Stage 1: tool setup
Cursor, Claude Code, Codex, GitHub Desktop, repo creation, commit/push workflow, and error notes.
← Back to selected work
Research project · supporting proof
Evidence-led GTM research for a B2B SaaS cold outreach pipeline.
This supporting project documents how I selected sources, rejected weak evidence, mapped advice to a pipeline, organized the work in GitHub, and manually reviewed AI-assisted research.
Verifiable artifact
The public repository documents the research process.
The repo documents tool setup, research decisions, saved sources, transcript collection, and why each practitioner belongs in the pipeline.
Research record
The repository contains source notes, inclusion and rejection decisions, pipeline mapping, and review details.
Stage 2: research project
Cold outreach pipeline for B2B SaaS, organized by source quality and pipeline coverage.
Manual verification
LinkedIn posts were checked against originals because automated extraction can drop or change words.
AI use disclosure
Claude helped fetch/draft, but source choices, exclusions, and final notes were reviewed by Amrutha.
SignalThe work shows evidence-led judgment: not collecting famous names, but building a usable map of what each source contributes.
Pipeline map
The expert set was selected to avoid repetition.
Each source owns a distinct layer of the same outbound process.
Account selection
Mark Colgan: account selection, channels, AI-assisted research, data quality.
Qualified list
Jan Brochwicz: ICP matrix, segmentation, disqualifiers, persona mapping.
Signal context
Anthony Natoli: turning a funding, hiring, or launch signal into a real reason to reach out.
Cold email
Jason Bay and Josh Braun: outbound frameworks, neutral questions, useful follow-up.
Phone + social
Kevin Hopp and Laura Erdem: permission-based cold calls and LinkedIn/social selling tied to pipeline.
Quality + deliverability
Petr Kaliuzhny, Nick Abraham, and Jen Allen-Knuth: sequence QA, inbox placement, and discovery/qualification.
Research decisions
The strongest proof is the rejection logic.
The README explains what was included, what was cut, and why.
Recent over famous.
Old or weak sources were replaced even when the person was well known.
Coverage over count.
Each expert needed to add a new layer, not repeat the same advice.
Manual checking.
Saved LinkedIn text was checked against live posts because automation can be unreliable.
Traceable notes.
Files include source links, dates checked, and short annotations.
Practical domain fit.
Cold outreach was chosen because it overlaps with real B2B SaaS selling experience.
AI with ownership.
AI helped speed up research, but every expert and source was reviewed before inclusion.
What the project produced
A documented operating map, not a loose content summary.
This supporting project combines structured research, source judgment, manual review, and public GitHub documentation.
Evidence disciplineClear source links, saved posts/transcripts, dates, and annotations.
GTM understandingPipeline stages from account selection to qualification.
JudgmentExplicit inclusion and exclusion reasoning.
Public artifactA working GitHub repo reviewers can inspect.