Jason M. Lemkin

With agents, that assumption breaks: customers and AI agents want the best result almost instantly and will switch or churn otherwise.

• The reason is agency — agents can make the change themselves, so they do not wait for a vendor relationship to repair.
• Lemkin's operating implication is that B2B software has to be legible and chosen by agents, not just by human buyers.

In one post, Lemkin writes that agents have zero vendor loyalty 'especially since they have the agency to just make the change … themselves.' In a 20VC recap, Lemkin notes agents trigger more than half of Vercel's deployments, up from under 3%, and describes asking an agent live on the show where to host an app; it picked Render, which Lemkin barely knew. Guest Dev Ittycheria's line in the same recap: 'If you're missing from an AI-generated answer, it's like being invisible on Google.'

A tension runs through the same recap.

• Ittycheria says developers show 'very little loyalty' and switch fast, while the enterprise is where software sticks.
• Lemkin's no-loyalty argument is therefore sharper for agent-mediated buyers than for large enterprise contracts.

Read original · Original 2

The argument is not that startups should be hostile, but that the first hires must be able to operate without a training program, a brand, or a long ramp. The priority is week-one shipping, selling without a known logo, and doing the basics without praise.

The post lists seven conditions in parallel: if you need training, help setting up your computer, a first month to get up to speed, maximum time off, an inability to ship in week one, an inability to sell without a brand behind you, or kudos for doing the basics — 'do not join an early stage start-up.' Each condition is about the person, not the company.

Read original

Founders may not yet be great CEOs, but they know the product, customers, market nuances and future more closely than an outside replacement, and they care more. When a non-founder executive shows almost the same level of care, Lemkin's priority is to promote that person fast, double their equity and keep them.

In one post, Lemkin writes that a founder might not be a great CEO yet, but anyone else 'wouldn't really know the product cold,' the customers, the market's nuances, the future, or care as much — so the founder is probably the right CEO.

• In another, the instruction for a VP or exec who cares almost as much as a founder is: 'Promote them fast.
• ASAP.
• Double their equity.
• And … never let them go.'

Read original · Original 2

Transparent pricing, free trials, proofs of concept, SMB customers, one or two large customers, and everyone doing customer support are things a startup does not need to do later, but doing them early improves the product. The reasoning is that pricing and direct customer contact shape what gets built, rather than merely supporting a finished product.

The post's list is deliberately ordinary and operational: transparent pricing, free trials, POCs, SMB customers, 'One Big Customer or two,' and 'Everyone doing customer support.' The framing is that these are things a company might postpone, but they help 'build a better product in the early days.'

Read original

The counterweight is that the buyer's priorities will change, so the founder should assume the offer will not return — roughly nine times out of ten.

• The decision reduces to whether the company is the founder's life's work and whether it is still accelerating.
• If it is not the life's work, Lemkin says take the strong offer, because the difficulty does not decrease.

The post asks: 'Is It Your Life's Work?' If yes and the company is accelerating, push on.

• If deep down it is not, take the offer, because 'you will regret it later.
• Because it never gets easier.' Lemkin closes with: 'It never does.
• You just get better.'

Read original

Peer conversations and market headlines are converted into numbered takeaways, each attached to a specific number, quote or test that a B2B operator or investor can apply. The tests include whether a company is a feature or a franchise, whether a revenue metric is durable once pilots are stripped out, and whether a valuation is worth paying before a growth round.

In the Anthropic recap, Lemkin warns that a reported 900% NRR almost certainly includes pilots converting, and estimates the durable number at 300% to 400%.

• Lemkin also says the worst number of the week was Broadcom backstopping $63B of debt so Anthropic could buy its chips, reading it as evidence that 'debt is an exhaustible source here.' Lemkin vetoed Instinct at a $30B valuation because of a preference for 100 days of traction in a growth round.
• In another recap, Lemkin's test for Salesforce's reported $2B acquisition of Listen Labs is whether the target is 'a feature or a franchise.' Guest Rory O'Driscoll's line closes the first recap: 'this is when the bill for AI comes due, and this is how it'll be paid.'

Read original · Original 2

A newcomer gets both his subject matter (AI-era distribution, acquisition math) and his method in one read.

A recap of a 20VC conversation with Dev Ittycheria, Rory O'Driscoll and Harry Stebbings, written as arrows rather than paragraphs.

