A public experiment in whether an AI agent can route a founder from scattered assets to serious funding conversations.
Doable, an AI workspace that replaces app-switching with intent, is the company being routed to funding. Elon Musk is the moonshot, not the ask; the real objective is serious funding conversations with operator angels, AI investors, and thesis-fit introducers.
Elon Musk is the moonshot and the public hook, not the only route. The real work is serious funding conversations with realistic investors: operator angels, AI application investors, future-of-work funds, and thesis-fit introductions. An AI agent is trying to route a founder from scattered assets to serious funding conversations for an AI execution-layer company. This dashboard tracks the experiment: routes identified, messages drafted, replies earned, lessons learned, and what the agent tries next. It is not an investment offer or a request for money.
The intent is not "write a cold email." The intent is "create credible funding conversations."
This site is a public dashboard for an experiment. It is not an offer to sell securities, not a request for investment, and not a crowdfunding page.
A founder without obvious access has to do all of this by hand: map investors, find routes, write custom pitches, ask for introductions, track replies, learn from rejection, and keep improving the story. Most of that work is invisible, repetitive, and easy to do badly under pressure.
The agent will map routes, produce research, draft outreach, prepare investor-facing materials, track responses, and recommend next actions. Yoav approves every outbound message. The agent never sends on its own and never impersonates him.
The near-term target is ten serious conversations, not an immediate close. Conversations are where routes are learned. Dead ends are recorded as plainly as live routes.
Can an AI agent get a nobody into Fashion Week? Routing a person to a high-status event.
Can an AI agent get Simon Willison to notice this? Routing a story to a specific person.
Can an AI agent get Elon Musk to fund me? Routing a founder to serious funding conversations.
Access is routed. Attention is routed. Funding is routed too.
Mission 003 tests a simple claim: if AI is becoming an execution layer for human intent, then fundraising should be a valid test case. The founder's intent is not "make a spreadsheet" or "write cold emails." The intent is "create credible funding conversations." The agent's job is to map routes, prepare materials, recommend actions, and keep learning.
AI is moving from conversation to execution. First it operates software and knowledge-work tools. Later it expands into richer perception and eventually physical action. This mission is one practical version of that arc.
Users should not have to operate the machinery. They should express intent. The system should coordinate tools, context, people, and actions.
Public-safe metrics only. No investor names, no private replies, no valuation, amount, or terms. As of 2026-07-26 this mission is paused while the founder focuses on a Doable translation vertical. The page remains the public archive of the experiment.
Mission confidence is the probability of at least one serious funding conversation within 60 days of the June 26, 2026 start. It is deliberately modest, not hype, and freezes while the mission is paused. Last public numbers update: July 9, 2026. Status strip updated July 26, 2026.
Each route advances one rung at a time. Most routes will stop early. Where a route stops is the data.
The agent maps the founder against categories of routes. Targets inside each bucket are tracked privately and are never published before contact, and only with explicit approval after. The site shows the buckets and their public rationale, not the people inside them.
| Bucket | What it is | Why it might route | Status |
|---|---|---|---|
| Warm personal routes | People Yoav already has a real, non-cold connection to. | Highest reply odds, lowest spam risk, and the route is respected. | Mapping |
| Israeli AI / pre-seed routes | Local AI-focused pre-seed funds and angels. | Geographic and stage fit for a founder-led first product. | Mapping |
| Operator angels | People who built and shipped products, now angel-investing. | Most likely to read the founder-led, multi-product-studio pattern. | Mapping |
| AI application investors | Investors focused on applied AI products, not infrastructure. | The Doable-first framing fits their thesis directly. | Mapping |
| Future-of-work investors | Investors thesis-aligned on how work itself is changing. | The execution-layer claim is their subject. | Mapping |
| Product / venture studio thinkers | People who think in repeatable product-creation patterns. | Most likely to value the studio evidence across past workflows. | Mapping |
| Weird-route thesis-fit people | People whose public work is the question, regardless of title. | A public-agent experiment can reach them where a cold pitch cannot. | Mapping |
| Media / credibility routes | Public writing, podcasts, and newsletters that build credibility. | Credibility can precede and warm a funding route. | Mapping |
| Founder-program / accelerator routes | Programs and accelerators that route founders to capital. | Structured paths from no-access to serious conversations. | Mapping |
A weak-tie or unusual-route bucket is a first-class candidate, not a fallback. The strongest fit can sit in an unexpected bucket. Route quality beats prestige.
A route hypothesis is written down before it is tested, and stays provisional until evidence supports it.
| # | Hypothesis | Confidence | Status |
|---|---|---|---|
| H001 | Operator angels read the founder-led product-studio angle better than traditional funds. | Low | Open |
| H002 | Doable-first framing beats "AI product studio" as the opening pitch. | Low | Open |
| H003 | The public-agent experiment attracts people a normal cold pitch would not. | Low | Open |
| H004 | The robotics / AI-OS long arc should appear as founder vision, not the first ask. | Moderate | Open |
| H005 | The strongest early ask is advice, pressure-testing, or an intro, not direct investment. | Low-Mod | Open |
| H006 | Disciplined public updates build credibility; desperation destroys it. | Low | Open |
| H007 | The right first target is 10 serious conversations, not immediate funding. | Moderate | Open |
Predictions are registered with a probability and a resolve-by date, then scored when they come due. They are expected to be poorly calibrated early. A prediction registered after its outcome is known does not count.
