing soak features.

Two-week pilot: from raw files to stakeholder brief

A compact two-week pilot converts raw files to a stakeholder brief using soak pipelines and Evidano automation.

This pilot validates soak pipelines and measures uplift from Evidano automation.

Day 0–2: Gather 50–200 transcripts or open survey comments; register in Evidano and run transcripts through Evidano’s custom-dictionary transcription plus PII redaction.
Day 3–5: Export cleaned text into a soak DAG: chunk → prompt labeling (themes, sentiment) → aggregate.
Day 6–9: Use Evidano to import soak outputs, run cross-segment frequency and co-occurrence analysis, and generate visualizations.
Day 10–12: Rapid QA: sample-check coded outputs, run calibration, and adjust prompts or codebook in soak.
Day 13–14: Produce a 2-page stakeholder brief with top 5 themes, segment differences, and recommended experiments, exportable from Evidano.

FAQ: LLM-assisted qualitative analysis

What is soak 0.13 and what does it do?

soak 0.13 is a DAG-based toolkit that composes text transforms and LLM prompts into repeatable pipelines for labeling and aggregating qualitative text. The toolkit helps replace ad-hoc scripts with multi-step flows: cleaning, chunking, LLM labeling, aggregation, and export, and ships examples and a docs site.

How does Evidano integrate with soak pipelines?

Evidano ingests raw transcripts, applies transcription and PII redaction, and exports cleaned text for use in soak DAGs; Evidano also imports soak outputs for calibration, analysis, and visualization. Evidano supports calibration tools and codebook imports to compare human and model labels before committing changes.

What security and licensing concerns should I consider?

You should review the AGPL-3.0-or-later license for soak before integrating it into closed-source systems, and treat model outputs from sensitive data as research signals requiring governance. Use Evidano’s PII redaction and encryption features to reduce exposure, and validate labels with human coders before taking action.

How quickly can I run a pilot combining soak and Evidano?

A compact two-week pilot can take raw files to a stakeholder brief by combining soak pipelines with Evidano’s ingestion, transcription, calibration, and visualization features. The pilot steps include gathering 50–200 transcripts, running transcription and PII redaction, exporting to a soak DAG, importing outputs into Evidano for analysis, and producing a 2-page brief.

Topics

LLM-assisted qualitative analysis
soak 0.13
DAG pipelines
qualitative coding
Evidano

Keep reading

Browse all articles

Graph Pipelines: LLM-assisted Qualitative Analysis

Key Takeaways

Findings Snapshot (summary)

Findings Snapshot

What soak 0.13 actually does

So what for research teams?

For UX & Product Researchers

For Policy / Health Analysts

For Data & Ops Teams

Turn this into an operational workflow (where Evidano helps)

Ingest & pre-process at scale

Consistent LLM labeling & calibration

Cross-segment analysis & visualizations

Security & reproducibility

Two-week pilot: from raw files to stakeholder brief

Limitations & governance notes

FAQ: LLM-assisted qualitative analysis

What is soak 0.13 and what does it do?

How does Evidano integrate with soak pipelines?

What security and licensing concerns should I consider?

How quickly can I run a pilot combining soak and Evidano?

Wrapping up: your next moves

Topics

Keep reading

Company

About

Preview Evidano

Account

Newsletter