Trossen Robotics · SDK Campaign Rollout

The SDK is live.
Now it travels.

A four-engine content campaign anchored to real events, real activations, and real metrics. Built from story flow session insights and external market intelligence. Designed to carry the SDK story forward with strategic intent and position Trossen as an infrastructure-grade physical AI platform.

PREPARED BY IAN NATANEK FOR TROSSEN ROBOTICS MAY 2026
01 — The Situation

In motion. Not yet moving.

The SDK is live and evolving — open source, actively maintained, shaped by the people using it. The engineering is real. The roadmap is open. But the story hasn't traveled yet. The market hasn't seen what this thing becomes at scale. This campaign doesn't restart — it amplifies what's already in motion and makes sure the right people are paying attention.

Two tracks, one campaign, six months of sustained execution

The SDK campaign sits at the intersection of two parallel priorities for the next six months.

Every piece of content serves one or both.

Track 1: Product Launches

The SDK is the first but not the last. This campaign is the template — an end-to-end system for how Trossen launches products going forward. Content specs, calendar, success metrics, full architecture. Repeatable.

Track 2: Company Repositioning

Trossen is shifting from "hardware company" to "physical AI platform company." The website still reads hardware. The SDK isn't on product pages. This campaign is the first proof point of the new identity.

The six-month window

Physical AI is in its infrastructure moment. Model companies are funded. Deployment is accelerating. But the data layer — the messy middle — is still fragmented. Everyone's running homegrown stacks. The window to become the standard is open right now and it won't stay open.

Six conferences between June and November put Trossen's exact audience in rooms together. Automate is in Chicago — home turf. CoRL has a workshop literally called "Making Sense of Data in Robotics." The external calendar is doing half the work. This campaign rides that wave.

02 — The Spine

One story, told everywhere.

Physical AI has a missing layer. Between the models (Physical Intelligence, DeepMind, Skild) and the deployment (factories, startups, labs) sits an infrastructure gap that nobody owns yet. Data collection, synchronization, embodiment management, fleet operations. Trossen is planting the flag in that gap. The SDK is the first proof point. The platform is the play.

That's the spine. Every piece of content — social post, investor deck, conference talk, guerrilla demo, partner brief — connects back to this. The expression changes per audience and channel. The story doesn't.

How the spine flexes

For Technical People

"Robotics has a data problem. We built open-source infrastructure to fix it." Lead with the architecture: hardware-agnostic, lock-free pipeline, Trust MCAP format, schema-first design. Show the work. Let them read the code.

For Money Gravitators

"We're building the platform layer of physical AI." Lead with the market position: first non-hardware product, open-source wedge to cloud, nobody owns this layer yet. The SDK is the entry point. The cloud is the revenue model. The billion-dollar thesis is the platform.

STORY FLOW SESSION INSIGHTS — MAY 12, 2026

Four messaging pillars surfaced from the in-person session with Anish, Matt, and Marc. These informed the campaign architecture below.

+

In the story flow session, the SDK was positioned against current approaches — ROS bags, HDF5, homegrown stacks — rather than competing products. The scaling problems between model companies and deployment aren't solved. A fuller competitive landscape analysis would strengthen this positioning, but the foundational bet is clear: build the consolidation layer for a space that's still fragmented.

THE MESSY MIDDLE
ANISH SHAH, LINKEDIN + SESSION

Between the brain and utilization sits a scaling gap: fleet management, embodiment decisions, data synchronization. The SDK lives here. Built for scale, not entry-level collection.

"PYTORCH OF DATA COLLECTION"
MARC DOSTIE, SESSION CLOSING

Equivalence pitch — immediately legible to technical and investor audiences. One of several possible hooks alongside "Robotics has a data problem" and "Infrastructure, not theater."

TRUST MCAP — UNDERSOLD
MARC DOSTIE, TECHNICAL DEEP-DIVE

Standardized open-source file format for robotics data. Designed with more forethought than MP3. Mentioned once in the announcement — needs its own content surface. Technical credibility play.

OPEN-SOURCE WEDGE TO CLOUD
ANISH SHAH, STRATEGIC FRAMING

SDK is the first non-hardware product. Open source by design. Natural extension to paid cloud layer (data storage, fleet management). The revenue model behind the open-source play.

"The infrastructure layer of physical AI is fragmented. Everyone's building their own. Trossen is building it for everyone."

THE STRATEGIC BET BEHIND EVERY PIECE OF CONTENT IN THIS CAMPAIGN
+
03 — The Foundation Piece

Before anything else, this video needs to exist.

Every activation, every demo, every blog post in this campaign works better when the audience has a baseline understanding of what the SDK actually does. The SDK page is technically impressive — serious engineering for serious people. But there's an opportunity to bridge "look at this beautiful arm" and "here's how it collects the data that trains the AI" in a way that reaches beyond the people who already know what a lock-free queue is.

90-SECOND SDK WORKFLOW VIDEO

Teleoperate → SDK records → convert → dataset lands on Hugging Face

One continuous shot. A human teleoperates an arm through a task. The SDK is visibly capturing — synchronized data streams, sensor inputs, timestamps. One command converts to LeRobot V2. The dataset pushes to Hugging Face. Ninety seconds. No narration needed — the workflow speaks. Overlay the data flow so viewers see what the SDK is actually doing underneath the physical movement.

WHERE IT LIVES

SDK page hero. YouTube. Every sales conversation. Conference booth loop. Investor email attachment. Partner outreach. The one link that answers "what does this actually do."