• Each point carries a number or a direct quote: OpenAI's run rate nearing $70B and up more than 70% since July; Ittycheria's 12-year view that developers show "very little loyalty" while enterprises stick.
• An anecdote about a Factory adviser becoming Cognition's CRO two hours after Factory cut ties, with Ittycheria saying he has never seen a CRO flip to a direct competitor.
• Vinod Khosla calling his own portfolio company "a struggling second tier competitor," and O'Driscoll's verdict that "it was not venture's finest hour."
• Reflection's Beam open-weight model and the claim that no one at Dreamforce wanted to run a Chinese model.
• Salesforce reportedly paying $2B for three-year-old Listen Labs, with Lemkin's test being whether you are "a feature or a franchise."
• Agents triggering more than half of Vercel's deployments, up from under 3%, and an agent picking Render live as a host.
• Ittycheria's line: "If you're missing from an AI-generated answer, it's like being invisible on Google."
• Oura pulling an oversubscribed $2.2B IPO.

For brands planning a campaign

Directly usable for AI-era positioning work. The agent-picks-Render story and the "invisible on Google" line are concrete ways to argue that a B2B company needs to be legible to AI answer engines, not just to human buyers — a keynote angle with a quotable source.

If you read only one post to understand his AI-era positioning thesis, read this one.

Lemkin argues that the classic SaaS playbook assumed a high degree of vendor loyalty, with NPS as the lighthouse metric, and that this assumption is gone.

• Customers and AI agents, in his framing, want the best results almost instantly and will switch or churn otherwise.
• The sharpest point is about agency: agents have zero vendor loyalty "especially since they have the agency to just make the change ... themselves."
• It is a short post with one claim and one consequence, not a survey.

For brands planning a campaign

A starting point for questioning retention assumptions in a SaaS go-to-market plan. If switching costs fall because agents can migrate themselves, the plan's churn model, onboarding investment and NPS-based health scoring all need re-examination.

Good for seeing how he reasons under uncertainty rather than just what he concludes.

A numbered rundown of a week on 20VC with Harry Stebbings and Rory O'Driscoll.

• The items: Anthropic filing an S-1 with $16.3B in first-half revenue, with Lemkin noting the five-year path from founding to IPO versus SpaceX's 24 years and asking why anyone would fund a company 10-12 years from liquidity.
• A warning that Anthropic's 900% NRR almost certainly includes pilots converting, with a durable figure of 300-400% as his estimate.
• Broadcom backstopping $63B of debt so Anthropic could buy its chips, read as evidence that "debt is an exhaustible source here."
• A view that Microsoft offering exits to 27% of staff is a worse signal than a layoff, with a preference for cutting 9% in one day.
• Cloudflare cutting 20% while growth accelerated, used to argue that the assumed "planes of stability" at $100M, $300M or $500M ARR no longer exist.
• Chinese open-weight models having peaked in US market share, with the Qwen file-upload incident as a security objection.
• And his own decision to veto Instinct at a $30B valuation because he wants 100 days of traction for a growth round.
• O'Driscoll's line closes it: "this is when the bill for AI comes due, and this is how it'll be paid."

For brands planning a campaign

Useful for pricing and growth-story diligence. His NRR read — stripping out pilots to find the durable number — is a practical check a B2B company can apply to its own metrics before taking them to market or to a board.

It answers the question of which early-stage sales and pricing motions are worth doing before they are strictly necessary.

A short list of practices framed as things you don't need to do later, but that help build a better product in the early days: transparent pricing, free trials, proof-of-concepts, SMB customers, one or two big customers, and everyone doing customer support. The framing matters more than the list — each item is something a company might drop as it scales, and Lemkin's point is that doing them early shapes the product rather than just the pipeline.

For brands planning a campaign

A concrete agenda for a pricing or GTM session: transparent pricing and free trials as early-stage decisions, POCs as a sales motion, and the trade-off between SMB volume and one or two anchor customers.

It shows the founder-facing register he uses in advisory settings and the personal test he applies to business choices.

Lemkin reasons that a strong offer usually means you have something still compounding, so it may be worth more in 12-60 months — but also that the buyer's priorities will differ by then, so assume the offer never comes back, "at least, 9 times out of 10."

• He reduces the decision to one question: is it your life's work?
• If it is and you are accelerating, push on; if deep down it is not, take the offer, because you will regret it later.
• He ends on the line that it never gets easier, "you just get better."

