Thomas Lah

Complex AI systems require substantial implementation and hands-on integration, a pattern they compare to the forward deployed engineer role and to how Salesforce and ServiceNow were built in the cloud era.

• The tradeoff is explicit: accepting low gross margins on professional services in year one can embed the AI product in the customer's system of work, paving the way for durable, high-margin recurring revenue.
• Lah also argues that automation shifts effort from transactional, reactive work such as level 1 support to transformational, outcome-based work in customer success and specialized professional services.
• The right internal targets, they say, are total gross profit and customer lifetime value, not year-one service margin.

Responding to an Andreessen Horowitz article, Thomas Lah argues that an obsession with pure product-led growth obscures how enterprise platforms have actually been built. They close by asking readers to examine their service portfolio and judge whether they are investing enough in the high-touch, transformational work required to build an AI moat.

Read original

They argue that outcome pricing keeps attention on actual customer results rather than activity, exposes weaknesses in product quality, onboarding and change management, and aligns go-to-market, product and support around the same scoreboard. The internal cost is the tension Lah identifies: the new model can cannibalize existing SaaS annual recurring revenue, so finance leaders must trade familiar predictability for a model that only wins when customers win.

Lah credits Intercom CFO Dan Griggs for underwriting the transition and co-founder Des Traynor for publicly discussing what it takes to make it work, drawing on an article at mostlymetrics.com.

Read original

Much of the AI investment in sales, they say, is layered onto CRM systems that were never designed around the buyer.

• TSIA's Model Driven Revenue Engine, as Lah describes it, would use AI to monitor every customer touchpoint in real time instead of relying on CRM fields and a Thursday pipeline review.
• The post summarizes two guest views: Stephen Messer, co-founder of Collective[i] and the guest on TECHtonic episode 132, argues against the forecast call and bolting AI onto an unsuitable CRM; OpenAI CFO Sarah Friar, writing on August 10, describes finance ambitions of a zero-day close and continuously updated forecasting built from statistical models, sales conversations, account evidence, operating data and finance judgement in one live view.

Lah places the episode released on August 7 alongside Sarah Friar's August 10 piece, presenting a revenue operator and a CFO describing the same destination from opposite ends of the building. They ask how many sales and finance organizations are ready for revenue forecasting in the AI era.

Lah extends the same AI-economics argument about services and pricing to forecasting: AI changes where effort, margin and decision signals sit, from services delivery and pricing models to revenue prediction.

Read original

The post is the clearest statement of the position running through the other two.

Thomas Lah responds to a piece from Andreessen Horowitz and restates TSIA's position on AI economics: AI will not remove services work but reorder it, because complex AI systems need heavy implementation and hands-on integration.

• The post points to the "forward deployed engineer" role and compares the pattern to Salesforce and ServiceNow in the cloud era.
• The argument spells out the tradeoff: low gross margin on professional services in year one is an investment in embedding the AI product in the customer's system of work, which produces durable, high-margin recurring revenue.
• AI automation also shifts effort from transactional, reactive work such as level 1 support to transformational, outcome-based work in customer success and specialised professional services.
• The closing question asks whether readers are spending enough on the high-touch services needed to build an AI moat.
• The post presents both the thesis and the opposing assumption that pure product-led growth is sufficient for enterprise AI.

For brands planning a campaign

For a B2B SaaS company weighing AI monetisation, this piece frames the central tension a campaign or session would need to resolve: where to accept low service margins deliberately to secure recurring revenue. It is a natural anchor for a keynote segment on services-led growth, and the Salesforce and ServiceNow comparison gives the argument a concrete historical reference point.

Where the services-led post is about strategy, this one explains what an outcome-based pricing switch does inside a company and names the specific people who made the call.

Thomas Lah writes about Intercom's move from seat-based pricing to outcome-based pricing for an AI product, drawing on an article on mostlymetrics.com.

• The post credits CFO Dan Griggs for underwriting the transition and co-founder Des Traynor for publicly discussing what the change requires.
• Its useful observation is that outcome-based pricing focuses attention on customer results and exposes weaknesses in product quality, onboarding and change management: the model only pays off when the customer gets a result.
• Lah also notes the internal cost that makes the decision hard: the new model will cannibalise existing seat-based SaaS annual recurring revenue.
• The takeaway is a CFO trading predictable revenue for a model that succeeds only when customers do.

For brands planning a campaign

A campaign aimed at finance and product leaders can use this example to open a discussion about willingness to disrupt existing revenue in exchange for outcomes. For a session or advisory engagement, it works as a named case; check the underlying Intercom and mostlymetrics material before presenting it to an audience as fact.

It is the best example of the connective, side-by-side style used in these posts.

Thomas Lah sets two pieces of evidence side by side.

