01 · Data + AI
Data monitoring & alerting
Watch any source — public registers, APIs, your own database — detect what changed, and deliver scored alerts, by digest or in realtime. The engine behind our own products.
tiny_prism_llc
· registered in Sheridan, WY
Data, agents, and
the systems that
ship them.
A one-person studio working four lines at once: Data + AI monitoring and alerting, AI agents that do real work, applications and systems built end to end — and our own niche products, live in production as the proof.
Our four lines of work
Domains we work in · client names withheld by default
Federal contracts / sam.gov
Securities / sec edgar
Trademarks / uspto
Marketplaces / reporting
Health-tech / device data
About · 01
A small studio
A small studio
with a long memory.
Tiny Prism LLC is a one-person studio registered in Sheridan, Wyoming. Two decades building for the web, the recent years spent on data, LLMs, and agents. Four lines of work share one pair of hands — which is exactly why they reinforce each other rather than compete.
The principal
Founder · Data, AI & systems engineer
Building for the web since 2005, fully remote with US and EU teams since 2011 — now split between Data + AI, agents, whole applications, and our own products.
The four lines are less separate than they look. Monitoring taught us to make data trustworthy; agents are what you build once the data is worth acting on; applications and systems are where both have to actually live; and our own products are where we pay the cost of every shortcut ourselves. Whichever one you hire, you get the other three's scar tissue.
What you actually get: a working slice within a week — a first scored record, an agent run you can read end to end, or a deployed page — then steady, demoed progress instead of status decks. A runbook and an architecture note when we're done, so the next person to touch it can.
If something is outside our focus, we'll tell you within a day and point you toward someone better suited. We'd rather pass than half-ship.
21yrs
Building for the web
4
Live in-house products
15yrs
Remote & async
40+
Projects delivered
What we do · 02
Four lines,
Four lines,
one set of hands.
Data + AI, AI agents, applications and systems, and niche products — grouped below by which of the four it belongs to. Anything outside them, we'll tell you upfront and point you to someone better suited.
01 · Data + AI
AI analysis & summarization
LLMs pointed at your data: plain-English summaries, relevance scoring, structured extraction, and evals — grounded in the source and always linked back to it.
01 · Data + AI
Data pipelines & integration
Ingest, dedupe, entity-match, and normalize from feeds, APIs, and databases — the unglamorous plumbing that makes the data, and the alerts on top of it, trustworthy.
02 · AI agents
AI agents & automation
Agents that actually do the work — real tools, retries, guardrails, and a trace you can read after the fact. Built so a wrong answer is visible and cheap, not silent.
03 · Apps & systems
Applications, end to end
A whole application, not a prototype — schema, screens, auth, billing, background jobs, deploys. Bring the idea and the domain; you get something on staging in week one.
03 · Apps & systems
Systems setup & integration
The systems you already run, made to work together — infrastructure and deploys, third-party integrations, admin and reporting layers, and the automation between them.
04 · Niche products
Niche products, end to end
A product for one narrow market, built the way we build our own — watchlists or workflows, accounts, billing, and the pages that earn the traffic. Shipped, and yours.
Hourly
Consulting, by the hour
Senior pair-hours across any of the four — source selection, scoring logic, prompt and eval design, agent architecture, or a second opinion before you commit a quarter.
—
Not sure which?
Send a brief paragraph. We'll tell you within a day whether it's a fit, and what it might cost.
Send a brief
Capabilities · 03
The four,
The four,
up close.
Each line stands on its own — and each one is better because the other three exist. Here's what sits behind them.
Monitoring & alerting
Continuous watch over public records and private feeds — ingested, deduped, entity-matched, and scored by an LLM — then delivered as a digest or a realtime alert the moment something changes, always linked back to the source record.
ingestion · entity-matching · scoring · realtime alertsAgents that do real work
Not a chat box. Agents with actual tools, retry and fallback paths, budget and permission guardrails, and an execution trace you can read line by line when something goes wrong — because eventually it will.
tool use · guardrails · traces · evalsApplications and plumbing
Whole applications — schema, screens, auth, billing, background jobs, deploys — and the less glamorous half: standing up the infrastructure and making the systems you already run talk to each other.
web apps · auth & billing · infra · integrationsProducts we own
Four live products where we are the client: we pick the niche, carry the hosting bill, answer the support email, and pay for every shortcut ourselves. It's the fastest feedback loop we have, and clients inherit what it teaches.
contractpulse · filingradar · trademarkwatch · hexagram castAlso in the kit — public-record ingestion · fuzzy & phonetic matching · relevance scoring · source-linking · prompt & agent evals · Rails & Postgres · background jobs · CI & deploys · CSV / API export
How it goes · 04
Four steps,
Four steps,
roughly six weeks.
Every project runs on the same cadence. You'll always know what's next, what we're blocked on, and what's shipping this Friday.
STEP / 01
Brief & read-through
An hour on a call, then a day in your repo and your docs. You get back a one-page plan with a fixed quote and a start date.
~ 2 days · free
STEP / 02
Something real inside a week
A working slice against your own data or domain — a first scored record, an agent run you can read end to end, or a deployed page. A visible stake in the ground that sets the cadence for everything after.
