Moritz Heininger writes about how businesses can put AI to work: preparing teams, redesigning workflows and introducing agents with clear tasks, access and oversight.
Their posts connect practical implementation to bigger questions about software strategy, jobs, sales operations and cybersecurity.
Readers looking for a grounded view of the move from AI experimentation to operational change will find concrete frameworks as well as arguments about what that shift means for companies.
Heininger’s distinctive angle is to treat AI adoption as a work-design and management challenge, not just a choice of model.
In a guide to introducing AI agents, they compare an agent to a new employee: define its task, provide appropriate access, start with human approval, test it on real cases and measure the cost of completed work.
On workforce change, they argue for understanding where work happens, enabling people and redesigning processes before cutting roles (post on AI-related job cuts.
They also look beyond individual workflows.
Their analysis of “headless” SaaS argues that as agents use APIs and structured data, software value may shift from the user interface to the underlying data and integrations.
In a post about measuring AI productivity, they make enablement central to whether adoption produces results, illustrating the point with a day of varied work they say they completed using AI.
These examples make their writing useful to business readers weighing how to introduce agents, assess software and change work without mistaking adoption for impact.
Moritz Heininger argues that most disappointing AI results stem from enablement and sequencing rather than technology.
Read more
In their posts, companies adopt tools without teaching people to use them, cut roles before re-cutting the work, and buy software that AI agents cannot operate.
They treat an AI agent as a new hire that needs a job description, system access, staged autonomy and a real cost measurement.
They follow frontier-model capability mainly as an operational and security problem.
A recurring priority is evidence: their own detailed work logs, named practitioners with numbers, and vendor announcements read through an operator’s lens rather than taken at face value.
AI results depend on enablement, not on the model
Heininger challenges the widely repeated finding that AI produces hardly any measurable efficiency gains, asking which companies were actually measured.
Read more
Their argument: adoption without training produces no results.
The missing layer is enablement—knowing which systems and context to use, which workflow to follow, and when to trust the model.
Companies, in this reading, do not have an AI problem but an enablement problem.
They describe a single nine-hour day producing ten projects: an event concept for a leadership dinner with 25 executives; a personalized email campaign reconciled contact by contact against the CRM; a complete landing page in corporate design; a competitive-intelligence analysis that involved de-anonymizing an anonymized M&A document; and a data study across 12,477 companies, delivered as an interactive report with five posts and 25 graphics.
They estimate the same output would take a full working week without AI, or closer to two with external providers.
They compare measuring AI in untrained teams to putting bicycles in an office and concluding nobody cycles to work faster because nobody learned to ride.
Onboard an AI agent the way you onboard an employee
Heininger’s practical framework treats a first AI agent as a new hire and walks through what is normally missing before launch.
Read more
The list covers a job description with one clear task, an accountable department and a three-month success measure; a workplace with access to ERP, CRM and Drive under a separate account with correct permissions; onboarding that moves from suggestions to approved actions to independent work; probation tested on real historical cases; rules for data protection, works council and the EU AI Act; integration into the tools the team already uses; feedback reviews of decisions and misses; and a true cost per completed task, including monitoring and maintenance. They argue the model choice is roughly as important as the laptop a new colleague receives.
The probation step is concrete: give the agent 50 real cases from the previous year, including the difficult ones, and compare its results with what people achieve.
The post opens with a Gartner expectation that more than 40 percent of agent projects will be discontinued by the end of 2027.
It also names job-loss anxiety as something that deserves an honest conversation rather than silence.
Heininger predicts that many SaaS providers will lose their user interface because AI agents need no polished front ends, only clean APIs and MCP connections to structured, machine-readable data.
Read more
Power shifts to whoever becomes the central data store for a function or an entire company, such as CRM, ERP, ticketing or a knowledge base, and away from providers with the nicest click path and thinnest data foundation.
Opening up as a data store, however, invites well-funded agent startups to pull out the context and bundle a free version.