WHY IT'S FIRST

If we can't explain what the SDK does in 90 seconds, neither can a researcher landing on the page. This video IS the plain-language explanation that doesn't exist yet. Everything else in this campaign points back to it.

▶ SPEC TREATMENT — SDK WORKFLOW VIDEO
FormatSingle hero film, 60–90 seconds
Cutdowns15s / 30s / 45s for social
ProductionHalf-day shoot, Trossen facility
Turnaround7–10 working days from approval
0:00 – 0:08
COLD OPEN — The arm, alone.
Single Trossen arm on a lab bench. Static. No context, no branding. Lit clean — product shot register. The viewer sees hardware before they see anything else. Three seconds of stillness, then the operator's hands enter frame on the controller.
0:08 – 0:25
TELEOP — The work begins.
Operator teleoperates the arm through a simple manipulation task — pick, place, pour, stack. Camera holds on the physical movement. Then: a subtle overlay fades in. Data streams. Synchronized timestamps. Sensor readouts. The SDK is visibly capturing underneath the movement. The viewer sees the invisible layer for the first time.
0:25 – 0:40
CAPTURE — What the SDK sees.
Split or overlay: physical movement on one side, SDK data architecture on the other. Joint states, camera streams, mobile base odometry — all synchronized, all timestamped, all flowing into a structured pipeline. This is the technical proof. The viewer understands: this isn't recording video. This is capturing structured, training-ready data.
0:40 – 0:55
CONVERT — One command.
Cut to terminal. One line typed. The dataset converts to LeRobot V2 format. No wrestling, no one-off scripts, no manual conversion. The pipeline that used to take weeks, compressed into a single command. Hold on the terminal output long enough for a technical viewer to read it.
0:55 – 1:10
PUSH — The dataset lands.
Browser: Hugging Face dataset page loading. The data that was just collected — structured, converted, published. Ready for model training. Cut back to the arm. Still. The data lives somewhere now. On-screen: "Teleoperate → SDK records → convert → dataset lands. That's it."
1:10 – 1:30
CLOSE — The assertion.
End card. "Stop wrestling your data pipeline. Start collecting." Trossen wordmark. SDK URL. GitHub link. Clean, no music swell, no voiceover. The workflow spoke.
VOICE & TONE: No narration. No music. Servo sounds, keyboard clicks, quiet room tone. Let the workflow carry the argument. The register is closer to a developer demo than a brand film — matter-of-fact, slightly flat, technically credible.

PRODUCTION: 90% captured fresh on-site at Trossen. Single operator, one camera, one light kit. SDK overlay graphics built in post. Social cutdowns (15s/30s/45s) automated from the master. The 30-second cut focuses on beats 2–4 (teleop → capture → convert) for technical audiences.

CAPTURE TOOLING: OBS Studio (free, open source — resonant with Trossen's own ethos) for multi-source compositing. Scene-based layout: arm camera feed + SDK data screen capture layered with adjustable prominence. Hotkey switching between "arm-prominent" and "data-prominent" scenes for the PiP effect. Descript for post-production — transcript-driven editing, automated captions, overlay compositing.
04 — Campaign Architecture

Four engines. One spine.

Four engines means you never go dark. Each one serves a different part of the funnel, produces a different kind of content, runs at a different cadence, and generates downstream deliverables — from social clips to investor materials.

ENGINE 01

Conference Anchors+

Planned content timed to when your audience is already gathered and paying attention. The tentpole moments every 4–6 weeks through November. Produces: blog posts, technical papers, speaker decks, social commentary, investor materials, conference prep kits.

Cadence: Event-driven Funnel: Credibility Audience: Technical + Investors
ENGINE 02

SDK in the Wild+

Applied demonstrations of the SDK working in environments nobody engineered for it. Produces: video clips, livestreams, "What I Learned Today" written receipts, blog recaps, case study slides, technical data visualizations, social posts.

Cadence: Weekly-capable Funnel: Awareness Audience: Broad / Top-of-funnel
ENGINE 03

Purpose Stories+

Disability community partnerships, assistive use cases, the "why this matters beyond robotics" content. Slower, deeper, more deliberate. Produces: long-form written stories, video documentaries, investor narrative material, publishable datasets, influencer collaborations.

Cadence: Quarterly Funnel: Loyalty Audience: Everyone
ENGINE 04

Home Base+

Local, authentic, community-embedded. Trossen as a place, not just a brand. The brewery. The neighborhood. Produces: iPhone-quality video, social posts, local press coverage, community event recaps, behind-the-scenes content, partnership announcements.

Cadence: Ongoing Funnel: Community Audience: Local + Community
[CLOSE]

What Conference Anchors Produce

  • Technical deep-dives timed to conference themes (Trust MCAP at ICRA, ROS comparison at ROSCon)
  • Speaker submissions and workshop proposals (RoboBusiness messy middle talk, CoRL data workshop)
  • Guerrilla activations at/around venues (arm outside Automate, livestreamed to LinkedIn/YouTube)
  • Ad buys in conference programs, sponsorship of relevant sessions
  • Real-time social commentary from Anish/Marc reacting to conference content, tying back to SDK

Data Contribution

Conference demos should produce publishable data. Trained operators, controlled setups even if the environment is wild. This IS a real contribution to the GitHub repo.

