Locked narrative (July 2026). Source of truth for deck copy. Public slides and PDF must match this outline. Claims stay inside Evidence and Methodology.
Public shareable version (no auth): haga.mushoodhanif.com/deck — branded slides with Download PDF (/haga-pitch-deck.pdf). Keep this MDX, the public deck, and the PDF aligned when content changes.
Plain-language rule: If a term needs explaining on the slide, delete it or replace with English. Methods detail lives on Lab / /methodology, not in the skim path.
Cut from public core: Expansion-as-table, Roadmap-as-table dump, Hub71 (meeting appendix only), and jargon-first framing (MuJoCo, CoTracker, Wilson CIs, OSC, static_hover, “Pillar 1/2” as slide titles).
Order rule (post–Slidebean review): story first (Problem → Why now → Solution → Evidence); Team before Ask (solo-founder honesty without leading with pedigree gaps). Technical evidence ≠ revenue traction — never invent LOIs/ARR.
Public core — 12 slides
1. Cover
Haga
Independent verification for physical AI.
One plain sentence: We stress-test robot policies and AI-generated worlds in simulation — then report what breaks, privately.
Tagline: Stress-test. Break. Report.
Buyer anchor (small): Private evaluations for robotics labs and sim teams.
Living prep · July 2026
2. Problem
Headline: Simulated success does not survive the real world.
Only big numbers on this slide:
- 89.4% — RLBench sim manipulation success
- 12% — Stanford HAI AI Index 2026 real household-task success
Visual job: Sim looks good → different contexts → real-world failure (three nodes). Label benchmarks on the nodes.
Supporting lines (short):
- Labs grade themselves. Generated worlds ship without an independent physics check.
- Heavy real-robot leaderboards exist at the top — small teams still lack a fast, private pre-check in simulation.
Footnote: Urgency signal from two published contexts (not one measured cliff). Do not say “fell from 89.4% to 12%.”
Speaker note: Lead with the Stanford gap as motivation. Market size is later. Do not invent dollars-of-delay without a cited source.
3. Why now
Headline: Physical AI is flooding with capital; independent verification is underserved.
Three drivers only (no table dump):
- Capital — $27.6B physical AI / robotics funding across 1,009 deals in 2025 (PitchBook).
- World models — generative environments are becoming training fuel; physics mistakes compound downstream.
- Standards — collaborative robot applications need documented validation (ANSI/A3 R15.06-2025). Better sim tools raise fidelity — they do not independently audit outputs.
Category signal (one line, optional): verification and sim tooling is already funded and moving (Patronus, Instance, Robocurve, Antioch, Bifrost).
Speaker note: TAM/SAM/SOM live on the Market slide — do not overload Why now with bottoms-up math.
4. Solution
Headline: Independent failure testing in simulation.
Visual job: Artifact in → stress-test → pass/fail report out (two inputs, one method).
| You give us… | We do… | You get… |
|---|---|---|
| A trained robot policy | Push until physics or task success fails | Pass/fail vs clear thresholds + failure cases |
| AI-generated video / worlds | Check whether physics stays consistent | Physics-consistency flags + score |
| (Next) a release pipeline | Score every version | Ongoing plausibility signal |
Public vs product: Methodology and aggregate findings are public. The product is private evaluation engagements.
5. Evidence
Headline: Pre-revenue. Technical proof under stress.
Kicker: Evidence — not “Traction.” Until paid engagements exist, say technical evidence.
Verdict (hero line): Policies that look perfect on paper collapse under stress.
Three degradation curves (hero numbers — not a spreadsheet):
| Task | Easy conditions → hard conditions |
|---|---|
| Lift | 1.00 → 0.26 |
| Stack | 0.96 → 0.20 |
| Pick & place | 1.00 → 0.24 |
World-model one-liner: Held-out CogVideoX check flags generated-video failures that real footage does not trigger (public Lab).
Footnote (required): No paid engagements or LOIs yet — we do not invent them. Sim-first · scripted baselines today · generative held-out still narrow · not a substitute for institutional real-robot leaderboards. Charts and methods: Lab · /methodology.
Door and full tables stay in Evidence — not on the skim slide.
6. Why Haga wins
Headline: Why Haga wins
Three edges — one sentence each:
- Private sim-first pre-check — Catch physics failures before expensive hardware evals and public leaderboards.
- One engine for policies and worlds — Same adversarial stress method across robot policies and AI-generated environments.
- Published numbers, private product — Public methodology builds trust; the sale is a confidential report with curves and disclosed weak results.
Footnote: Proof today is public Lab cliffs and held-out physics checks — not customer ROI claims we have not earned.
5-second answers this slide must unlock:
- What does Haga do? → Independent sim stress-tests that report what breaks.
- Edge vs RoboArena / Robocurve / Instance? → Fast private pre-check spanning policies and generated worlds — not a public leaderboard, not hardware-only, not video-only.
- Are results good enough to care? → Perfect-looking policies fall to ~0.2–0.26 under stress; that is the product proof.
7. Competition
Headline: Everyone else covers one lane. Haga sits in the empty wedge.
Axes (state explicitly): public ↔ private · policy ↔ world / data.
Map zones (not a five-row table):
- Public leaderboards — RoboArena and peers: institutional, real-robot, late-stage trust.
- Hardware eval — Robocurve: closest policy competitor; real robots.
- Sim / data platforms — Antioch, Bifrost, training-loop tools: build or train; they do not independently audit.
- Video physics only — Instance: closest world-model sibling; not policies.