That pushes incumbents toward their own agents, bundles and pricing by compute and storage instead of seats.
For mid-market buyers, the advice becomes a procurement test: for every new software purchase, check how agent-ready it is. If the answer is only that work is in progress, buy something else.
Heininger argues that AI-driven restructuring fails when companies get the sequence wrong: understand where work actually arises, enable people, then rebuild processes.
Read more
Cutting roles first, in this account, moves the work onto those who remain and loses the knowledge of those who leave. The correct order takes longer than a quarter, and that delay is the point.
They cite 97,000 US job cuts in May, 40 percent attributed to AI, the highest May figure since 2020, and contrast it with a hiring-manager survey in which more than a third of those who cut a role because of AI later filled the same or a very similar role. A former lululemon CIO, quoted by Heininger, coined the term "AI washing" for claiming more AI activity than is real under result pressure; Heininger adds "AI wishing" for treating AI as a magic wand over a hard problem.
The post extends the enablement argument into a sequence: capability building first, process redesign second, headcount changes last.
Frontier model gains are a security operations problem
Heininger reads frontier-model announcements through an operator’s lens.
Read more
New offensive capability, they stress, is a by-product of general improvements in coding, reasoning and autonomy rather than a targeted feature: better coding automatically means better hacking. Their operational conclusion is that defenders are strengthened in the long term, the short term is turbulent, and any company still measuring patch cycles in weeks rather than hours has a problem.
Responding to Anthropic’s announcement about Claude Mythos Preview, they report that the predecessor model Opus 4.6 produced working Firefox exploits twice in several hundred attempts, while the new model reached 181.
It also found a 27-year-old bug in OpenBSD, a 17-year-old remote code execution bug in FreeBSD and a 16-year-old bug in FFmpeg, the library behind most video services.
According to the account, engineers without security training ran the model in the evening and had a working exploit by the morning.
The model claims and figures come from Anthropic’s announcement, not from independent testing by Heininger.
Teach AI by building it layer by layer into real work
Heininger’s teaching method is applied and layered rather than tool-focused: participants build what they call a personal AI operating system and apply each layer directly to their own work, with the stated goal of orchestrating instead of executing.
Read more
The practitioner events they promote follow the same preference for applied evidence, with operators presenting measured cases instead of product demos.
The Agent Builder course runs eight weeks across six layers: set up an own agent; give it context about how you work; connect it to mail, calendar and CRM through MCPs; turn recurring tasks into skills; set automations that run without a person as the trigger; and finally build an individual app or piece of software.
The promoted Scalantec AI Summit in Berlin followed the same logic, with five named operators presenting real numbers: Julius Göllner of REP on what moves pipeline across 30 SaaS companies; Roman Hipp of BetterContact on reaching $3M ARR with three people; Dirk Ramm of charles on a lead-scoring rebuild that produced 50 percent more meetings at constant volume; Nicolas Schell of Scalantec on where AI matters across a GTM organisation; and Heininger on turning CRM leads and LinkedIn engagement into qualified pipeline.
The course is a pilot cohort with Monday input, Friday consulting and one-to-one coaching.
The summit was limited to 60 curated seats and four round tables of 15.
This post gives newcomers the core argument behind Heininger’s AI-agent work and explains how it could change software decisions.
Read more
Heininger argues that software dashboards are moving into the background because AI agents need clean APIs and MCP connections, not polished user interfaces.
Agents can handle tasks such as contract data, customer status, and ticket closing by accessing structured, machine-readable data.
In this reading, market power shifts to providers that become the central data store for a function or an entire company, while providers with the best click path and thinnest data foundation lose ground.
A second-order risk: once a vendor opens itself up as a data store, well-funded agent startups can extract the context and bundle a free version.
Heininger’s countermeasures are to build proprietary agents, offer bundles, and price based on compute and storage rather than seats.
For mid-market buyers, the advice is to check how agent-ready new software is and walk away if the only answer is that work is in progress.
