Hi, I'm Drew. Here's the work.

▼

Systems builder · Automation architect · Privacy-first engineer

Here's the work.
BLAP.

I'm Drew. I build complete systems, from an empty server all the way to production: data pipelines, compliance engines, AI agents and the apps people hold in their hands. Five of them are below. Each one started with a real problem nobody had solved well.

See the flagship Hire me

The Work

Flagship · iOS & Android

AnonyMaps

Find an anonymous recovery meeting near you — without anyone finding you.

Problem #1 — the map didn't exist

Recovery meetings are listed by hundreds of independent local and county groups, each with its own website and its own format. Someone who needs a meeting right now has to find the right list, read the addresses, and drop them into a map one by one just to see what's close.

Problem #2 — the people were the product

Many of those local sites rely on third-party services that collect visitor data. A categorized list of IP addresses belonging to people in recovery is exactly the kind of data that gets resold. The people most in need of anonymity were the least protected.

The answer

  • One map, every source. An ingestion pipeline unifies 150+ independent AA and NA meeting feeds — different formats, different reliability — into a single searchable, distance-sorted map, refreshed on a schedule.
  • IP stripping. Requests are stripped of the user's IP before anything leaves the device's path.
  • Your search stays on your phone. No accounts. Location and activity are processed locally; the server holds public meeting locations and nothing about the person looking.
  • Safety over completeness. Meetings are deliberately withheld in countries where admitting to addiction can carry life imprisonment or death — a listing there could cost someone everything.
  • Built to survive bad data. Feeds that return HTML, broken JSON or nothing are skipped and retried; missing coordinates are geocoded; re-imports never duplicate. One failing source never stops the rest.
  • 150+meeting feeds unified
  • 18pipeline workflows
  • 1,100+successful import runs
  • 0user accounts, by design
Download on the App Store Get it on Google Play

n8n · JavaScript · Supabase / Postgres · Cloudflare Workers, Tunnel & Access · Geocoding · iOS / Android

Live · Multi-tenant SaaS

LeadFunl

A compliant lead marketplace that connects retirees with financial advisors.

Advisors need prospects who actually qualify for a Roth conversion, and generic leads waste their money. I built the whole funnel: ads lead to a survey and a calculator that writes a personalized PDF report. Leads are qualified on real criteria and on consent. AI voice agents follow up and book appointments. A weighted engine then routes each lead to subscribing advisors, straight into their own CRM.

  • Compliance as code: a consent gate with a full audit trail, STOP/revocation handling, DNC holds and calling-window enforcement.
  • Multi-tenant by design: row-level security, per-tenant isolation, and offboarding that cuts all access in one action.
  • Self-healing ops: errors become a work queue. An AI diagnoses each one and alerts arrive in real time.
  • Full audit: every line of workflow code was read (10,111 of 10,111), and the critical consent defects were fixed.
  • Mobile app: a React Native app that records advisor calls and turns them into structured notes.
  • 77workflows
  • 50+database tables
  • 7edge services

n8n · Supabase (RLS) · Cloudflare Workers · GoHighLevel · Stripe · Meta Conversions API · Vapi / Bland AI voice · Expo / React Native · Metabase

Coming Soon

What's on the bench right now. Built the same way as everything above.

Built · Pre-launch

TotalAutomation.ai

A white-label social media department that runs itself, with a human holding the brakes.

A business fills in its brand style guide once. From then on the platform researches trends, writes posts, generates images and video, and publishes. It also responds to comments, measures what works and re-optimizes. Every piece of content passes four human approval gates before it goes live.

  • Zero-trust secrets: a custom token broker at the edge. No credential lives in a workflow, a database or a credential store.
  • Four-gate review: media, copy, preview and deployment, with idempotent publishing so nothing double-posts.
  • Cost-gated rollout: every workflow that spends money is switched on deliberately, one at a time.
  • Respond, never initiate: engagement is reply-only, a hard rule built into the system.
  • 31workflows
  • 410automation nodes
  • 4approval gates

n8n · Claude · Perplexity · FLUX · Meta Graph API · Supabase · Cloudflare Workers KV · React portal

Proof of concept · Oct 2026

AI Voice Intake & Attorney Routing

An AI voice agent replaces cold call centers for accident victims seeking representation.