Week 1
STEP / 03
Build, Friday demos
Every Friday: a 15-minute recorded walkthrough of what shipped, what's in review, and what's next. No status decks, no theatre.
Weeks 2 – 5
STEP / 04
Handoff & 30 days of support
Runbook, test suite, and an architecture note you can hand to your next hire. Thirty days of follow-up included, whether you need it or not.
Week 6 +
Case study · 05
The work,
The work,
one project at a time.
A recent engagement — typical scope, typical pace. The list below spans all four lines: data and reporting, AI work, and full applications. Client names are withheld by default; references and fuller write-ups are available on request.
Case study — lessons marketplace · name withheld
Two-plus years owning the data behind a live marketplace, end to end.
A consumer marketplace for scheduled, payable lessons — where every booking, payout, and refund had to reconcile. One engineer owned the data and reporting across releases and integrations: keeping numbers trustworthy, surfacing what changed, and making sure the money and the records always agreed.
Ask for the write-up2+ yrs
Sole ownership, data & code
3
Systems: scheduling · payments · reporting
1
Senior engineer, end to end
Selected engagements · 2018 – 2026
client names withheld
06
AI writing assistant on the OpenAI API
LLM drafting and editing flows — prompts, retries, structured output, and evals.
ai · llm · summarization 2023 – 24 05Lessons marketplace — data & reporting
Featured above. Multi-year ownership: payouts, refunds, reconciliation, reporting.
data · reporting · reconciliation 2022 – 24 04Content publishing platform
A content pipeline publishing into Google Docs, Medium, HubSpot, and WordPress.
pipelines · apis · automation 2021 – 22 03Retail marketplace — catalog & admin
A multi-retailer marketplace — catalog data, storefront, and admin reporting.
marketplace · catalog · admin 2021 – 23 02Health-device client portal
Patients and clinicians reviewing device data — charts, configuration, exports.
device data · charts · exports 2019 – 20 01Crypto wallet — internal analytics
Dashboards for wallet usage statistics and operational reporting.
analytics · dashboards · sql 2018 – 19
Line 04 · In-house · 06
Products we
Products we
build for ourselves.
The fourth line, and the clearest statement of what the other three can do. Three of these run the same Data + AI loop we build for clients: watch an official public record, score what actually matters, send a plain-English alert linked to its source. The fourth takes the same discipline — reproducible computation, cited sources, nothing hidden — somewhere else entirely. All four are live, in production, and run at our own expense.
ContractPulseFederal contract intel
Find the right federal opportunity before your competitors do.
Monitors SAM.gov and USAspending, summarizes every new opportunity in plain English, and sends only the alerts that match your NAICS, agencies, and competitors.
4
Federal data sources
Daily
Digest + urgent alerts
SAM.gov · USAspending
Visit site
FilingRadarSEC filing alerts
The SEC filings that actually move stocks — in plain English.
Watches every insider buy, activist stake, and 8-K on SEC EDGAR, scores the ones that matter, and sends a plain-English alert the moment they're filed.
6,000+
Filings read a day
<60s
Wire to inbox
EDGAR · Form 4 · 13D/G · 8-K
Visit site
TrademarkWatchUSPTO brand monitor
Know the moment a similar trademark appears.
Monitors USPTO records and alerts you in plain English when a new application may conflict with your brand — risk scored, with the official record one click away.
650K+
Marks filed a year
<24h
Filing to inbox
USPTO · TSDR
Visit site
Hexagram CastI Ching casting engine
Cast a hexagram from anything — and check the math.
Resolves coins, yarrow stalks, a sentence, a photo, or a song into the same six-line structure, shows the full computation trace behind every cast, and grounds the optional deep reading in the classical text.
8
Casting methods
64
Hexagrams archived
I Ching · SHA-256
Visit site
Each one is an independent product, built and run in-house — every alert and every reading links back to its source.
Want one built for your niche?
Engagements · 07
Two ways
Two ways
to work together.
Hourly when you have the team and just need senior eyes — on data, on an agent design, on an architecture call. Fixed-scope projects when you need the thing itself built end to end. Same two shapes across all four lines: no retainers, no minimum contract length past the first ten hours.
$150/ hour
Senior pair-hours on data, AI, agents, or the systems around them. Best when you have the team — you just need a second set of experienced eyes.
- 10-hour minimum, topped up as needed
- Async reviews on your data, prompts & code, same business day
- Weekly pairing session or two, your cadence
- Slack / Discord / GitHub access
- No long-term commitment, cancel any Friday
From $5k
A scoped deliverable — a monitoring pipeline, an agent, a whole application, or a niche product. Fixed scope, fixed bid, weekly demos.
- Written plan & quote within 2 days of the brief
- Something working inside the first week
- Friday demo + changelog, every week
- Handoff runbook and an architecture note
- 30 days of follow-up support, included
Get in touch · 08
Tell us what to build.
A paragraph is plenty — the data you want watched, the work you want an agent to take over, the application you need standing up, or the niche you want a product in. We read every brief and reply within one business day with a yes, a no, or a referral to someone better suited.