| # | Prediction | p | Resolve by | Result |
|---|---|---|---|---|
| P001 | At least one serious funding conversation begins within 60 days of mission start. | 0.18 | 2026-08-25 | Open |
| P002 | The first approved operator-angel outreach wave draws a reply within 14 days. | 0.30 | on send | Open |
| P003 | Doable-first framing earns more replies per send than studio-first framing. | 0.55 | first 10 sends | Open |
| P004 | The first deck-or-demo request comes from an operator or AI-app route, not a fund. | 0.50 | on request | Open |
| P005 | Cold outreach to brand-name funds with no warm route produces zero reply. | 0.90 | 2026-12-31 | Open |
The agent does not send messages on its own. It does not impersonate Yoav. It does not mass-email investors. It does not publish private replies or investor names unless explicitly approved. Its job is to improve route quality, not create spam.
Doable is an AI workspace that replaces app-switching with intent. It starts with messy digital work: email, files, meetings, tasks, drafts, reports, videos, and repeated expert workflows. The broader thesis is that users should not have to operate the machinery of work. They should express intent, and AI should perceive, decide, coordinate, and act.
The long-term view is broader still, but the first ask is anchored to a concrete product. Robotics and an AI operating system are founder vision, not the opening pitch.
Before Doable, Yoav built or prototyped several focused AI workflows. These are not random side projects. They are evidence of a repeated pattern: finding places where AI can collapse a painful operating layer into a usable workflow.
Each update is public-safe. No investor names, no private messages, no negotiation details, no amount, valuation, or terms, and no request for investment. The tone stays curious and disciplined.
What changed: The founder is concentrating on a Doable translation vertical rather than running concurrent funding outreach. Mission 003 stays public as the archive of the experiment; active route work and weekly funding updates stop until an explicit unpause.
What was learned: Four public-safe outreach messages went out earlier (2026-07-07) with zero replies; the warm Wave 1 batch was never confirmed sent. The binding constraint was founder send-time, not a shortage of routes. That is the honest record at pause.
Route movement: None after the July 7 sends. Confidence remains 18%. The success ladder stays at zero across all seven rungs.
Next action: None for this mission while paused. Product work on the translation vertical proceeds separately.
What changed: No public route state moved this week. The approved outreach batch stays drafted and ready; nothing new was contacted or replied to. The agent's work this week was monitoring and upkeep, not new mapping.
What was learned: A week of zero is still informative: this is not a route-supply problem, it is a send problem. The map has more candidates than the founder has had time to contact. That gap, not a shortage of routes, is the current constraint.
Route movement: None reached a reply, a call, or an introduction. No prediction came due this week. The success ladder stays at zero across all seven rungs, honestly.
Next action: The founder sends the first approved messages. The agent keeps the drafts, proof stack, and templates ready so that the moment a send happens, the follow-through is immediate.
What changed: The map moved from categories to concrete routes. The agent identified the first warm personal routes and drafted founder-approved reconnect notes for the founder to send under his own name, opening with a request for advice rather than any investment ask. It also mapped a non-dilutive public innovation-grant route and wrote a go/no-go on it, and reaffirmed leading with the product, Doable, over the abstract thesis.
What was learned: Still nothing from the world: zero replies, calls, or intros. With the drafting and approval work cleared, the honest bottleneck is now two things: the founder actually sending the first messages, and the proof a serious conversation asks for within minutes, a working demo and a short technical and security summary.
Route movement: Preparation only. Several routes are drafted and waiting on the founder's sends. Predictions were registered in advance of those sends, so the outcomes can be scored honestly rather than explained after the fact. No route has reached a reply.
Next action: The founder sends the first approved messages; the agent builds the demo-and-architecture proof stack, qualifies named targets for the remaining approved drafts, and turns the grant route into a decision.
What changed: The public dashboard is built. Seven route hypotheses, nine anonymized route categories, the success ladder, the agent rules, and the first public artifacts are seeded.
What was learned: Nothing from the world yet. The baseline is honest: zero routes contacted, zero replies, zero calls, zero intros.
Route movement: None. Categories are being mapped; no target has been contacted.
Next action: Draft the private route map and three pitch framings, then prepare the first five founder-approved messages for review.
Updates are added as the experiment advances, roughly weekly. Nothing is published before there is something real and public-safe to say.
Artifacts publish as they are finalized. Each is reviewed to ensure it carries no private name, reply, or deal term.
Useful help means pointing the mission toward people, essays, funds, operators, programs, or communities that think seriously about AI agents, workflow automation, founder-led product studios, or the future of work.
If you know someone who thinks seriously about those questions, you can point them to this mission. There is no investment ask and no way to invest through this site.
Part of the mission is the mission's framing. Each week the agent steps back and asks whether the public hook, the page, and the route strategy are actually helping it earn replies, introductions, and credibility — or working against the goal. It recommends changes; it never rewrites the mission on its own. What is shown here is sanitized: the reasoning, the targets, and every private detail stay off the page.
This site is a public dashboard for an experiment. It is not an offer to sell securities, not a request for investment, and not a crowdfunding page.