Haga’s wedge: Fast, private, sim-first physics verification for robot policies and generative world-model outputs.
Full map: Competitors.
8. Product path
Headline: Private reports now. Continuous scoring next. Comparative stress data as the moat.
| Stage | When (from wire) | Measurable gate |
|---|---|---|
| Now | ≤4 mo | 2 live design-partner evaluation reports delivered |
| Next | ≤6–9 mo | World-model cohort n≥30 held-out · continuous-scoring design · first ML/eval hire |
| Moat | 2027+ | Comparative stress dataset across policies + world models |
Do not inflate the bottoms-up verification market into the full physical-AI capital stack on this slide.
9. Go-to-market
Headline: B2B distribution into labs and sim teams, not consumers.
Channel: robotics / physical-AI labs, simulator teams, and world-model research groups that already run internal evaluation. Haga sits inside their existing workflow, not in front of it.
Funnel (three steps only):
- Awareness — Public Lab + methodology → technical credibility.
- Intake — Account-based outreach + 48-hour path into a scoped private evaluation.
- Paid report — Design partners validate pricing; LOIs only when real.
Footnote: Continuous scoring is product expansion after WTP — not a near-term sales channel.
Speaker note: No LOIs yet. Intake → scoped report workflow is built; Priority A outreach is live. Do not invent logos, ARR, or customer counts. Status detail: GTM · Design-partner pipeline.
10. Market
Headline: Bottoms-up verification spend — not the full physical-AI capital stack.
Slide heroes (lead with bottoms-up):
- TAM ~$39M — ~1,080 buyers × mature recurring verification ACV
- SAM ~$8M — ~430 reachable ICP accounts · near-term blended ACV
- Y3 SOM ~$0.8M — capacity-constrained (25–35 won accounts)
Backdrop line: $27.6B physical-AI / robotics capital in 2025 (PitchBook) — Haga sells into that stack as a thin paid verification layer.
Supporting line: Working hypotheses for diligence — not booked ARR. Full equations and ASP ranges: TAM–SAM–SOM.
Speaker note: Do not quote robot-software CAGRs as Haga’s TAM. Do not claim $27.6B as TAM. Do not publish a public rate card — ASP is modeling input only.
11. Team
Mushood Hanif — Founder
Product, evaluation systems, full-stack. Built Haga end-to-end: stress-test harness, public Lab evidence, methodology, and intake → report product.
Concrete (facts only — no invented pedigree):
- Shipped: policy stress cliffs (e.g. Lift 1.00→0.26) + held-out CogVideoX physics check — public Lab.
- Gate A equity / IP side letter signed. Delaware C-Corp via Atlas from a runway-sized check.
- Solo by design today — not multi-founder optics. First hire: ML / robotics-eval engineer (written scorecard); advisors none formal yet.
Detail: Bios · Hire scorecard.
12. The ask
Headline (locked): Raising $1.0M to earn the right to raise a Series A
Secondary path: For program-sized bridge checks or residencies, use the Bridge raise deck — first-check band only; do not place a $1.0M ask on a micro-check slide.
What $1M buys (outcomes — never “founder runway” as the title):
- 2 live design-partner evaluation reports (≤4 mo)
- World-model scoring at depth (n≥30 held-out ≤6 mo)
- Continuous-scoring product design (≤6 mo)
- First ML/eval hire (≤9 mo)
Use of funds (on-slide split): 40% team · 25% compute · 12% product · 12% GTM · 11% legal + buffer — Use of funds.
Secondary line only (never the title): Pre-seed SAFE · ~12 months operating horizon · allocation maps to outcomes — not founder runway.
Contact: mohdmushood@yahoo.com · haga.mushoodhanif.com · linkedin.com/in/mushood-hanif
5-second answer this slide must unlock: What does $1M buy that unlocks the next VC round? → Delivered partner reports, world-model depth, platform design, and a second builder.
Soft-raise hygiene (dataroom — not a public slide)
Soft raise / reachout discipline: active cold outreach is paused. The $1.0M pre-seed pack is the institutional target only, not active outbound. Bridge and warm program replies are the current path until an evidence trigger lands live partner report, grant award, substantive reply, or multi-model held-out).
- Gate A — equity + IP signed — Founder equity & IP (Ready 2026-07-17)
- Evidence pack Ready — landed
- Technical distribution live — landed; Priority A outreach unlocked
- Demand signal (async): ≥10 ICP conversations or ≥3 serious eval/LOI threads — strengthens pitch; does not block meetings — Pipeline
- Cold outreach pause: no cold capital outreach until evidence triggers — warm replies / program inboxes only until then
- Delaware C-Corp (Stripe Atlas) from runway-sized check → SAFE / wire → Investor sharing
Bridge track (separate): micro-checks / residencies use the Bridge raise deck — do not put the $750k–$1M band on a $1k–$10k First Check slide.
Risks: Appendix.
Meeting appendix — Hub71 & Abu Dhabi
Not a default public or dataroom slide. Share only in Hub71 / MENA-focused conversations. See Disclosure policy.
Commitment: Sole founder relocates long-term to Abu Dhabi; HQ and first hires in ADGM.
12-month plan in UAE:
- Partner compute for held-out multi-model eval cohorts (world-model scale-up)
- Engage SAVI / mobility corporates for sim-first verification pilots — no invented LOIs
- Hire first ML/robotics eval engineer in Abu Dhabi; deliver 2 paid design-partner reports
Capital path: Hub71 SAFE as bridge; MENA industrial + physical-AI ecosystem vs. building alone from Pakistan.