For brands planning a campaign
Useful as a workshop segment on software selection and build-versus-buy decisions: it turns an abstract platform shift into procurement criteria mid-market clients can put to vendors.
The clearest practical framework in these posts for anyone planning a first agent project and asking what must be in place before launch.
Read more
Heininger compares a company’s first AI agent to a new employee and walks through what is usually missing.
The job description names one task, the responsible department, and how performance will be measured after three months.
The workplace includes access to ERP, CRM, and Drive, with a separate account and appropriate permissions.
Outdated documents, they note, can make even a strong colleague produce errors.
Onboarding starts with suggestions, moves to actions with approval, and only later allows independent work.
Probation means testing the agent against 50 real cases from the previous year, including difficult ones, and comparing its results with people’s.
Rules cover what the agent may see, who will spot a manipulated email, and prior clarification of data protection, works council, and EU AI Act questions.
The team section puts the agent in Outlook, Teams, and CRM and calls for an honest conversation with anyone worried about their job.
Feedback reviews ask what the agent decided and why, and where it underperforms.
The final point is to measure cost per completed task—including monitoring and maintenance—before starting.
Heininger adds a Gartner expectation that more than 40 percent of agent projects will be cancelled by the end of 2027, and argues that the model itself is roughly as important as the laptop a new colleague receives.
For brands planning a campaign
A ready-made structure for a client workshop on agent adoption: each heading can become a checklist item to work through before a pilot starts.
This post explains why Heininger sees enablement, not model choice, as the central problem, using a concrete day rather than a general claim.
Read more
Heininger challenges the recurring finding that AI produces hardly any measurable efficiency gains and asks which companies were actually included in the measurement.
The account describes nine hours of work starting at 9 a.m.: an event concept for a leadership dinner with 25 executives; a personalized email campaign reconciled against the CRM contact by contact; a complete landing page in corporate design; a fixed integration problem between two tools; a competitive intelligence analysis that involved de-anonymizing an anonymized M&A expose; a consultant profile as PDF; a podcast concept with launch plan; a data study covering 12,477 companies, delivered as an interactive report with five posts and 25 graphics; and preparation for three AI driver's license sessions starting the following week.
Heininger estimates the same output without AI would take a full working week, or closer to two with external providers for design, data analysis, and writing.
The point is not that they work unusually fast, but that knowing which systems, context, and skills to use, which workflows to follow, and when to trust the model changes the result.
The comparison is a bicycle placed in an office: after three months, nobody arrives faster because nobody learned to ride.
The conclusion: companies do not have an AI problem but an enablement problem.
For brands planning a campaign
A strong opening for a talk to sceptical leadership teams: it answers the objection that AI has not delivered and reframes the missing layer as enablement, not technology.
This post explains Heininger’s thinking on AI and the workforce, a key consideration for any client weighing headcount changes against capability building.
Read more
Heininger cites 97,000 job cuts in the United States in May, 40 percent of them attributed to AI, the highest May figure since 2020.
The number Heininger says receives less attention comes from a survey of hiring managers: more than a third of those who cut a role because of AI filled the same or a very similar role again afterwards.
A former lululemon CIO has coined the term "AI washing" for the moment a company under pressure to show results claims more AI activity than is actually happening, and Heininger adds "AI wishing" as its twin, the belief that AI is a magic wand that removes the need to do the work.
What is missing in these cases, Heininger writes, is never the model but the sequence: cutting roles before re-cutting the work only moves it to the people who remain and loses the knowledge of those who leave.
The companies Heininger works with do it the other way around, first understanding where work actually arises, then enabling people, then rebuilding processes, which takes longer than one quarter.
The post ends by asking how many AI announcements in someone's environment have produced a measurably different working day within twelve months.
For brands planning a campaign
Suitable for change-management and HR audiences: it offers a specific caution against headcount-first AI plans and a sequence clients can follow instead.
This post shows what Heininger teaches and at what depth—a quick way to assess the shape and level of a talk or workshop.
