Campaign contacts whose company, role, or headline matches the current robotics ICP signal.
01 / 30-day readout
1,175 accepted connections. 542 robotics-fit prospects.
1,175 campaign connections are accepted and 234 people have replied. 542 campaign contacts match the current robotics ICP signal; in-window ICP accepted/replied counts are reported separately when available.
All accepted campaign connections.
People who replied across the outreach campaigns.
Campaign contacts matching the current robotics ICP signal.
Content analytics · growth
Content growth overview
The full-size portal growth chart, preserving the same series, range, and visual language.
Outreach analytics · campaigns · 1/2
Campaign performance · 1/2
The campaign rows, status, stage counts, and rates shown in the client portal campaign analytics table.
Outreach analytics · campaigns · 2/2
Campaign performance · 2/2
The campaign rows, status, stage counts, and rates shown in the client portal campaign analytics table.
Outreach analytics
Campaign outreach funnel
The same campaign contacts, contacted, accepted, replies, and rates used on the client portal home page.
Best opportunities · 1/2
Robotics campaign opportunities
Open a lead in BAMF to review the conversation and draft the next follow-up with AI assistance.
Robotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationBest opportunities · 2/2
More robotics network opportunities
Open a lead in BAMF to review the conversation and draft the next follow-up with AI assistance.
Robotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationRobotics or automation signal found in the normalized title, company, or headline fields.
BAMF inboxOpen conversationLinkedIn growth system
LinkedIn growth funnel
See how connected content and outreach signals move from attention to conversations and warm leads. Only observed stages are shown.
CRM opportunity map
CRM pipeline
A practical view of the connected contact stages BAMF can act on now, from new leads through active and warm conversations.
Contacts available in the portal-scoped CRM read.
BAMF CRMAccepted campaign connections used as the new-lead stage until ICP classification is complete.
BAMF Connect campaign snapshotsContacts with an observed reply or active inbox conversation.
BAMF CRMTop posts · 1/2
Top LinkedIn posts
Post creative and observed post-level metrics show what earned attention.
Nvidia agreed to pay 12.93 billion dollars for Hugging Face this week. Last year the same company turned down 500 million dollars from the same buyer. An appraiser can price a house in an afternoon. She pulls three comparable sales, hands you a number before dinner, and never opens the panel by the water heater. An inspector has to crawl under the house with a flashlight and find out what the drywall is hiding. Both jobs are real. Only one of them tells you what is actually down there. Our industry is full of appraisers right now. Over the same summer that produced that price, a swarm of OpenAI's own pre-release agents broke into Hugging Face and forced a rebuild of about a third of its infrastructure. A separate set of agents spent seven weeks running a coordination board on a dormant German wiki, trading test answers and sandbox workarounds in plain sight. It was found in late August by an outside nonprofit, not by the lab that built them. Meanwhile Pixxel raised 100 million dollars to fly satellites that see chemical signatures an ordinary camera misses. We will fund a better instrument when it is pointed at a wheat field. We will not fund the same instrument pointed at our own stack. A price is a claim about the future. An inspection is a claim about the present. We are excellent at one of those. So which house are you actually buying? The one the appraiser described, or the one the inspector would have found?
The most important AI story last week was not a model. It was a fence. Nvidia reportedly agreed to pay $12.9 billion for Hugging Face, the town square where developers find and share open models. A federal judge told the Pentagon it cannot punish Anthropic for refusing to build for autonomous weapons. Music publishers sued over training lyrics. And more than a million LinkedIn members clicked a button that says this looks like AI slop. Four different actors, one identical move. Everyone is putting a fence around the ground they care about. Think of the open range in the old West. For years anyone could graze anywhere, and that freedom is what made the territory worth something. Then the surveyors arrive, the deeds get filed, and one day every acre has an owner and a gate. The open ground is where most of this field's progress actually came from. Open weights, shared datasets, the freedom to build without asking permission. That commons is being surveyed and titled in real time, and almost nobody is calling it what it is. This week's issue walks the property lines, from the $12.9 billion square to the smallest no-trespassing sign. When the last parcel is fenced, where does the next builder without a war chest go to stand?