For brands planning a campaign

Relevant for founder and executive audiences weighing a raise against an exit, and a useful sample of the framing he brings to advisory conversations with leadership teams.

Read it to calibrate tone as much as content — this is the direct, list-driven voice he uses when the audience is founders and operators.

Seven parallel statements of the form "if you need X, do not join an early stage start-up": training, help setting up your computer, a first month spent getting up to speed, maximum time off, an inability to ship in week one, an inability to sell without a brand behind you, or kudos for doing the basics.

• Each is a condition on the person, not on the company.
• The post states expectations without qualification or reassurance.

For brands planning a campaign

Suited to hiring, culture and early-team keynote material, or as a pointed opener for a founder audience discussing what the first ninety days should demand.

In one 20VC recap, the post notes agents now trigger more than half of Vercel’s deployments, up from under 3%, and repeats a guest’s warning that missing from an AI-generated answer is like being invisible on Google. Lemkin’s closing line: developers switch, executives switch, agents pick — make sure they pick you.

For brands planning a campaign

For a B2B SaaS go-to-market, pricing, and AI-era positioning brief, Jason M. Lemkin can speak or advise on answer-engine visibility, agent-driven buying, and how product stickiness must survive zero-loyalty switching behavior.

Examples: Source 1 · Source 2

One lists what people should not expect from an early startup: training, help setting up a computer, a first month to get up to speed, maximum time off, week-one shipping without support, selling without a brand, or kudos for basics. Another says that when a non-founder executive cares almost as much as a founder, promote them fast, double their equity, and never let them go.

For brands planning a campaign

Lemkin can support leadership offsites, hiring-bar workshops, and founder-mode culture sessions for B2B software teams that need to raise expectations, retain high-ownership operators, and align equity with founder-level commitment.

Examples: Source 1 · Source 2

The approach is reassurance backed by founder-specific knowledge rather than a generic leadership credential.

For brands planning a campaign

This is useful evidence for founder coaching, CEO peer sessions, or a keynote on why founder-led product and customer knowledge matters when a company is deciding whether to bring in outside leadership.

Examples: Source 1

The list treats pricing and customer contact as early product inputs, not later operational polish.

For brands planning a campaign

For a pricing, packaging, or product-led growth brief, Lemkin can challenge teams to test transparent pricing, trials, POCs, and support exposure earlier than they might plan, especially when SMB and large-customer signals need to shape the roadmap.

Examples: Source 1

But the buyer’s priorities may change, so the offer may never return. The decision comes down to whether the company is the founder’s life’s work and whether it is still accelerating; if it is not, he says take the strong offer because it never gets easier.

For brands planning a campaign

Lemkin can contribute to founder, board, or investor sessions on exit timing, M&A readiness, and the emotional and strategic tradeoffs between selling and compounding, rather than treating acquisition interest as a purely financial calculation.

Examples: Source 1

They cover OpenAI’s reported run rate nearing $70B, developer switching versus enterprise stickiness, a CRO moving from Factory to Cognition, Vinod Khosla’s public criticism of a portfolio company, Reflection’s open-weight Beam model, Salesforce’s reported $2B acquisition of Listen Labs, Anthropic’s S-1 and reported $16.3B first-half revenue, Microsoft buyouts for 27% of staff, Broadcom backstopping $63B of debt, Chinese open-weight model risk, Cloudflare’s cuts, and Lemkin’s own investing takes.

Lemkin uses the 20VC format to synthesize guest interviews, reported numbers, and his own numbered takes into short market reads for a B2B software audience.

Several figures and quotes come from guests or reports mentioned in the posts and are not independently verified by the posts themselves; treat them as commentary and reported claims, not audited data.

For brands planning a campaign

A brand can evaluate Lemkin for keynote, panel, or advisory work that translates AI and venture-market headlines into operating implications for B2B software leaders, provided the session distinguishes reported claims from independently verified figures.

Examples: Source 1 · Source 2

The posts show recurring involvement with the show but do not state ownership, employment, or payment.

View example

The posts do not state an ownership or employment relationship.

View example

The subject overlap with Lemkin is the B2B tech audience and the habit of turning a working experience into a lesson a reader can apply directly. Both also build a named body of independent work rather than only commenting in the moment.

Khan writes from the buyer's side of the table, covering buyer psychology, collateral damage to the buyer's team, and clarity-driven leadership, and that work centers on a book and a podcast about the buyer's seat. Lemkin writes from the operator and investor side, working in revenue metrics, valuation tests, acquisition decisions and weekly market analysis.