Read more
Heininger says the same question followed every AI driver's license session: people use AI daily and want to get out of the chat window.
The Agent Builder is an eight-week advanced course in which participants build their own AI operating system, layer by layer, applying each layer directly to real work.
The six layers are setting up their own agent, giving it context about how they work, connecting it to mail, calendar, and CRM through MCPs, turning recurring tasks into skills, setting automations that run without a person as the trigger, and finally building an individual app or piece of software.
The stated goal is to orchestrate rather than execute.
A pilot cohort starts on 31 August, with a fixed rhythm of Monday input, Friday consulting, and one-to-one coaching for each participant’s own project.
A free information session is announced for 22 July at 12:00, with the first ten places at half price.
For brands planning a campaign
The six layers offer a reusable workshop outline and show where beginner prompts end and agent work begins.
This post shows how closely Heininger follows frontier-model capabilities and why those changes affect ordinary IT operations, not just security specialists.
Read more
Heininger responds to an Anthropic blog post about Claude Mythos Preview, which autonomously found and exploited security vulnerabilities in major operating systems and browsers.
The comparison is numeric: the predecessor model Opus 4.6 succeeded twice in several hundred attempts to write a working exploit for known Firefox bugs, while Mythos Preview reached 181 successes.
The model found a 27-year-old bug in OpenBSD, a 17-year-old remote code execution bug in FreeBSD, and a 16-year-old bug in FFmpeg, the library behind most video services.
According to the account, engineers without security training started the model in the evening and had a working exploit by morning.
Heininger stresses that these abilities were not specifically trained but emerged as a by-product of general improvements in code, reasoning, and autonomy—so better coding automatically means better hacking.
Anthropic deliberately keeps the model closed for now and starts Project Glasswing to protect critical software before comparable models become widely available.
Heininger expects defenders to be stronger in the long term and turbulence in the short term, and argues that any company still measuring patch cycles in weeks rather than hours has a problem.
For brands planning a campaign
Relevant for clients with security or infrastructure concerns: it frames rapid patch cycles as a management decision, not a technical detail.
This post shows where their work intersects with revenue operations and pipeline generation, often a deciding question for sales-led clients.
Read more
Heininger describes how Ammar Kamran of Scalantec built snipKI’s complete go-to-market system, revealing almost 40,000 leads in the company’s HubSpot CRM that were previously little understood.
The post promotes a Scalantec AI Summit in Berlin on 2 July, focused on AI in sales and scaling revenue without additional headcount.
Five operators will present numbers and use cases.
The programme includes a fireside with Julius Gollner of REP on what actually moves pipeline across 30 SaaS companies; a keynote by Roman Hipp of BetterContact on reaching $3M ARR with three people; a keynote by Dirk Ramm of charles on a lead-scoring rebuild that produced 50 percent more meetings at the same volume; a live demo by Nicolas Schell of Scalantec on where AI matters across the whole GTM organisation; and Heininger’s session on converting CRM leads and LinkedIn engagement into qualified pipeline.
Four round tables with 15 seats each follow, then food, drinks, and an open bar.
They mention Clay tables, Claude Code workflows, and lead-scoring logic as things participants will see, and describe 60 curated seats for sales, RevOps, and founders.
For brands planning a campaign
Useful for clients focused on pipeline results rather than enablement theory: it shows the sales-operations side of the work and the operator-level cases Heininger brings together.
Moritz Heininger argues that weak measured productivity gains from AI say more about training than technology.
Read more
In one post, they counter the claim “there are hardly any measurable efficiency gains from AI” with a day log: ten projects in nine hours, including an event concept for a leadership dinner, a personalised CRM-matched email campaign, a corporate-design landing page, a tools integration fix, a competitive-intelligence analysis, a consultant profile PDF, a podcast launch plan, and a study across 12,477 companies with an interactive report, five posts and 25 graphics.
Their conclusion: companies do not have an AI problem; they have an enablement problem.
Teams never taught how to work with AI will not show results.
