BRIAN KENNETT
№ ——.——·——:——:—— ———

Brian Kennett

Twenty years on the commercial side of local media — now building the software the business actually needs.

I.Practice

  • Media planning & order management
  • AI prospect research & competitive intelligence
  • Political, legal & creative compliance review
  • Contract-aware pricing engine
  • Real-time inventory across digital, print & email
  • Signal-weighted sales forecasting
  • Document intelligence & semantic search
  • Agentic order assistant
  • HubSpot CRM bridge
  • Enterprise cost governance

II.Dispatches

HubSpot CRM bridge

30+ structured line-item fields, bidirectional sync

8m ago

Fulfillment details panel

26 product configs with inline defaults

27m ago

Media order generator

contract-aware pricing bridge, era 1 / era 2 formats

1h ago

Sales forecasting engine

daily deal sync + weekly pipeline snapshots

3h ago

Ask Brian assistant

real-time agent chat over the operating ecosystem

6h ago

III.Live

▸

AI prospect research

competitor scoping, SEO, social signals

stage 4 of 7

▸

Document intelligence

submission parse + index

queued

▸

Weekly pipeline forecast

deal scoring, revenue projection

24h cron

▸

Compliance review

political, legal, creative

human in loop

▸

Inventory sync

digital, print, email audiences

running

▸

Order sync

CRM deals + line items

idle

IV.This Week

AI Spend
$47.12
LLM Calls
2,864
Deploys
7
Prod Tasks
12
◆deploy succeeded · hubspot crm bridge · 8m◆ai prospect research · 14 competitors scoped◆capability gap · closed without a vendor contract◆compliance review · 4 items escalated to legal◆12 background tasks active◆document intelligence · legal submission parsed◆weekly forecast · pipeline $1.84m◆contract-aware pricing · 505 products · 951 rates◆ad inventory sync · 29 exclusive placements refreshed◆email audience sync · 47 segments◆agentic order assistant · 312 plans generated◆media planning · $40m in annual revenue orchestrated◆deploy succeeded · hubspot crm bridge · 8m◆ai prospect research · 14 competitors scoped◆capability gap · closed without a vendor contract◆compliance review · 4 items escalated to legal◆12 background tasks active◆document intelligence · legal submission parsed◆weekly forecast · pipeline $1.84m◆contract-aware pricing · 505 products · 951 rates◆ad inventory sync · 29 exclusive placements refreshed◆email audience sync · 47 segments◆agentic order assistant · 312 plans generated◆media planning · $40m in annual revenue orchestrated
◆
LeaderThe mission

grew a local digital advertising business to $40 million in annual revenue with industry-leading margins and a 92% client retention rate. While running that operation, I watched enterprise tools burn through our budget — $22,000 a month for platforms that delivered generic reports nobody acted on.

No CS degree. No engineering team. No venture funding. Just twenty years of domain expertise and a conviction that the people who understand the problem should be the ones building the solution.

So I built it. All of it — while still running the business. An entire operating ecosystem — media planning, order management, competitive intelligence, sales forecasting, document processing, inventory management — powered by agentic AI that actually understands what our clients need.

Not automation. Intelligence. Systems that research prospects, analyze competitors, score deals, and generate strategic recommendations tied to real client needs. This isn’t where the industry was headed — it’s where it needed to be decades ago.

Why? Because local journalism is dying — and nobody else is going to save it. Every dollar of efficiency I create funds another reporter, another investigation, another story that holds power accountable. If the business side can’t sustain the mission, the mission disappears. So why not me.

Case № IThe capability gap

The old platform cost twenty-two thousand three hundred dollars a month, and couldn't do any of the interesting things.

What the vendor charged for was a reporting surface. What the business actually needed was everything underneath it: agentic prospect research, contract-aware pricing, real-time inventory across digital, print, and email, compliance workflows, document intelligence, forecast-grade sales data, and a CRM bridge that understood the work. None of it came in the contract. None of it was on the roadmap. None of it existed in the market then. None of it exists in the market now, either.

So it got built. On the same stack every modern company already pays for — Postgres, React, TypeScript, a handful of AI providers — in the hands of the person who lived inside the problem. The cost difference is real, but it isn't the point. The point is that no vendor contract at any price was going to close this gap. It took someone who carried a number for twenty years and decided to stop waiting.

Case № IIReframe

This isn't automation.
This is intelligence.

The media industry has been stuck with tools designed for a different era. So I built what should have existed all along.

The industry standard

  • ×Static reports nobody reads.
  • ×Generic recommendations for every client.
  • ×Tools built by engineers who never carried a number.
  • ×Months of dev cycles for routine changes.
  • ×$22K a month for dashboards that collect dust.

What I built

  • ✓Agentic research that treats each prospect as its own problem.
  • ✓Strategic recommendations tied to actual client needs.
  • ✓Built by the operator who lives inside the problem.
  • ✓New capabilities shipped in hours, not quarters.
  • ✓An entire operating system for less than the price of one dashboard seat.
◆
Case № IIIThe capabilities

Enterprise-grade. Solo-built.

A short catalog of what has shipped and what it does.

I.

Agentic AI prospect research

A multi-stage intelligence pipeline: business resolution, competitor discovery, SEO profiling, social analysis, and strategic recommendations. Not keyword reports — real competitive intelligence powered by multiple frontier models.