[CLOSE]

What SDK in the Wild Produces

  • "What I Learned Today" daily receipts — what the SDK captured, what the synchronization looked like, what the embodiment adapted to
  • Different embodiments at different venues = the hardware agnosticism pitch made physical
  • Livestreams that become clips that become shorts that become blog posts
  • QR codes at every activation leading to GitHub/community
  • One stunt → four channels → weeks of content yield

Data Contribution

Primarily a content play, not a data quality play. The SDK is running — that's real. But public participation produces messy data. Frame it as "community collection" datasets with appropriate caveats. Proves robustness, not research quality.

[CLOSE]

What Purpose Stories Produce

  • Partnerships with disability influencers and assistive technology advocates
  • Documented use cases: a robot learning to assist with daily tasks, adapted to specific needs
  • The "why this matters at a human level" content that lives on Anish's LinkedIn and the blog
  • Investor narrative material — purpose + platform = compelling story

Data Contribution

These could produce genuinely valuable datasets. Assistive robotics data collected carefully with trained operators could be some of the most valuable open-source contributions in the entire campaign. The long play.

[CLOSE]

What Home Base Produces

  • Alter Brewing demos — robot learning to pour a beer at the local brewery
  • Local business partnerships — mechanics, restaurants, community spaces
  • Behind-the-scenes of Trossen's actual workspace and team
  • Community events — "come see the robots" open house
  • Ambient content: doesn't need to be polished. iPhone footage. Authentic because it's not trying.

Data Contribution

Low pressure. The data might be interesting. It goes on GitHub as community collection. It shows the SDK working in environments nobody engineered for it. That has its own value.

05 — Content Taxonomy

What actually gets posted.

Events, demos, and activations produce raw material. But what are the actual content units that get published consistently? Nine content types across three categories. Some are tied to events and demos. Others run on their own cadence regardless of what else is happening. Together they keep the campaign producing consistently.

Technical

Trust MCAP Deep-Dive

Architecture walkthrough, format comparison vs ROS bags/HDF5, design decisions. The credibility play. Marc's voice.

Monthly • Blog + GitHub docs

Technical

SDK Tutorial Series

"First data collection in 15 minutes." Step-by-step walkthroughs, new sensor integration guides, configuration recipes. Reduces time-to-first-capture.

Bi-weekly • Blog + YouTube + Docs

Technical

Release Notes / Changelog

Every SDK update gets a human-readable changelog. What changed, why it matters, who asked for it. Ties back to "support informs product."

Per release • Blog + GitHub

Narrative / Founder

Anish's Industry Posts

Messy middle framing, physical AI market commentary, platform thesis. His LinkedIn voice is already strong — increase cadence, tie to SDK proof points.

Weekly • LinkedIn

Narrative / Founder

Operator Essays

Courtney's register — "support as product," "constraint not theater." Long-form pieces that carry brand weight. The voice that already exists on the blog.

Monthly • Blog

Narrative / Founder

Applied Demo Case Studies

Each activation becomes a written case study: what was the setup, what did the SDK capture, what were the results. Technical enough to be useful, narrative enough to be shareable.

Per activation • Blog
06 — The Calendar

The content calendar hits when attention peaks.

A content calendar isn't "post every Tuesday." It's content timed to when your audience is already gathered and paying attention.

This is what it looks like when four engines run simultaneously. Every dot is an activation, a content drop, or a campaign moment. The visual grid is the overview. The detailed timeline below it is the playbook.

Week 1
Week 2
Week 3
Week 4
JUN
ICRA Vienna
— Trust MCAP drop
— Marc: social commentary
— Tutorial #1
— Anish LinkedIn post
AUTOMATE
— Guerrilla activation
— Livestream
— Alter Brewing
— Automate clips
— WIL receipt
— Blog: Automate recap
JUL
— Tutorial #2
— Community spotlight
RSS Sydney
— Data methodology blog
— Sports arena activation
— Baseball pitching demo
— WIL receipt
— Anish LinkedIn
— Tutorial #3
— Operator essay
AUG
— WIL series launch
— Mechanic shop
— Purpose story #1
— Disability partnership
— WIL receipt
— Tutorial #4
— Community spotlight
— Local activation
— Anish LinkedIn
— SDK release notes
— WIL receipt
SEP
— Local activation
— Tutorial #5
— Community spotlight
— Operator essay
ROSCon Toronto
— Marc: workshop/talk
— ROS comparison content
IROS Pittsburgh
— Content amplification
— Anish LinkedIn
OCT
— WIL receipt
— Tutorial #6
— Local activation
RoboBusiness
— Anish: Messy Middle talk
— Purpose story #2
RoboBusiness clips
— Community spotlight
— WIL receipt
— Operator essay
— SDK release notes
NOV
— Tutorial #7
— Local activation
— Anish LinkedIn
CoRL Austin
— Data workshop
— "Making Sense of Data"
— WIL receipt
— CoRL clips
— Community spotlight
— 6-month retrospective
— Campaign retro
— "What we learned"
2027 planning
CONFERENCE ANCHORS
SDK IN THE WILD
PURPOSE STORIES
HOME BASE
HOT WEEK

Event-by-event playbook

THE DOTS ABOVE → THE DETAILS BELOW

May
Campaign Infrastructure

Finalize Spine Bible. Rebuild SDK landing page with narrative framing. Add SDK to all product pages (one-line fix). Submit speaker proposals for RoboBusiness + CoRL workshops. Begin outreach for disability community partnerships.

Foundation Website fixes Speaker submissions
NOW
June
ICRA 2026 — Vienna

While every robotics researcher is gathered and paying attention to data pipeline discussions, Trossen drops technical content timed to the conversation.