Uses a first-person time log and a bicycle analogy — putting bikes in the office and concluding nobody cycles faster — to locate the failure in enablement, not the tool.
For brands planning a campaign
A strong angle for a workshop or talk for mid-market clients whose AI programmes are delivering disappointing returns: focus on capability building, workflows and judgement about when to trust the model, using a before-and-after productivity structure.
Moritz Heininger frames agent adoption as a hiring problem: would you let a new employee start without a job description, access or onboarding?
Read more
The post lays out the missing pieces: one clear task, an accountable manager and a three-month success measure; accounts with correct permissions for ERP, CRM and Drive; a ramp from suggestions to approved actions to independent work; probation tested on 50 real cases, including hard ones; rules for data protection, works council and the EU AI Act; integration into the team tools people already use; feedback reviews of decisions and misses; and the true cost per completed task, including control and maintenance. They cite a Gartner expectation that more than 40 percent of these projects will be discontinued by the end of 2027, and say the model choice matters about as much as the laptop.
Maps each stage of employee onboarding onto the agent lifecycle, giving companies a checklist to work through before launch.
Acknowledges that many agent projects fail: the Gartner discontinuation estimate is the premise, not a footnote, and the post says job-loss anxiety needs an honest conversation.
For brands planning a campaign
A ready-made structure for a client workshop or playbook on agent rollout and governance: job description, access, staged autonomy, probation cases, compliance, team integration and cost measurement — all in buyer language.
In a headless-SaaS argument, Moritz Heininger predicts that many SaaS providers will lose their user interface because AI agents do not need polished front ends; they need clean APIs and MCP connections to structured, machine-readable data.
Read more
They say power shifts to whoever becomes the central data store for a function or an entire company — CRM, ERP, ticketing or knowledge base — and away from providers with the prettiest click path and thinnest data foundation.
Opening up as a data store invites well-funded agent startups to absorb the context and bundle a free version into their agent, pushing incumbents towards their own agents, bundles and pricing by compute and storage instead of seats.
Their practical advice to mid-market buyers: for every new software purchase, check how agent-ready it is.
If the answer is “we're working on it”, buy something else.
For brands planning a campaign
Suits content or a session for SaaS vendors and product leaders on API and MCP readiness, data ownership and pricing-model change, or a buying-criteria segment for mid-market software buyers.
Moritz Heininger describes Anthropic's Claude Mythos Preview finding and exploiting security holes across major operating systems and browsers, including a 27-year-old OpenBSD bug, a 17-year-old remote code execution bug in FreeBSD and a 16-year-old FFmpeg bug behind many video services.
Read more
They contrast the predecessor model's two successful Firefox exploits in several hundred attempts with 181 for the new model, and report that engineers without security training obtained working exploits overnight.
Their point for business readers: these abilities were not specifically trained; they are a side effect of better coding, reasoning and autonomy.
Better coding means better hacking.
Anthropic withheld the model publicly and started Project Glasswing to protect critical software.
They conclude that defenders gain long term, the short term is turbulent, and any company still measuring patch cycles in weeks has a problem.
Reads a vendor announcement through an operator's lens: separates the capability numbers, the cause — general model improvement — and the defensive response, then states the operational consequence.
Notes that the capability is an unintended by-product, not a targeted feature, and that the model was deliberately not released publicly.
For brands planning a campaign
A segment for executive audiences that puts AI risk and security readiness alongside adoption, with a current technical example of why patching and monitoring timelines need to shorten. No security-specialist audience required.
Moritz Heininger cites 97,000 US job cuts in May, the highest May figure since 2020, with 40 percent attributed to AI.
Read more
The number Moritz says nobody discusses comes from a survey of hiring managers: more than a third of those who cut a role because of AI later filled the same or a very similar role.
Moritz quotes the former lululemon CIO's term "AI washing" — claiming more AI progress than is real under result pressure — and adds its twin, "AI wishing": treating AI as a magic wand over a hard problem to avoid the work.