II.

Enterprise cost governance

Real-time spend tracking across fifteen-plus service providers. Budget alerts, coverage gap detection, spend reconciliation, and forecast-aware procurement.

III.

Contract-aware pricing engine

Five-level rate fallback: contracted override, tier rate, catalog lookup, legacy rate, safe default. Handles volume discounts, seasonal multipliers, and contract drawdown tracking.

IV.

Full-stack HubSpot bridge

Bidirectional CRM sync of companies, deals, contacts, fulfillment pipeline, behavioral events, and thirty-plus structured line-item properties. Not a connector — an operating bridge between systems.

V.

Real-time inventory

Live availability from the ad server, email lists, and print placements. A hold system with expiry, conflict detection, calendar visualization, and multi-source reconciliation.

VI.

Signal-weighted forecasting

Pipeline scoring across six dimensions: deal stage, activity, Compass data, velocity, temporal patterns, and historical win rates. Commit / likely / upside / longshot tiers, automated weekly snapshots.

VII.

Domain-aware AI assistants

Custom copilots embedded directly into the workflow. Not generic chatbots — trained on the data, the clients, the products. Answers grounded in real business context and historical performance.

Case № IVThe ecosystem

Not a project. A platform.

What started as one replacement became a connected operating ecosystem. Planning, forecasting, compliance, document intelligence, and orchestration all share infrastructure, data contracts, and AI capabilities across a family of applications.

The flagship

Media Planning & Orders

An end-to-end operating system for selling media: prospect research, contract-aware pricing, rate exceptions, order generation, amendment workflows, and full CRM bridge.

  • ·Agentic prospect research with multi-provider AI
  • ·Five-level contract-aware rate resolver
  • ·175+ relational tables, production Postgres
  • ·Real-time inventory from ad server, email, and print
  • ·Bidirectional CRM sync for deals and line items
  • ·PDF & deck generation in the strategist's voice
React·TypeScript·Postgres·Clerk·Trigger.dev

The next generation

Agentic Analysis Pipeline

A 3,000-line agentic pipeline that resolves a business, discovers competitors, enriches SEO and social signals, runs strategic analysis, and returns a ready-to-review plan.

  • ·Seven-stage agentic orchestration
  • ·Multi-model provider routing
  • ·Monorepo with shared schema and packages
  • ·Durable execution via Trigger.dev
TypeScript·Trigger.dev·Drizzle·Neon

The operations layer

Forecasting & Operations

The ops layer across account management, execution, and revenue. Signal-weighted pipeline forecasting, weekly snapshots, fulfillment health, and Slack-integrated alerts.

  • ·Six-dimension pipeline scoring
  • ·Thirty / sixty / ninety day revenue forecasts
  • ·Fulfillment health + SLA tracking
  • ·Enterprise cost governance dashboard
Next.js·Neon·HubSpot·Snowflake

The review orchestrator

Political, Legal & Creative Review

Submission intake, automated first-pass analysis, human review loops, legal escalation, and a full audit trail for the messy regulated workflows weak software collapses on.

  • ·AI-assisted compliance evaluation
  • ·Human-in-the-loop review and escalation
  • ·Submission + artifact storage
  • ·Event-driven fulfillment handoffs
React·Express·Clerk·Cloudflare R2

Operational memory

Document Intelligence

Parsing, indexing, and semantic retrieval over enterprise documents. Multi-tenant processing with chat over the corpus and access controls.

  • ·LlamaCloud parsing pipeline
  • ·Vector search + semantic retrieval
  • ·Tenant-scoped access
  • ·Chat over the corpus
Next.js·Clerk·Neon·LlamaCloud
Case № VThe method

The future isn't learning to code.
It's knowing what to build.

AI did not replace the need for expertise. It unlocked the people who have it.

Not no-code.

Full TypeScript, React, Node, Postgres. Production-grade architecture with real engineering patterns — nothing abstracted behind a drag-and-drop builder.

AI as collaborator.

Claude Opus, GPT-5, and Gemini are not writing code on their own. They are thinking partners that amplify domain expertise into working systems.

Domain expert > dev team.

Twenty years of media sales knowledge means I build what actually solves the problem — not what an engineer guesses the problem might be.

Ship in hours, not quarters.

New capabilities go from idea to production in a single session. No sprint planning. No ticket backlogs. No waiting.

Case № VIOn the record

Five things worth saying once.

Quotable notes from the margins of the work.

On building

“Every line of software is an opinion about how the business should work. When you outsource the software, you outsource the opinion.”

On buying software

“A vendor who doesn't understand your work isn't a partner. They're rent. And the lease always goes up.”

On AI

“AI didn't make me a developer. It let twenty years of domain knowledge finally build itself into something usable.”

On the job

“A newsroom's survival is decided by the work that never gets a byline — ad ops, pricing, compliance, forecasting. That's the whole job.”

On local news

“The cliff is close. When a local newsroom goes dark, the stories only a local reporter would have found go dark with it. I run a $40 million agency and build the software underneath it. Both exist for one reason — keep the newsroom open.”

Case № VIIIn conversation
LIVE · CONVERSATIONAL · GROUNDED IN THE WORK

Ask Brian

A conversation with the operator who built the systems on this page. Ask about the work, the tradeoffs, or the decisions a software vendor would never have made.

Brian Kennett

Pick a question below or write your own.