• The first is episode 132 of the TECHtonic podcast, released on August 7, featuring Stephen Messer, co-founder of Collective[i], which the post describes as an AI sales intelligence network compared by the Wall Street Journal to "Waze for sales."
• Messer's argument, as summarised, is that the weekly forecast call is guesswork, consumes selling time and does not improve accuracy, and that much current AI spending in sales sits on top of a CRM built around something other than the buyer.
• The second is a piece by OpenAI CFO Sarah Friar dated August 10, which describes two ambitions for the finance function: a zero-day close and continuously updated forecasting, built from statistical models, sales conversations, account evidence, operating data and finance judgement in one live view.
• Lah concludes that a revenue operator and a CFO are describing the same destination: revenue forecasting is becoming model driven rather than meeting driven, which is what TSIA calls the Model Driven Revenue Engine.
• The post closes by asking how many sales and finance organisations are ready for that shift.

For brands planning a campaign

This is the strongest starting point for a sales or finance audience: it pairs two credible voices from different parts of a business around one operational change. For a briefing or advisory session, the forecast call as a contested practice offers a low-friction way to discuss how a company's revenue data and processes would need to change if forecasting were model driven rather than review driven.

In a post responding to an Andreessen Horowitz article, they say complex AI requires heavy implementation and hands-on integration, comparing that work to the cloud era at companies such as Salesforce and ServiceNow.

• They frame professional services with initially low gross margins as an investment in embedding an AI product into the customer's system of work, and describe services as the engine of AI adoption and the ultimate moat.
• The post also says transactional, reactive services give way to transformational, outcome-based services.

For brands planning a campaign

A B2B SaaS company evaluating AI monetization can ask Lah to connect services spending to durable recurring revenue, not just short-term margin. For a keynote or advisory session, they can make a practical case for how much high-touch implementation to fund and how to price it.

Examples: Source 1

They treat the change as more than packaging: it disrupts seat-based SaaS economics and forces a focus on customer results, exposes weaknesses in product quality, onboarding and change management, and aligns go-to-market, product and support around the same scoreboard. The post also stresses the CFO conviction required because the model can sacrifice existing SaaS ARR.

For brands planning a campaign

For a brief on AI monetization, Lah can speak to pricing-transition tradeoffs, internal alignment and the revenue risk of moving away from seats. The concrete Intercom example gives a keynote a case to reason from rather than an abstract pricing argument.

Examples: Source 1

They say the weekly forecast call is guesswork, consumes selling time and does not improve accuracy, and that much AI spending in sales is layered onto CRM systems not designed around the buyer. They point to Friar's finance ambitions, including a zero-day close and continuously updated forecasting, and conclude that revenue forecasting is going model driven, not meeting driven.

For brands planning a campaign

A B2B SaaS company can use Lah to challenge forecasting rituals and make the case for AI-assisted pipeline review. This topic fits a session on operating-model change across sales and finance, with a clear contrast between CRM-field reporting and real-time customer signals.

Examples: Source 1

One post explicitly quotes 'TSIA's Take' against an Andreessen Horowitz article; another credits mostlymetrics and names Intercom's CFO and co-founder; a third pairs a TECHtonic episode with a CFO essay. Recurring formats include LinkedIn commentary, references to named operators and companies, numbered framework points, and podcast cross-promotion under the TSIA and AI Economics tags.

The three posts follow a repeatable pattern: cite third-party work, add a TSIA framework or take, name relevant operators or companies, and close with a direct question to readers.

For brands planning a campaign

A brand can expect Lah to bring a framework-led, source-heavy approach rather than purely personal opinion. For a keynote or advisory session, ask for a tailored TSIA-style framework and concrete company examples so the session produces decision criteria, not just market commentary.

Examples: Source 1 · Source 2 · Source 3

They also host TSIA's TECHtonic podcast, where they introduce Stephen Messer as "my guest." For a formal affiliation, ask what title Lah holds and whether they are a TSIA employee.

View example

They cite the piece as outside support for the argument.

View example

They use the company's decision as a case of AI-era business-model redesign.

View example

They comment on OpenAI's published writing.

View example

They distinguish advisory work from forward-deployed engineering and tie the argument to AI economics and professional services strategy.

Bo Di Muccio's posts rebut the prediction that AI will collapse services, using ServiceNow's internal AI results as a central example. Thomas Lah frames the same territory through TSIA models, focusing more on pricing, margin tradeoffs, and revenue forecasting.

Read original · Original 2

This overlaps with Thomas Lah's interest in AI economics and operating models.

Ryland draws on military strategy and physical AI deployments in factories, stadiums, and defense, with a policy and accountability lens. Thomas Lah stays closer to enterprise software and services businesses, pricing models, and forecasting.

