BISKIN RESEARCH · FIELD REPORTS FROM INSIDE THE WORK

Reports on the AI agent economy, written by someone shipping in it.

Most "AI research" today is written by people who've never shipped an agent. Biskin Research is the opposite: field reports from inside the work. Who's winning, who's hiring, where the margin actually goes, which tools survive the next 12 months. Free top-of-funnel issue per report. Paid full report for operators who need depth.

Quarterly
Cadence
FREE
Preview issue
$299
Full report
50+ pp
Per report

Analyst reports are written by people who don't build.

Look at any "AI agents market map 2026" PDF. It's a logo salad made by someone who's never shipped a retry loop, never debugged a hallucinating tool call, never watched a demo fail on stage. The insights are vibes. The numbers are scraped from press releases. The predictions are safe enough to be useless.

Biskin Research is the opposite. Every report is written while I'm building in the same space. I'm an operator, not an analyst. When I say "this stack won't hold past 50 agents," it's because I tried to run 50 agents. When I say "the margin in this category goes to the orchestration layer," it's because I've watched it happen in my own P&L.

The free preview is the first 12 pages — the thesis, the methodology, the top-3 takeaways. Enough to decide if the full 50+ page report is worth $299 to you. If it's not, don't buy it. Simple.

One thesis. Fifty+ pages of evidence.

REPORT 01 · LIVE Q2 2026
Who actually wins the AI agent economy in 2026?
Draft · Q2 2026 · 50+ pp

The thesis.

The AI agent market splits into three layers in 2026: model, orchestration, and interface. Most value accrues to orchestration — the layer retail analysts underweight because it has no flashy models and no consumer brand. Here's who's positioned for it, who's burning cash in the wrong layer, and where the next $10B-hiring-year will land.

  • Layer-by-layer margin analysis (with real numbers, not TAM fiction)
  • 15 companies I'm watching — 5 I'd bet on, 5 I'd avoid, 5 I can't call
  • The two hiring signals that predict winners six months out
  • Where the tool builders vs. tool users actually make money
  • Three bets that look obvious but aren't
$299 · full report
Get the preview →
REPORT 02 · Q3 2026
The retention cliff for AI-native SaaS.
Upcoming · Q3 2026

The question.

2025 AI SaaS pitches all showed hockey-stick ARR. 2026 is when month-6 retention numbers start coming in. The preliminary data: a lot worse than the founders claim. This report breaks down where it holds, where it falls off a cliff, and what the survivors are doing differently at the product level.

  • Retention curves by category (vertical SaaS, horizontal, infra)
  • The wedge features that anchor retention past month 6
  • The wrapper companies that are actually building moats
  • A founder's playbook for fixing a leaky bucket
TBD · subscribe for preview
Get notified →
REPORT 03 · Q4 2026
The solo-operator stack.
Upcoming · Q4 2026

The field report.

What a one-person company actually looks like in 2026 when you've got agents, MCPs, cheap inference, and a sub-$5k tooling budget. I'm running this experiment live — seven product lines, solo, public. The report is the receipts: what worked, what burned money, what compounded.

  • The tool stack (with actual monthly costs, no sponsors)
  • Which agents saved the most time, ranked by hours saved
  • Where AI fails and you still have to do it yourself
  • The "100-hour weeks are a lie" ground truth
TBD · subscribe for preview
Get notified →

Get Report 01's free preview when it drops.

Twelve pages — full thesis, methodology, top-3 takeaways. Enough to decide if the 50-page version is worth $299. Plus a heads-up on every future report, priority-of-access to the operator Discord where I share signal I don't post publicly, and the occasional long email when something in the space changes faster than the quarterly report can capture.

~ Quarterly cadence. Free preview + operator Discord. Unsubscribe one-click.