Read original · Original 2

That is the same short-post structure Lemkin uses when he turns a market headline or a 20VC segment into a numbered take with a consequence attached.

Badhwar stays inside application security: vulnerability eradication, insecure packages and patterns generated by agents, and how quickly teams fix existing risk versus letting new risk in. Lemkin's territory is SaaS go-to-market, startup hiring, acquisition and valuation judgment, with security appearing only as one topic in his market recaps.

Read original · Original 2

Lemkin does the same with a week of 20VC interviews, converting specific events into a general test a reader can reuse. Both write short, plainly argued posts that end in a recommendation.

Boice works in automotive manufacturing and OEM customer experience, reading quality studies, survey data and brand-level sales results. Lemkin works in B2B software, venture funding and AI, so the evidence base and the decisions being advised are different even when the underlying argument — break the silos, connect the signals — is close.

Read original · Original 2

Lemkin when the session is about AI-era positioning for a security or developer-facing product, and the room needs someone who turns a model release into a buyer's decision.

• The reviewed posts do that work on a single event: Anthropic shipped Opus 5.5 two months after Opus 5, release cycles moved from six months to a few weeks, developers take whichever model is fastest or cheapest that day, and the code plus every dependency it pulled in is already in production.
• The argument lands on a practical instruction — stop reacting to this release and have a plan for the next one — and on a second effect worth planning for, that better code-reading models surface more vulnerabilities in open source a team already depends on.
• A second post frames risk as a two-sided measurement: how fast existing risk is fixed, and how often new risk is introduced, with the second number telling a team whether it is getting ahead.
• For a security product's positioning workshop, that is closer to the buyer's daily problem than a broader market read.

The reviewed material is narrow: cyber, application security, AI-generated code, and vulnerability queues.

• There is no reviewed evidence on pricing, retention, founder hiring, fundraising, or agency-era go-to-market, so a general SaaS growth or operating keynote would lose Lemkin's range.
• Two posts is also thin evidence of speaking style, format, and commercial availability, so a paid engagement would need confirmation first.

Examples: Source 1 · Source 2

Lemkin when the campaign decision is about why sellers lose deals they should win — buyer psychology, the internal cost a purchase imposes on the buying team, and pitch differentiation.

• The reviewed post describes the buyer's side of the table directly: while a seller pitches a product, the buyer is silently adding up internal hours drained from the team, the collateral damage of switching, and how identical this pitch sounds to the last five heard that week.
• That is usable as workshop material for a sales-enablement or go-to-market session in a way that a market-recap format is not.
• The other reviewed post shows a public author platform around clarity and leadership — a book with a launch event, a signing, and a reader-facing announcement — which suits a session or event that wants a leadership-and-clarity theme alongside commercial content.

The reviewed evidence covers buyer psychology and a leadership book, not pricing, retention, or market analysis, so a keynote built on portfolio-level venture commentary would not be supported.

• One of the two reviewed posts is mainly a book and podcast announcement, which shows platform and promotion rather than the substance of a talk.
• With two posts, there is no evidence of session format, repeat speaking, or commercial terms.

Examples: Source 1 · Source 2

Lemkin when the audience is automotive — an OEM, a supplier, or a company selling customer-experience and data platforms into vehicle brands — and the session needs current industry examples with an operating moral attached.

• The reviewed posts read two recent situations closely: Ford's quality turnaround, where vehicle engineering, manufacturing, supply chain, and quality were put under one industrial system and expanded to include digital, design, and global teams, followed by top placement in an initial-quality study and a 41-point reduction in problems per 100 vehicles; and Nissan's nineteen consecutive months of US retail growth against a worst global sales year since 2003, set against Chinese brands taking nearly 10% of the European market and one in four cars sold in Australia.
• Both posts carry that industry reading into a customer-experience argument about breaking departmental walls so survey, reputation, service, and behavioural data sit in one view under one accountable team.
• For an automotive brief, those specific examples do work Lemkin's posts do not.

The subject matter is automotive and customer experience, not B2B SaaS go-to-market, pricing, or AI-era positioning, so the advantage only appears when the audience or product is automotive-adjacent or needs a CX-organisation story. Two posts is thin evidence of range beyond that vertical, and there is no evidence here about formats, availability, or commercial terms.

Examples: Source 1 · Source 2