— Trust MCAP technical deep-dive blog post (Marc's voice)
— Real-time social commentary on ICRA data pipeline talks
— Timed social posts tying ICRA themes back to SDK

Production path: 30-minute recorded conversation with Marc. He walks through Trust MCAP design decisions, cross-embodiment forethought, comparison to existing formats. Recording becomes a transcript-edited blog post with embedded video clips. One session → written deep-dive + video inserts for SDK page.

Blog post Social Technical audience
JUN 1–5
Automate 2026 — Chicago (Home Turf)

45,000+ attendees. McCormick Place. Trossen's backyard. The biggest swing in the first three months.

— Guerrilla activation outside venue: "Teach Me" or "Your Robot is Lonely"
— Livestream on LinkedIn + YouTube with QR to GitHub
— Trained operator running live teleop demos
— Post-event: clips, recap blog, case study slide, social for weeks

Guerrilla Livestream Content factory Humanoid pavilion
JUN 22–25
Alter Brewing Activation

Robot learns to pour a beer at the local brewery in Downers Grove. Bartender as teleoperator. Livestreamed, iPhone-quality welcome. "What I Learned Today" receipt: pour attempts, data points captured, synchronization across inputs. Content: "We taught a robot to pour a beer. Here's what the data looked like."

Home base Local viral Community
JUN (FLEX)
July
RSS 2026 — Sydney

High-signal academic robotics conference. Content play: blog post on data collection methodology that researchers can reference. If any Trossen team is attending, live social commentary from sessions. If not, timed content drop on SDK's approach to the problems being discussed.

Blog post Academic audience
JUL 13–17
Sports Arena Activation

Robot attempting to dribble a basketball outside a Chicago sports venue. Absurd. Attention-grabbing. People stop, film, share. SDK capturing every attempt — movement data, sensor synchronization. The receipt: "147 dribble attempts. 3 bounces. 12,000 data points. The SDK doesn't care that it failed — it captured every failure perfectly."

Guerrilla Spectacle Clips
JUL (FLEX)
August
"What I Learned Today" Series Launch

Formalize the content format. Weekly SDK-in-the-wild sessions at local businesses: mechanic shops, restaurants, bars, community spaces. Different embodiment at each. Different pressure sensors, different configurations. Each one produces a receipt. The series becomes the ambient content layer.

Series format Weekly cadence Hardware agnosticism proof
AUG (ONGOING)
First Purpose Story

Disability community partnership goes live. Influencer collaboration documenting assistive robotics data collection. Different register entirely — this isn't spectacle, it's purpose. Lives on Anish's LinkedIn, the blog, in the investor narrative. Answers "why does this matter" at a human level.

Purpose Assistive Long-form
AUG (TARGET)
September
ROSCon 2026 — Toronto

The ROS community IS the competitive positioning audience. SDK's whole pitch: you don't need ROS bags anymore.

— Side-by-side data pipeline comparison: SDK vs. ad-hoc ROS bag workflows
— Marc: speaker or workshop sponsor
— Content framing: "we felt the same pain you feel"

Competitive positioning Workshop Technical
SEP 22–24
IROS 2026 — Pittsburgh

Back-to-back with ROSCon. Content from ROSCon amplified during IROS week. SDK positioning for the intelligent systems community. If Trossen is present: live demo with SDK collecting data from multiple sensor types simultaneously.

Content amplification Demo
SEP 28–OCT 1
October
RoboBusiness 2026 — Santa Clara

Money gravitators. The event's thesis — bridging research and deployment — IS the messy middle.

— Anish: "The Messy Middle of Physical AI" TED-style talk
— LinkedIn post turned into 20-minute presentation
— SDK as proof point, platform repositioning as the frame
— Investor-facing engine at maximum output

Speaking slot Investor audience Platform repositioning
OCT 14–15
November
CoRL 2026 — Austin

Conference on Robot Learning. The culmination of six months of credibility building.

— Target: "Making Sense of Data in Robotics" workshop demo or poster
— Pre-conference blog on data collection methodology
— If accepted: the SDK sells itself in that room

Workshop target Robot learning audience Capstone
NOV 9–12
Six-Month Campaign Retrospective

Metrics review against benchmarks. What worked, what didn't. Content: "What we learned launching an open-source SDK" — transparency piece that feeds the community narrative and sets up 2027 planning.

Retrospective Transparency 2027 setup
NOV (END)
07 — Applied Demonstrations

Real-world SDK, not lab SDK.

These aren't just stunts. They're applied demonstrations of the SDK working in environments nobody engineered for it. Each one shows a different capability, produces weeks of content, and proves something specific about the platform. Design moments that produce content, not content that hopes to become moments.

🏪

"Teach Me" at Automate

McCormick Place, Chicago • June 22–25

Trossen arm outside the venue. Livestreamed on LinkedIn + YouTube. Attendees walk up, teleop the arm, create training data. QR code to GitHub. "Watch 47 strangers teach a robot to pick up a coffee cup."