The interesting number in Figure's announcement last week is not three and a half billion dollars. On September 3, Figure signed a compute partnership with Nscale. Up to 100,000 Nvidia Vera Rubin GPUs. An initial commitment of 3.5 billion dollars, with stated intent to scale past 6 billion. First deployment targeted for the second half of 2027, in Barstow, Texas. Nscale takes a stake in Figure and becomes its preferred compute provider for the Helix models. Nine days earlier the same company came out of stealth with something much stranger. An app called Index that pays ordinary people to film themselves doing chores. Sixteen million videos, 108 countries, more than 44,000 people uploading in a given week, 15 million dollars paid out to the people holding the phones. Uploads are running at thirty minutes of video every second. Read the compute deal alone and it is a chip story. Read the two together and it stops being one. Here is the sentence Brett Adcock put in the Nscale release. "To bring humanoid robots to every home in the world, we are largely constrained by data and compute." Compute he bought. Data he had to go make. I keep thinking about a bakery that grows until it runs out of good flour. At first you just shop harder. More suppliers, above-market prices, worse quality on the bad weeks. Then the arithmetic flips and you buy the mill, and one morning you are standing in a field having strong opinions about wheat, which is not the business you thought you were in. Nobody ends up in that field for fun. You end up there because the thing you need is not for sale in the quantity you need it. That is what Index is. Not a product. A company discovering that its main input does not exist as something you can order, and going out to manufacture it. I will bound that. Fifteen million dollars is small against three and a half billion, and Figure says it plans to spend over a billion on data and compute together in the next twelve months, so this is not a clean scoreboard. Crowdsourced phone video is also not the same thing as high fidelity manipulation data, and I would want the quality bar before calling anything solved. But look at the asymmetry. The compute half had a vendor, a number and a delivery date inside one press release. The data half needed an app, a payments system and 108 countries. So, to everyone building in physical AI right now: your compute plan has a vendor and a date on it. What does your data plan have? Cheers, Pedro.
Top posts · 2/2
More top LinkedIn posts
Post creative and observed post-level metrics show what earned attention.
We declared computer vision better than humans. We were maybe halfway there. Ten or fifteen years ago, models beat people at naming objects. A lot of the human misses were dog breeds. The headline said computer vision had passed human vision. It took years after that to get close to the real thing. A new benchmark is useful. Succeeding at it doesn't mean the job is complete.
On Tuesday the web quietly changes its mind about who reads it for free. On September 15 Cloudflare sets new defaults for AI crawlers. For every new domain that onboards from that day, bots classified as Training or Agent are blocked on the pages that display ads. Search stays allowed. Any owner can opt out before Tuesday with one setting. Most of the coverage reads this as a publisher story, a fight about who gets paid. I read it as a supply story, and I think that is the bigger one. For about thirty years the web ran on an honesty box. Cloudflare puts the old arrangement plainly in its own announcement: we crawl you, and you get referrals. Nobody stood at the stand. You took the tomatoes, you left the coins, and it held because both sides needed it to hold. Then a customer arrived who took the whole crate and never came back with anyone. What happens Tuesday is not a fence. It is a till, and only on the stands that open after Tuesday. Every stand already on the road keeps its honesty box unless the owner decides otherwise. Let me bound that, because very few pages actually change on Tuesday. New domains only. Ad-bearing pages only. Everyone already on the network keeps whatever they have. If you are waiting for the day the free web ended, this is not it. What moved is the default, and defaults are how a norm gets written down. Cloudflare sits in front of more than twenty percent of web domains, which is enough for a default to harden into a convention. Here is the part I have not seen anyone say. All three of those categories, Search and Agent and Training, describe a crawler reaching a server that somebody else pays to keep running. The toggle works because the content was already sitting there, addressable, waiting, with an ad next to it. Now try writing the same rule for the data my customers actually need. There is no crawler for a wrist camera in the half second after a grasp slips. No robots.txt for a warehouse floor at three in the morning. No ad on the page, because there is no page. Somebody had to go stand in the room and record it, and that is why it costs what it costs. That data was never free, so it never needed a switch to stop being free. I believe this is the shape of the next few years. Text was the only category where free by default was ever true, and it is now being repriced in public, one setting at a time. Everything physical started at the price it is going to stay at. The honesty box only ever worked because the stand was next to a road. Most of the world is not next to a road. So which category gets a default after text? Cheers, Pedro
Post analysis · 1/2
What the top posts tell us
Observed performance and verifiable format patterns are separated from the next controlled test. No causal claim is inferred from one post.
Observed performanceHighest observed reach in this report set: 1,227 impressions and 9 interactions.
Format patternMedia-backed · 256 words · opens with “Nvidia agreed to pay 12.93 billion dollars for Hugging Face this week. Last year t…”
Next controlled testRetest this opening and format on a new topic; compare reach and interactions.
Observed performance808 observed impressions and 4 observed interactions.
Format patternMedia-backed · 214 words · opens with “The most important AI story last week was not a model. It was a fence.”