Read original · Original 2

Polamero's posts center on founder-led scale, capacity, utilization, delegation, and sales-marketing orchestration. Thomas Lah focuses on AI economics, outcome-based pricing, and model-driven revenue forecasting.

Read original · Original 2

Both posts focus on that question.

• The first argues that the predicted collapse of the services layer has not happened: Anthropic launched a services firm, OpenAI acquired consulting firms, and Microsoft, AWS, and Google each built AI-services efforts, with more than $9 billion committed in about two months to close the gap between capable models and business outcomes.
• The post distinguishes advisory work from the build itself and treats Forward-Deployed Engineering as a distinct operating model with its own risk, margin, talent, and success measures.
• It warns that lumping professional services and forward-deployed engineering together misprices both.
• The second post cites ServiceNow's reported elimination of about 2.5 million work hours from its own operations to argue that AI compresses delivery hours, not the value of the outcome.
• Services organizations, it says, should move from implementation billing to pricing and operating around adoption and outcomes.
• For buyers deciding how to structure and price a services motion around an AI product, this is more direct than a keynote framed around the broader market.

The material focuses on the economics and operating model of professional services, PS 2.0, and forward-deployed engineering.

• It does not cover the SaaS seat-to-outcome pricing transition at Intercom or the revenue-forecasting cadence linking pricing to sales and finance operations.
• Buyers choosing Bo Di Muccio should expect an argument centered on services organizations; if the session also needs a software pricing-change example or a finance-operations review, they should source separate material.
• The ServiceNow example is a secondhand account, so check its figures and customer-growth claims against the original before using them with clients.

Examples: Source 1 · Source 2

The May post uses John Boyd's OODA loop to argue that companies applying observe-orient-decide-act to AI strategy are moving faster.

• It cites PepsiCo's reported three-month path to a 20% throughput lift at a Gatorade plant with Siemens and NVIDIA; BMW's virtual factory across thirty-plus production sites, cutting collision checks from four weeks to three days; and Unilever's Tinsukia plant, reducing trial time by eighty-four percent.
• The post names pilot purgatory as the failure mode and closes with five concrete operator actions.
• The July post extends the argument to physical AI: assembly lines, stadiums, borders, and critical infrastructure.
• It describes the World Cup as a civilian AI surveillance deployment and measures Ukraine's drone defense by cost per defended night rather than intercept rate.
• It then pushes accountability beyond the audit log, proposing an independent panel to certify the result and payment gated on that outcome instead of a consumption meter.
• For B2B buyers whose AI products serve industrial, logistics, security, or public-safety operations, this is a closer match than a services-margin argument alone.

The tradeoff is domain and evidence.

• These posts are aimed at operators who carry a 10-K and answer for an EBITDA line, but their examples largely concern physical AI, defense, and industrial deployments—not B2B SaaS monetization.
• They offer neither a seat-to-outcome SaaS repricing case nor a services-investment thesis like the Thomas Lah posts.
• The posts also rely on external reports, Substack material, and primary-source receipts that they do not include.
• Verify the PepsiCo, BMW, Unilever, World Cup, and Ukraine figures before repeating them to clients.
• Buyers seeking a pricing and packaging framework for a software product would get the decision-speed and accountability half of the conversation here, not the pricing half.

Examples: Source 1 · Source 2

The Stacked to Scale episode with Tom Tonkin and Marc Rosenkoetter covers the shift from founder-led growth to sustainable scale: why most growth pain is identity-based rather than operational; the move from personal heroics to institutional capability; codifying expertise without commoditizing it; delegation and accountability; capacity management and utilization cliffs; and turning tribal knowledge into scalable systems.

• The post on intent data makes the demand-side case: intent data does not create revenue by itself; the gap is orchestration, because sales and marketing see different versions of the prospect and follow-up depends on memory.
• It argues that AI can connect the system so sales knows who is engaged, marketing knows the next message, and the CRM becomes a source of truth rather than a record graveyard.
• For buyers whose bottleneck is staffing, delegating, and scaling delivery—or routing demand inside a professional services firm—this alternative speaks more directly than a market-level AI economics keynote.

Thomas Lah's posts focus on running and scaling a professional services organization, change management, utilization, and marketing-sales orchestration.

• They do not offer a framework for pricing an AI product around outcomes or address trading year-one services margin for a durable platform moat.
• Buyers choosing Thomas Lah should expect a practical discussion of capacity, delegation, and demand handoffs.
• If the session also needs an AI-monetization or outcome-based-pricing thesis, they should add another speaker or a separate module.
• One post promotes a podcast; the other is a short marketing-process essay.
• Together, they show the topics and argument style, but not a full advisory methodology.

Examples: Source 1 · Source 2