YIELD: Livestream → highlight reel → clips → blog post → SDK landing page video → social for weeks
▶ SPEC TREATMENT — AUTOMATE (Repeatable Event Activation) +
FormatLive activation + livestream, full day
Crew1 operator, 1 camera, 1 host
LocationOutside McCormick Place, Chicago
OutputLivestream + 20+ clips + 1 recap film
PRE-EVENT
Setup & staging.
Trossen arm on a portable table outside the venue entrance. Signage: "Your Robot is Lonely" or "Teach Me." QR code displayed prominently linking to GitHub/community. Livestream goes live on LinkedIn + YouTube 30 minutes before doors open. SDK running, data capture active from minute one.
MORNING
Attendee engagement — arrival wave.
Attendees arrive, see the arm. Host invites them to teleop — "teach the robot something." Each interaction is an SDK data collection session. Camera captures reactions, attempts, conversations. The livestream shows the SDK data overlay in real time — viewers at home see what's being captured as each person interacts.
MIDDAY
The receipt goes live.
First "Earned It / Learned It" receipt posted to social mid-day: "47 people have taught the robot so far. 312 episodes captured. 84,000 synchronized data points. Here's what the SDK learned this morning." Real numbers, real data, posted while the event is still running. Creates FOMO for afternoon attendees.
AFTERNOON
Peak traffic + signature moments.
Conference breaks drive foot traffic. Camera operator captures the best interactions for the highlight reel. Host identifies interesting attendees for brief on-camera reactions: "What did you just teach it?" The livestream audience peaks. QR scans tracked in real time.
POST-EVENT
Content production sprint.
Same-day: final receipt with full-day numbers. Day 2: highlight reel cut (2–3 minutes). Week 1: 10+ social clips extracted. Week 2: blog post recap with data visualizations. Month 1: case study slide for investor deck showing QR-to-clone conversion metrics.
The activation is simultaneously marketing, content production, SDK demonstration, community building, and data collection. One day, five functions, weeks of output.
🍺

Alter Brewing Pour

Downers Grove, IL • Flexible

Robot learns to pour a beer at the local brewery. Bartender is the teleoperator. The content: what the data looked like, what the SDK captured, what went wrong, what the arm learned about tilt and flow rate.

YIELD: Local-viral video → "What I Learned Today" receipt → SDK proof-of-concept → community content
🏀

Robot Dribble

Chicago Sports Venue • Flexible

Robot attempting to dribble a basketball. Absurd and shareable. The receipt: "147 attempts. 3 bounces. 12,000 synchronized data points. The SDK captured every failure perfectly." Infrastructure doesn't need success. It needs clean data.

YIELD: Spectacle video → shorts → "failure as data" narrative → SDK reliability proof
🔧

Mechanic Shop Session

Local Auto Shop • Flexible

SDK collecting data in a real environment where physical AI has actual applications. Different pressure sensors, different embodiment config. Shows hardware agnosticism in a context people understand: dirty, loud, real.

YIELD: "What I Learned Today" receipt → industrial use case video → technical blog on sensor adaptation
🚒

Fire Station Demo

Local Fire Station, Downers Grove • Flexible

Emergency services + robotics. Right down the street. Robot assisting with equipment handling, hose manipulation, tool sorting. Real-world application with genuine public interest. The kind of local partnership that makes news and demonstrates SDK capturing data in high-stakes environments.

YIELD: Local news potential → first responder use case → community goodwill → "What I Learned Today" receipt

Assistive Robotics Partnership

Partner-Dependent • Q3 Target

Disability influencer collaboration. Documenting assistive data collection — a robot learning daily tasks adapted to specific needs. Not guerrilla. Purpose. The SDK's hardware agnosticism means genuine adaptability for different accessibility needs.

YIELD: Long-form story → Anish LinkedIn feature → investor narrative → valuable open dataset
🏥

Hospital / Care Facility

TBD • Requires Partnership

Healthcare is where robotics is heading. A controlled demo in a care facility — robot learning to assist with routine tasks. Data quality matters here: trained operator, careful setup. The content serves the purpose story engine AND the technical credibility engine.

YIELD: Purpose story → technical case study → publishable dataset → conference presentation material
📣

"Your Robot is Lonely"

Any Conference Venue • Repeatable

An arm outside a conference handing out flyers that say "Your Robot is Lonely." Livestreamed. QR to community/GitHub. A repeatable format that works at any event. The arm IS the marketing. The SDK IS the infrastructure recording it all.

YIELD: Repeatable guerrilla template → event-specific clips → ongoing series

Robot Pitching Machine

Local Baseball Field • Flexible

Robotic arm throwing pitches — 20% curveballs, 20% fastballs, the rest mixed. SDK capturing trajectory data, joint angles, release timing for each pitch type. Different movement profiles, same infrastructure. The data becomes a training dataset for pitch classification.

YIELD: Spectacle video → pitch data visualization → "diff configs, same SDK" proof → sports community crossover
🌟

Earned It / Learned It

Daily Format • Any Activation

The content format that ties it all together. After every activation, the same-day receipt: what the SDK learned, what data was captured, what the milestones were. Real-time exemplification. A daily proof of life that adds up to an undeniable body of work.

YIELD: Daily social content → weekly compilation → monthly highlight reel → quarterly retrospective
🎬

SDK Encounters Series

Tabletop / Studio • Weekly Episodes

Recurring video format. Each episode: one human task, one robotic attempt, one SDK receipt. Crocheting (fine motor dexterity, thread tension data). Whack-a-Mole (synchronization proof — why lag creates delay back propagation, made intuitive). The task is the hook. The attempt is the content. The receipt is the proof. Lives on the Learn page alongside existing series, gets cut into clips for social.

YIELD: Weekly episodes → social clips → technical data showcases → Learn page series
🤟

ASL / Sign Language

Studio + Partner • Purpose Story Tie-in

A robotic arm learning sign language. Hand shape data, finger positions, transitional movements between signs. Assistive technology with real emotional resonance. Connects directly to disability community partnerships. "Teaching a robot to communicate."