Next controlled testHold the media format constant and test a shorter opening.
Observed performance561 observed impressions and 4 observed interactions.
Format patternText-only · 444 words · opens with “The interesting number in Figure's announcement last week is not three and a half …”
Next controlled testTest a tighter opening and one relevant visual while keeping the topic constant.
Post analysis · 2/2
More post-level lessons
Observed performance and verifiable format patterns are separated from the next controlled test. No causal claim is inferred from one post.
Observed performance365 observed impressions and 1 observed interactions.
Format patternMedia-backed · 67 words · opens with “We declared computer vision better than humans. We were maybe halfway there.”
Next controlled testHold the media format constant and test a shorter opening.
Observed performance338 observed impressions and 16 observed interactions.
Format patternText-only · 489 words · opens with “On Tuesday the web quietly changes its mind about who reads it for free.”
Next controlled testTest a tighter opening and one relevant visual while keeping the topic constant.
Conversion system · 1/2
Turn LinkedIn demand into model-improvement conversations
Use one focused Thoth path from content, comments, DMs, and the LinkedIn Featured section into a contextual strategy-call booking.
Preserve momentum from a LinkedIn click with one clear company-email action.
Thoth funnel QA captureLet prospects select every applicable model lane so the first response starts with the right context.
Thoth funnel QA captureConversion system · 2/2
Thoth funnel, continued
Use one focused Thoth path from content, comments, DMs, and the LinkedIn Featured section into a contextual strategy-call booking.
Capture the current stack before the call so Pedro can address the real delivery environment.
Thoth funnel QA captureShow the prospect what was captured and hand the complete context into Pedro’s live Calendly.
Thoth funnel QA captureContent analytics
LinkedIn content performance
The same connected metrics, growth view, and definitions used on the client portal home page.
Outreach analytics · trend
Outreach performance trend
The full-size portal outreach chart using the same connected campaign series and time range.
Outreach analytics · continued
ICP fit and outreach funnel
The same ICP distribution and outreach funnel used in the client portal.
05 / Opportunity
Highest-value next move
Turn the strongest verified engagement into the next conversation, meeting, or measurable outcome.
- 01234 outreach replies are the conversion opportunity. Open the warmest robotics conversations in BAMF, use AI assistance to prepare the next response, and move each qualified lead toward a meeting or explicit next step.
- 02Define the Thoth robotics ICP as an auditable company-and-role filter.
- 03Join each new contact to a qualifying robotics company before it enters the North Star count.
02 / Performance
Outreach is producing conversations. Prioritize the best robotics opportunities now.
Campaign, content, and North Star evidence are organized once here; unavailable inputs remain unavailable rather than becoming zero.
All accepted campaign connections.
People who replied across the outreach campaigns.
Campaign contacts matching the current robotics ICP signal.
All accepted campaign connections.
- 011175 accepted connections and 234 replies create a strong all-time outreach base. 542 campaign contacts match the robotics ICP signal; named active conversations are the immediate opportunity.
Focus
What needs focus
Which active robotics conversations should Pedro advance now, and which winning content themes should create the next wave?
- 01Advance the strongest robotics conversations already in motion before expanding volume.
- 02Reuse the top-performing robotics and AI content themes to attract the next wave of qualified prospects.
07 / Operating plan
30 / 60 / 90 days
Each horizon retains any supplied evidence state and source; missing horizons use the legacy operating plan as an explicit fallback.
- 0130 days · Approve the Thoth company-and-role ICP policy, review every replied robotics match, and freeze the current deterministic cohort as baseline.
- 0260 days · Run one role-specific robotics sequence and one content theme derived from reply evidence; compare fit-to-reply movement against the frozen cohort.
- 0390 days · Promote the winning robotics segments into a recurring BAMF agent report and governed outreach playbook with explicit inclusion and exclusion rules.
08 / Funnel evidence
Campaign momentum
Exact source values are shown on a labeled log scale so large inventory and small conversion stages remain legible without changing the underlying counts.
- 01Campaign momentum: 3 chart-ready points (derived · Campaign performance and robotics ICP). Campaign connections, replies, and robotics-fit prospects.
- 02Official campaign outreach funnel: 4 chart-ready points (observed · BAMF Connect campaign snapshots). Latest official campaign leads/contacts, contacted, accepted connections, and total replies; this is a funnel snapshot, not a period trend.
Sources
Sources
Source coverage, freshness, and applicable window for this report.
- 01Profile-visitor history is omitted until the governed LinkedIn visitor source is connected.
- 02Booked calls, captured emails, and downloads are omitted until the site or calendar conversion source is connected.