A caller talks to an AI intake agent that opens with clear disclosures. Their state is captured first, and that reserves the next eligible law firm from a fair-rotation pool. The firm's own intake script then runs, so the caller never answers the same questions twice. A qualified caller is committed to that firm. Anyone else is released back to the pool.

  • Reserve before you ask: a 15-minute fair-rotation hold. Newer firms go first, then whoever was served least recently, and busy firms are skipped.
  • Evidence by design: every signed agreement is stored as the exact rendered text with a cryptographic hash, a timestamp and the call record.
  • Compliance up front: plain disclosures on every page and every call. No testimonials, no case-result promises, no tracking pixel.
  • Legal questions go to counsel: open regulatory questions are written down and handed to a lawyer, not guessed at.

Vapi voice AI · n8n · Supabase · Cloudflare Workers & Access · Python deploy & test harness

New · Coming to the App Store & Google Play

AI Transcription

Intelligent call notes for financial professionals. Every client conversation becomes a record you can stand behind.

Regulated professionals have to prove what was said, and when. Most of them are still writing notes by hand after the call, or using transcription tools that treat a conversation as a wall of text. I built this one for the retention and audit standards that advisors, insurance agents and lenders actually face.

Every call, live or recorded, turns into structured, searchable, downloadable notes, organized by client, date and topic in a back-office portal.

It fits any regulated service. The core isn't tied to finance. Change the summary template and the retention rules, and it serves any profession that needs a clean record of what was said.

  • Dual-pass AI: two processing passes on every call, for accuracy that holds up when someone reads it later.
  • Announced on every call: a recording disclosure plays automatically at the start of every call, following call-recording consent standards and practices.
  • Notes built for compliance: structured summaries made for documentation and audit trails, not just readability.
  • Searchable back office: every conversation is organized by client, date and topic, and can be found in seconds.
  • Retention by design: records are kept for regulatory retention, and transcripts download ready to file.
  • Subscription-gated access: when a subscription ends, portal access ends automatically, with no manual offboarding.
  • 2AI passes per call
  • 3+regulated industries
  • 2app stores

Proprietary · Browser extension

Measure Twice

Makes any AI slow down and do its homework before it commits to a decision.

AI builders are fast and sure of themselves, and that is the problem. One confident guess about a version, an API or a setting can quietly cost weeks of rework. Measure Twice sits between you and whichever AI you work with, and it won't let a decision that matters be made on a hunch.

  • Works with any AI: it runs in the browser you already use, alongside the assistant you already use.
  • Research matched to the stakes: the bigger the decision, the deeper the AI has to dig before it's allowed to be confident.
  • Every corner, not just the obvious one: the AI considers the whole project before it touches one piece of it.
  • Months become days: in my own builds, work that used to drift for months now lands in days, and sometimes in hours.

The method is proprietary and kept confidential. What it produces is everything on this page.

Browser extension

Tattletale

Keeps a card breach from becoming a loss, and tells you exactly who leaked it.

When a store gets breached, you usually find out from the fraud. Then you cancel the card, update it everywhere, and never learn which business let it go. Tattletale turns that around.

  • Traces the leak: a proprietary algorithm uses tracking technology to trace which vendor is responsible for your stolen information, helping you assign liability.

How I Think

Tools change every quarter. These rules don't. They came from real failures, written down so they never happen twice.

01

Protect the most vulnerable user first

Design for the person with the most to lose. If the data could hurt them, don't collect it. Then there is nothing to leak.

02

Read before you write

Current docs, then the source, then the live system. Confident guessing is the most expensive habit in engineering.

03

Prove the instrument before you trust a zero

"No errors found" from a check that's never caught an error means nothing. Test it on a known positive first.

04

Compliance is a feature, not a footnote

Consent, revocation and audit trails are built into the pipeline. If a rule isn't enforced in code, it isn't enforced.

05

Isolate the blast radius

Every business gets its own automation instance, its own data and its own front door. One failure stays one failure.

06

Humans hold the brakes

AI does the volume. People approve what matters. Anything that skips, retries or suppresses says so out loud.

The Toolbelt

Why "BLAP"?

On the superhero TV shows of the '60s and '70s, every punch landed with a word splashed across the screen: POW. BAM. THUD. ZOK. The strangest of them all was BLAP.

That's what this site is for. No long windup. Here's the work, it lands, and you feel it. BLAP.

Let's Build Something

I'm open to roles in automation and systems engineering, and to privacy-first product work, especially where the problem is messy and the people it serves deserve better.

Find me on Wellfound