YIELD: Purpose story → assistive tech showcase → meaningful open dataset → influencer collab
🎨

The Sistine Chapel Reach

Studio • Signature Piece

Two robotic arms recreating the Creation of Adam — reaching toward each other. Instantly iconic. Everyone knows the image. Underneath: the SDK synchronizing data from TWO embodiments simultaneously, coordinating movement profiles, recording approach trajectories. The image becomes a poster, a LinkedIn hero, an event banner, a t-shirt. The data becomes a dual-embodiment synchronization showcase.

YIELD: Signature visual → campaign hero image → dual-sync technical proof → merchandise potential
▶ SPEC TREATMENT — SISTINE CHAPEL REACH +
FormatSignature photo + 30-second film
ProductionControlled studio shoot, 2–3 hours
HardwareTwo Trossen arms, facing each other
OutputHero image + film + social cuts + print assets
SETUP
Two arms. One table. Dark background.
Two Trossen arms mounted facing each other, approximately 18 inches apart. Dark backdrop — black or deep charcoal. Single dramatic light source from above, slightly warm. The visual reference is unmistakable: the Creation of Adam. The gap between the fingertips is the frame.
THE REACH
Both arms extend simultaneously.
Slow, deliberate movement. Both arms reaching toward each other from opposite sides. The SDK is capturing synchronized data from both — dual-embodiment coordination, approach trajectories, joint angles from two separate systems harmonized into one dataset. The movement takes 8–10 seconds. Camera captures from the side, low angle, fingertips approaching center frame.
THE MOMENT
Fingertips meet. Hold.
The grippers touch — or almost touch. Hold the frame for three full seconds. This is the hero photograph. Shoot stills during this hold. The image needs to work as: LinkedIn hero banner, conference booth backdrop, presentation title slide, t-shirt, poster. One image, ten surfaces.
THE DATA
Overlay reveals what the SDK captured.
For the film version: after the moment, a data overlay fades in. Two synchronized data streams — one per arm — running in parallel. Timestamps aligned. The visual makes the technical point visceral: the SDK doesn't just capture one embodiment. It synchronizes across two simultaneously. This is the dual-embodiment proof that Marc's architecture enables.
CLOSE
Pull back. Both arms retract.
Arms return to rest. On-screen text: "Two arms. One SDK. Synchronized." Trossen wordmark. The film version is 30 seconds total — short enough for any social platform, striking enough to stop a scroll.
PRODUCTION: Controlled environment, Trossen facility. Two arms, one camera, one overhead light, dark backdrop. This is the lowest-logistics, highest-yield shoot in the campaign — 2 hours of setup and capture produces the signature visual asset for the entire six months. The still photograph alone justifies the shoot.
TABLETOP

Start this week.

Arm + SDK + camera. SDK Encounters series, Sistine Chapel reach, ASL. No venue needed, no travel. Produces weekly content from Trossen's own space.

LOGISTICS: Minimal
LOCAL

Venue partnership + transport.

Alter Brewing, mechanic shop, fire station, baseball field, sports arena. Different embodiment at each — the hardware changes, the SDK stays constant. Proves hardware agnosticism without saying the word.

LOGISTICS: Moderate
EVENT

Planning + travel.

Automate guerrilla, conference demos, "Your Robot is Lonely" at venues. Logistics planning, potential booth costs, trained operators. Highest yield per activation.

LOGISTICS: Full planning cycle

ALL THREE TIERS → CONTENT + "EARNED IT" RECEIPTS + DOWNSTREAM DELIVERABLES FOR INVESTORS, SALES, PARTNERS

08 — The First 90 Days

This is the production plan.

Everything above is the architecture. This is the committed sequence — fourteen moves in 90 days, sequenced by dependency and impact.

CONFERENCE ANCHORS
SDK IN THE WILD
PURPOSE STORIES
HOME BASE
Week 1–2: Foundation
MAY 12–25
Add SDK to every product page
WidowX AI lists seven third-party software products and not the one Trossen built. One-line fix per page. A researcher who finds you via the arm gets a path to SDK.
Rewrite homepage hero
Replace workstation specs with "Robotics has a data problem" or equivalent operator voice. Zero new writing. Material already published.
Submit speaker proposals: RoboBusiness + CoRL
Anish: "The Messy Middle of Physical AI" at RoboBusiness (Oct 14–15). Marc: target CoRL "Making Sense of Data" workshop (Nov 9–12). Deadlines approach. Submit now.
Begin shooting the 90-second SDK workflow video
Teleoperate → SDK records → convert → dataset lands on Hugging Face. One continuous shot with data overlay. The foundation piece every other deliverable references. Marc directs the technical accuracy.
Draft investor market positioning brief
"The infrastructure layer of physical AI is fragmented. Here's why Trossen consolidates it." Two pages. For every capital conversation in the next six months.
Begin Neuracore video clip production
Matt's 83-minute Neuracore interview — nine timestamped segments already identified. 15 downstream assets. Zero new capture. The content pipeline starts producing from day one.
▶ CLIP BREAKDOWN + DESTINATIONS
0:00 Founding story → About page · brand video opener · conference intro reel
3:00 Leap into manipulators → Product pages · evolution narrative
15:50 Humanoids vs. real-world needs → LinkedIn clip · contrarian social content
18:30 Data collection bottleneck → SDK landing page embed · LinkedIn post · standalone clip
27:30 Modern ML pipelines beyond ROS → Technical blog · developer audience
34:20 VLA models explained → /ai overview page · 10-min educational explainer
47:20 Early commercial use cases → Sales enablement · customer/market stories
53:00 Why research drives robotics → University lab sales narrative
1:18:00 Final thoughts on robotics future → Category-level POV · brand video close

+ pull-quote graphics, multi-part blog series, newsletter excerpts, conference reel → 15 assets from one existing video. No new shoots required.

