GTM MEMO  ·  FDM / SCRAPCHEF APRIL 21 2026
INTERNAL MEMO

A go-to-market plan for Foresight Data Machines.

TO
Dr. Tom Kirk, CEO
FROM
James Jackson Leach
RE
First marketing hire
DATE
April 21, 2026
READ
~6 minutes
01

Premise.

The role is two jobs stitched together. First, grow Foresight inside steel using the accounts you already have. Second, take a real AI company out of steel into the next heavy industry without losing the credibility you have built.

This memo covers both. It is operator-level. Proof points are inline, deliberately brief. I would rather talk them through with you than pre-polish them.

02

Week one. Day one.

  1. Ask you three questions. Which deal is stuck. Which two customers you would walk through fire for. What "hire worked" looks like at day 90.
  2. Get Rachael's target-account list. Sit in on the next three sales calls, muted, taking notes.
  3. Ask an engineer to walk me through ScrapChef — physics, not pitch — until I can draw the architecture on a napkin.
  4. Book a flight to Hamburg. Stand next to the furnace. Everything I write afterward will be better for it.
Proof Mort Subite — Austin. I own the bar at 308 Congress. I poured the beer and read the Belgian trade press before I wrote a word of marketing for it.
03

Ninety days.

DAYS 01 — 30

Listen. Audit. Ship one small win.

  • Fifteen structured customer interviews. Output: one messaging doc the team can memorise.
  • Clean UTMs. Event tracking on the demo flow. One weekly dashboard you will actually open.
  • Rewrite the ScrapChef homepage above-the-fold around the strongest piece of customer evidence we have.
Proof Rebirth — rebirth-ai.com. Full B2B site, brand, ten deliverables, three essays, live analytics. Shipped in a weekend.
DAYS 31 — 60

Build what compounds.

  • ArcelorMittal Hamburg case study, done properly. Numbers. Named operators if permission allows.
  • Biweekly technical posts co-written with an engineer. One bylined to Steel Times International or AIST.
  • AISTech 2026 locked. ESTAD and the European EAF Conference scoped.
  • Inside the Furnace — episode one in production.
Proof Longhorn Meat Market — Austin. Wrote the merchandising, product-page, and wholesale-buyer content. Audience was butchers. The lesson is you write the way the audience writes.
DAYS 61 — 90

One campaign. One press hit. One dinner.

  • One integrated campaign on one theme — e.g., the real cost of mix decisions in an EAF — shipped across paper, social, webinar, paid, sales enablement.
  • One tier-one press placement. Three trade-press placements in parallel.
  • First self-hosted dinner. Twelve people, invite-only, London or Austin.
  • Episode one shipped.
Proof Crowdstake — pre-seed launch. Patent, memo, deck, landing page, outbound, Wefunder, Capital Factory, investor dinners at Mort Subite. All off one positioning statement.
04

Targets.

I will not promise pipeline numbers before a baseline exists. Operational targets I hold myself to:

MetricDay 30Day 60Day 90
Long-form assets published013
Technical posts shipped025
Conferences confirmed (12 mo)scoped4–6 locked6–8 locked
Self-hosted events001 done
Press placements013–4
Video episodes01 in post1 shipped
Demo bookings (marketing-influenced)baseline+25%+50%

If I cannot connect a metric to revenue or customer depth, I do not report on it.

05

Creative bet.

Inside the Furnace.

<$60K · 6 episodes

Short-form documentary on YouTube. Eight to twelve minutes per episode, shot inside a customer mill. One technical problem per episode. Reference: Wendover, Apple design docs. Not corporate B2B video.

Audience: metallurgists and operations leaders who watch Practical Engineering on their lunch break. Zero companies in heavy-industry AI are making this. Cost of being wrong: one quarter of video spend. Cost of being right: an eighteen-month moat.

Ship one. Judge after three. Clear kill criteria set up front.

Proof Dirty Weekend — Tribeca Film Festival. Executive producer. $9M of private capital raised and managed. I know what broadcast-quality video actually costs.
06

What I would need.

  • Engineering timeOne hour per engineer per month. I draft 80%. They make sure I am not lying.
  • Your voiceFirst integrated campaign and first tier-one press story need you on the record. After that we build out others.
  • Permission to killIf something is not working at day 60 or 90, I stop it. Includes things I argued for.
  • Customer championOne person at ArcelorMittal willing to go on the record. Most important variable in whether this function works.
  • Budget shapeWeek-two conversation. Not for approval. So I know the shape. Events will be the largest line.
  • BoundariesI use a wheelchair. Site visits are non-negotiable. Plan logistics early.
07

The cross-industry playbook.

Heavy industries do not share buyers. They share buyer behaviour.

A mini-mill plant manager, a foundry director, a cement ops lead, an aluminium smelter engineer — different press, different conferences, same buying pattern. Peers over vendors. Pilot over big-bang. Skeptical of AI. Operators influence the decision as much as the budget holder. The real conversation happens at the dinner, not the booth.

  1. Find the anchor customer first.

    No marketing launch into a new vertical. Commercial closes an anchor. Marketing wraps the story. ArcelorMittal Hamburg was a commercial outcome we amplified, not an asset we manufactured.

  2. Translate. Do not transplant.

    Steel messaging will not work in aluminium even when the product does. Technical credibility is the one thing that carries across.

  3. Hire one industry native per vertical.

    Retired plant engineer, trade-mag editor, ex-ops director. On contract. Credibility filter and relationship opener. Cost is small. Time saved is large.

  4. Find the three rooms that matter.

    Every heavy industry has a short list of conferences, publications, committees that serious buyers watch. First move in a new vertical is a paper, a byline, a committee seat — not a campaign.

  5. Port the infrastructure, not the content.

    Video capability, event playbook, case template, PR relationships, brand system, dashboard. Content reused verbatim kills credibility. Infrastructure reused compounds.

Proof Crowdstake — patent and platform. One system served three customer types (founders, investors, creators). Built for reuse on day one, not retrofit in year two.
08

Brand architecture.

FDM / ScrapChef is already set up for cross-industry expansion. Protect it.

Foresight Data MachinesScrapChef & future product brands
Role The company. The scientists. The lab. The product. What operators and plant managers touch.
Carries Technical credibility. Recruiting. Investors. Research. Cross-industry press. Product messaging. Demos. Case studies. Industry events. Merch.
Scales by Staying constant. Same brand in steel, aluminium, cement. Proliferating. New product brands per vertical on the same FDM platform.
Implication Modest refresh. Becomes the brand that speaks to analysts, journalists, investors. Stays sharp and steel-focused. Not diluted into "AI for metals."

We never rebrand to enter a new industry. We add a product on the FDM platform.

09

Year one, built for year two.

  • Case templateNot a case study. A template with a slot for different nouns.
  • Production capabilityNot video. Director, DP, edit pipeline. Ports to Inside the Foundry.
  • Event playbookNot events. Paper + booth + dinner + capture + follow-up, documented.
  • Brand systemNot guidelines. A system that already accommodates product brand two and three.
  • PR relationshipsNot hits. Journalists who cover industrial and climate-tech broadly.
  • DashboardNot reports. Multi-brand, multi-vertical from day one.

None of this adds cost. It adds discipline. The marketer who skips it makes choices in steel that have to be unwound in year two.