DESIGN PARTNERSOPEN CALL · 4 TO 6 ANCHOR SLOTS · APPLICATIONS OPEN UNTIL Q4 2026 VERTICALSINDUSTRIAL · AUTONOMY · EARTH OBSERVATION · ROBOTICS MUTUAL COMMITMENTDATA · DOMAIN ACCESS · DEPLOYMENT REPS · 12 MONTHS CONTACTPARTNERS@BULUTLABS.COM · 48-HOUR RESPONSE
Bulut Labs/Design partners

Open call. Four to six anchor design partners.

We are not yet selling a product. What we are doing — for a small, deliberately chosen group of operators across four verticals — is co-developing the BULUT-1 model line with the people who will eventually use it. This is the brief. If it sounds like you, write to us.

Slots 4 to 6 anchor partners Verticals industrial · mobility · EO · robotics Term 12 months · renewable Applications open until Q4 2026
§ 01 — What this is, and what it is not

Not a pilot. Not a paid pilot. An exchange.

A design partnership with us is a structured, twelve-month exchange. We trade access to the model line, the research team and our roadmap influence — for access to your domain, your data and your operators' time. Cash does not need to change hands.

If we sell you a model in 2026, you will be disappointed. If we build a model with you across 2026 and 2027, you will be ahead of every competitor when 2028 arrives.

— from the design-partner brief deck

What we offer

  • — Direct access to the founder and the research team. No account managers. No salespeople.
  • — A seat at the design table for BULUT-1's vertical-specific evaluation harness.
  • — First access to BULUT-1 Mini under research licence when it ships.
  • — Co-authorship on the public technical report we will publish about the partnership, if you want it.
  • — Right of first refusal on the equity round we eventually raise, with terms that recognise the partnership predates it.

What we ask

  • — Permission to capture multimodal data on a representative slice of your operations.
  • — A small, named technical contact who can answer four questions a month.
  • — Honest feedback. We need to hear what does not work.
  • — Twelve months of patience. We will not have a polished product on day one.
§ 02 — Where we are recruiting

Four verticals. One anchor partner each, ideally.

A short brief on each vertical and the kind of operator we are hoping to talk to. Not exhaustive — adjacent operators welcome to write in.
— 01 · Industrial monitoring
manufacturing · ports · energy

The vertical we expect to find our first anchor in.

Continuous video understanding of factory floors, ports and energy infrastructure. Anomaly detection, predictive maintenance, OSHA-grade incident reconstruction with full causal chain of attention. The discipline of the operator who runs this kind of system is a perfect match for ours.

Ideal partner: a mid-size industrial operator running 5–50 sites with existing camera infrastructure and a small data-science team.

1
anchor slot
12 mo
term
Q3 2026
target start
— 02 · Autonomous mobility
trucking · last-mile · simulation

For fleet operators who want a foundation backbone.

Closed-loop simulation, edge-case mining and behavioural prediction for L4 trucking and last-mile robotics. We do not build vehicles — we believe the stacks that go into them deserve smarter eyes than a per-task CNN ensemble. The physics of long-tail edge cases is exactly where foundation models earn their keep.

Ideal partner: a regional fleet or trucking operator with at least one year of vehicle-side telemetry and an in-house ML team.

1
anchor slot
12 mo
term
Q1 2027
target start
— 03 · Earth observation & climate
satellite · drone · ground-station

Where multimodal joint reasoning is obviously the unlock.

Joint training across satellite, drone and ground-station data to nowcast wildfire spread, urban flooding and methane plumes at sub-kilometer resolution. The bottleneck for ground-truth-augmented EO models has always been ground data — we are building it. Civil-protection agencies and EO research groups especially welcome.

Ideal partner: a climate or civil-protection programme with EO data access and a willingness to co-publish.

1–2
anchor slots
12 mo
term
Q2 2027
target start
— 04 · Robot foundation policies
humanoid · quadruped · arm

The most ambitious vertical. Hardest to bring online.

Generalist manipulation and locomotion policies fine-tuned from BULUT for humanoid, quadruped and arm-robot OEMs. We will need a hardware partner who tolerates an early-stage foundation backbone and can co-develop the action-codification we need. Likely the last vertical we activate.

Ideal partner: a robotics OEM with a working hardware platform and a small but skilled embodied-AI team.

1
anchor slot
12 mo
term
2028
target start
§ 03 — Who qualifies

A short list. We will be unsentimental.

We will say no to most operators who write in. The criteria below are how we will choose the four to six. None of them are about company size.
CriterionWhy it mattersHow we will assess
Operates real physical infrastructureIf your business does not depend on the physical world running, we are not the right fit.first call · 30 min
Has a small, sharp technical teamWe work with engineers, not procurement. The technical contact is the partnership.technical conversation
Comfortable with research-velocity timelinesTwelve months is the minimum useful horizon. Demo-by-Friday energy will burn out fast.honest expectation-setting
Open to data sharing under provenance disciplineWe can only co-build if we can train on your domain data — under our consent ledger.legal review jointly
Willing to co-publish if appropriateWe will not announce shipping. We may want to publish what we learnt — with you.case-by-case
Mission-alignedWe will not partner with operators whose use of physical AI we would be ashamed of.conscience-based · two-way veto
§ 04 — How it works

Five conversations. Sixty days from email to MoU.

A deliberately slow process. We are choosing partners for a year of close work — there is no rush. Each step is mostly with the founder.
DAY 0–7 · INTRO

30 minutes. Two-way.

You email partners@bulutlabs.com with a paragraph on what you operate and why physical AI matters in your context. We reply within 48 hours. We schedule a 30-minute call. Either side can say no without explanation.

DAY 7–21 · TECHNICAL FIT

Two technical conversations.

One with your technical lead, one with our founder. We discuss what data you can share, what evaluation harness would matter to you, and what your team's research velocity looks like. We are honest about what BULUT-0 can and cannot do today.

DAY 21–35 · LEGAL DIAGNOSTIC

The provenance conversation.

We share our consent ledger architecture and walk through how your data would be ingested, redacted, trained on and audited. Most partnerships fall apart here, which is good. The ones that survive this are durable.

DAY 35–60 · MOU

Memorandum of understanding.

We sign a short, plain-English MoU outlining scope, data access, evaluation milestones, mutual rights and exit clauses. No cash changes hands unless your team's time costs you more than data-flow plumbing.

DAY 60+ · BUILDING

Twelve months of actual work.

Quarterly reviews with you. Weekly stand-ups with your technical lead. Open Slack with our research team. Honest mid-year retrospective. Renewal conversation in month 11. Public co-published learnings if mutually useful.

§ 05 — Who we are not talking to

The honest list of nos.

We will say no to certain partner categories regardless of fit. We list them below so we do not waste anyone's time, ours included.
— NO

Defence-end-use

We do not partner with operators whose primary use case is offensive military or surveillance.

— NO

Mass-surveillance public

Even with consent ledgers, we will not train on populations that did not meaningfully agree.

— NO

Acquisition exits

If your goal is to acqui-hire us before BULUT-1 ships, we are not the right fit. We are a five-year-patient company.

— NO

"AI strategy" consulting

We do not run AI strategy workshops. We build models. We are happy to recommend people who do consulting.

— NO

Resale partnerships

We do not white-label. We do not embed silently. Our partners' presence in our customer list is part of how we earn trust.

— NO

Anything we cannot defend in public

If we would be embarrassed to put the partnership on the website in a year, we should not start it now.

Want to build the next era together?

Email a paragraph on what you operate and why physical AI matters to you. Replies in 48 hours, in three languages, from the founder.