Week 3–4: First Content
MAY 26 – JUN 8
Publish 90-second video → SDK page hero + YouTube
The video goes live. Embed on rebuilt SDK landing page. Upload to YouTube. Share across LinkedIn. This is the "before" — everything after this has a reference point.
ICRA week: Trust MCAP deep-dive drops
Technical blog post timed to June 1–5 while every robotics researcher is paying attention. Marc's voice. Architecture walkthrough, format comparison vs ROS bags/HDF5, design decisions that make it last.
First SDK Encounters episode: shoot + post
Tabletop. Arm + SDK + camera. Crocheting, Whack-a-Mole, or another first episode. The format establishes: one task, one attempt, one receipt. Posted to Learn page + social. Weekly from here.
Week 5–7: Automate
JUN 9–29
Alter Brewing activation
Robot learns to pour a beer at the local brewery. Warm-up for Automate. Tests the activation format, the livestream setup, the "Earned It / Learned It" receipt. Low stakes, high charm. Content for weeks.
Automate 2026 — the big swing
June 22–25, McCormick Place. 45,000+ attendees. Home turf. Guerrilla activation outside the venue: "Teach Me" or "Your Robot is Lonely." Livestreamed. QR to GitHub. Trained operator running the arm. One activation → months of content.
Post-Automate content production
Cut the livestream into clips. Write the Automate recap blog. Produce the "Earned It" receipt. Build the case study slide for the investor deck. Extract social posts for two weeks. One event, full content pipeline.
Week 8–12: Cadence
JUL – AUG 10
SDK Encounters: weekly episodes continue
The tabletop series is now a rhythm. New episode every week. Each one adds to the body of work. The Learn page grows. The social clips compound. The GitHub repo gets more visible.
Sistine Chapel Reach: the signature shot
Two arms. The Creation of Adam. Dual-embodiment synchronization captured by SDK. This produces the campaign's hero image — poster, LinkedIn banner, event signage, t-shirt. Schedule it when the content pipeline needs a spike.
Begin disability community outreach for Purpose Story
Identify influencers and advocates. Start conversations about assistive robotics data collection. ASL demo as potential collaboration format. This is a Q3 deliverable — the groundwork starts now.

At the end of 90 days:

The SDK has a hero video on every surface. The website carries the operator voice. Trossen showed up at the biggest automation event in North America with a guerrilla activation that's still producing content. A weekly video series is building an audience. Speaker slots are locked for fall. The investor brief is in circulation. And the "Earned It / Learned It" receipt has 12+ entries proving the SDK works everywhere it goes.

09 — Downstream Deliverables

Content becomes capital.

Every activation, every conference moment, every demo produces raw material. That material doesn't just become social posts — it becomes the investor narrative, the sales enablement, the partner briefs. One flywheel, multiple outputs. Here's what the four engines produce beyond content.

Investor Materials

Platform Repositioning Deck

Pitch deck module showing Trossen as physical AI platform, not hardware company. SDK as proof point. Cloud as revenue model. Conference traction and activation metrics as evidence of market pull.

Updated quarterly • From Engine 1 + Engine 3 outputs

Investor Materials

Traction One-Pager

Single page: SDK adoption numbers, GitHub activity, conference presence, community growth, activation highlights. The "here's what happened since we last talked" document. Numbers, not narrative.

Monthly • From metrics funnel data

Investor Materials

Market Positioning Brief

The "nobody owns this layer" argument in two pages. Physical AI infrastructure gap, Trossen's position, SDK + cloud thesis. For board conversations and investor intros.

Once, updated as needed • From spine + market context

Internal / Ops

Launch Playbook Template

This SDK campaign becomes the template for every product launch going forward. Documented process: spine → engines → calendar → activations → metrics → downstream. Repeatable by design.

Once, post-campaign • This document is the prototype

Internal / Ops

Conference Prep Kits

Per-event package: talking points aligned to spine, demo scripts, activation logistics, QR code assets, follow-up email templates, social post templates. Everything needed to show up ready.

Per conference • 2 weeks before each event

Internal / Ops

Quarterly Narrative Update

Board-ready summary of campaign performance, market positioning progress, and repositioning milestones. Ties the content activity back to company trajectory.

Quarterly • From all engines
10 — Decisions + Actions

What needs answering. What needs fixing.

Open questions that need answers before certain campaign elements can execute, and structural fixes on existing surfaces that cost near-zero and clear the path.

Open questions

Who teleoperates at activations? A trained Trossen operator produces publishable data. Public participation produces messy data but great content. Both are valid — they serve different purposes. Can't pretend mode two is mode one.

Is there AI baked into the demo? The SDK captures data. The AI happens downstream. "Watch a robot learn" is misleading if it's collecting data, not learning in real time. The honest framing is better: "This is what infrastructure looks like. The magic comes later." If trained policies exist, show the full loop.

Staffing for activations. Does Trossen have people who can travel with an arm? Or is it always Downers Grove? This constrains Engines 2 and 4 more than 1 and 3. The Home Base engine is inherently local. SDK in the Wild can expand as the team grows.

Conference budget. Speaker submissions are free. Guerrilla activations outside venues have logistics costs. Booth space at Automate is expensive. Ad buys in conference programs have rates. The campaign is presented expansively — Matt and Anish narrow based on budget reality.

Cloud product timeline. How far along is the cloud layer? If it exists, the SDK-to-cloud narrative gets featured prominently. If it's in development, the narrative is forward-looking. This affects how aggressively we push Pillar 4.

OOI and egocentric data collection. Anish mentioned these are coming soon. Are they part of the SDK campaign or separate launches? If they're bundled, the campaign gets bigger. If separate, they're the next exercise using this same four-engine template.

Customer stories + case studies. Labs are using Trossen hardware. If any SDK users can be named — or even anonymized — each one becomes a case study for the website, a one-pager for sales conversations, and potential video content. Options range from full named partnerships to anonymized usage profiles to "representative examples" based on real patterns. The activations themselves also become case studies — Trossen is its own first customer for the "What I Learned Today" format.

Structural fixes

Placement and architecture problems, not writing problems. The material already exists across blog posts, essays, and product descriptions — it just needs to be surfaced and placed where people can actually find it. Near-zero effort, high unlock.

Add SDK to product pages

The WidowX AI page lists seven third-party software products and not the one Trossen built. A researcher who finds you via the arm has no path to SDK. One-line fix. Do it this week.

Rewrite homepage hero

Current hero rotation: workstation specs, product line names, UGV taxonomy. The operator voice is one click below in Latest News. "Robotics has a data problem" should be the homepage hero. Zero new writing required.

Build a roadmap surface

The roadmap exists in fragments across three posts. "Help shape the roadmap" assumes someone can see it. A /roadmap page or a section of the SDK landing page. The material to build it is already published.

Feature Trust MCAP

Mentioned once in the announcement. Marc called it one of the most important storage decisions. It needs its own section on the SDK page, its own blog post, and a comparison to existing formats. The credibility play is sitting there undersold.

Rewrite SDK landing page

Currently paragraph-heavy and engineer-only. PyTorch became the dominant AI framework not through marketing but by removing friction — engineers stopped fighting the tool and did the work. The SDK page should lead with that same promise: "Stop wrestling ROS bags. Start collecting." The 90-second workflow video as the hero. Community-tested framing. Make a researcher feel like this tool removes problems, not adds complexity.

11 — Success Metrics

Measure the funnel, not the noise.

GitHub stars are vanity metrics. Bessemer Venture Partners — one of the top open-source investors — ignores them. What matters is the conversion funnel: from awareness to activation to retention to revenue signal. Each engine gets measured against this.

Awareness
QR scans
Page visits
Social impressions
Conference encounters
Interest
GitHub clones
Docs page views
— SDK page time-on-page
Newsletter signups
Activation
SDK installed
First data collection run
First episode recorded
"Time to first capture"
Retention
Repeat usage (30-day)
Issues filed
— Community posts
External contributors
Revenue Signal
Cloud signups
Hardware inquiries from non-Trossen users
Partnership inbound
Investor conversations

Benchmark context

Most emerging open-source projects would be in rare territory reaching 100 unique monthly contributors consistently. Only 2% of the top 10,000 GitHub projects have ever hit 250 monthly contributors for 6+ months. Trossen doesn't need to be Kubernetes. It needs to be the standard in its niche — robotics data collection.

A realistic 6-month target: 50–100 monthly contributors, 500+ GitHub stars, measurable conversion from QR activations to SDK installs.

What "1,000 people try the SDK" means

Anish named this benchmark. To make it real: define "try" as "completed first data collection run." Track with a lightweight telemetry ping (opt-in). 1,000 first-run completions in 6 months is ambitious but achievable with the four-engine approach. That's ~40/week sustained.

The QR-to-clone pipeline

Every guerrilla activation gets a unique QR code. Track: scans → GitHub visits → clones → first run. "347 QR scans at Automate led to 89 GitHub clones led to 23 first-run completions in 30 days." That's the number Anish takes to investors.

The Commitment

The 90-day plan is the handshake.

Section 08 is the production plan — fourteen moves in sequence, each unlocking the next. The calendar in Section 06 shows six months of density anchored to real industry events. The engines, the demos, the content taxonomy, the downstream deliverables — they're all ready to run.

What's needed now is one decision: greenlight the first 90 days. The plan is in Section 08. The calendar is in Section 06. Everything else cascades from there.

The SDK is live and evolving. The arms are already in labs — and those labs are the first stories waiting to be told.

This is the campaign that carries it.

CAMPAIGN INVENTORY — 37 ITEMS PROPOSED

White = in the 90-day plan · Gray = extended arsenal

SPEC TREATMENTS 3
SDK Workflow Video
Automate "Teach Me"
Sistine Chapel Reach
CONTENT 12
Neuracore Clips (9 segments)
Trust MCAP Deep-Dive
SDK Tutorial Series
Release Notes / Changelog
"Earned It" Receipts
Conference Commentary
Community Spotlights
Anish Industry Posts
Operator Essays
Demo Case Studies
SDK Encounters Series
Customer / Lab Case Studies
BUSINESS + INVESTOR 6
Market Positioning Brief
Traction One-Pager
Platform Deck Module
SDK Integration Brief
Customer Comms
Quarterly Narrative Update
ACTIVATIONS 9
Alter Brewing Pour
Robot Dribble
Mechanic Shop
Fire Station Demo
Assistive Robotics
Hospital / Care Facility
"Your Robot is Lonely"
Baseball Pitching
ASL / Sign Language
WEBSITE 5
SDK on product pages
Homepage hero rewrite
/roadmap surface
Feature Trust MCAP
SDK page rewrite
INTERNAL 2
Launch Playbook Template
Conference Prep Kits