What is missing, Moritz argues, is the order: understand where work actually arises, enable people, then rebuild processes.
Companies that cut roles first push the work onto those who remain and lose the knowledge of those who leave.
Pairs a labour-market statistic with a hiring-manager survey, then uses two named terms to distinguish announced AI activity from changed work.
Acknowledges that the right order takes longer than a quarter — and says that delay is the point.
For brands planning a campaign
Useful as a change-management counterweight in workforce-planning workshops: gives clients a sequence — understand, enable, rebuild — and language to challenge AI cost-cutting narratives internally.
Teaching formats: cohort course and practitioner summit
Two formats show how Moritz Heininger packages practical AI work.
Read more
The Agent Builder course runs eight weeks and builds what they call a personal AI operating system layer by layer: setting up an agent, giving it context about how you work, connecting it via MCP to mail, calendar and CRM, turning recurring tasks into skills, adding automations that run without you as the trigger, and finally building a custom app or software.
It is a pilot cohort with a fixed rhythm — Monday input, Friday consulting and one-to-one coaching — plus an info session and early-bird pricing for the first ten places.
Separately, they promoted the Scalantec KI Summit in Berlin, where five operators presented real numbers and use cases on AI in sales: pipeline discussions from 30 SaaS companies, a three-person team reaching $3M ARR, a lead-scoring rebuild that lifted meetings 50 percent at constant volume, and CRM and LinkedIn engagement turned into qualified pipeline.
The summit also featured round tables, food and a curated room of 60 sales, RevOps and founder attendees.
Combines a structured cohort course with live, numbers-led practitioner events; both focus on applying each step to real work rather than demoing tools.
For brands planning a campaign
A brand can evaluate two collaboration formats: co-host or sponsor a hands-on cohort workshop on agent building, or partner on a curated practitioner event where named operators present measured results. The event is invitation-limited and mixes talks with small round tables.
A post promotes the Scalantec KI Summit, names Moritz Heininger as a participant, and describes the session on turning CRM leads and LinkedIn engagement into qualified pipeline.
Read more
Ask whether they had a formal business relationship with the event and whether the promotion was paid.
Moritz Heininger writes in German about how companies adopt AI, with a recurring focus on enablement, agent design and the operational consequences of new model capabilities.
Read more
Start with the posts that treat an AI agent as a new hire and the one-day productivity log, then follow the thread into agent-ready software procurement, workforce sequencing and frontier-model security.
The practical material is strongest in the checklists, course structure and operator events; shorter announcement posts offer current context rather than a full method.
For adjacent perspectives on German-language B2B leadership and hiring practice, compare Melissa Schröder on leadership and recruiting, Marcel Eberhardt on proposals and cultural debt, and Johannes Becker on structured recruiting speed.
Both write in German about how workplace conditions shape performance, drawing on everyday work rather than abstract theory.
Read more
Schröder’s posts on leadership language and lost motivation share the practical, people-focused framing of Heininger’s argument for enablement.
Schröder focuses on recruiting, leadership behaviour and motivation; Heininger on AI agents, enablement programmes and software strategy. Readers seeking hands-on AI adoption will find Heininger more specific, while Schröder offers more on day-to-day management and hiring culture.
Both use structured commercial examples and operational frameworks from a B2B services context.
Read more
Eberhardt’s proposal structure and cultural-debt breakdown echo Heininger’s agent-onboarding checklist and one-day productivity log: each turns a business problem into a step-by-step model with concrete costs or outcomes.
Eberhardt focuses on proposal quality, offer-to-win rates and the cost of poor hiring; Heininger on AI agents, enablement and agent-ready software. Eberhardt is a useful comparison for readers seeking sales-process depth rather than AI implementation guidance.
Both write from an operator’s perspective about building teams and processes, with numbers to back them up.
Read more
Becker’s breakdown of 108 applications, video filters and a fast hiring process parallels Heininger’s use of concrete metrics and staged workflows in AI adoption.
Becker focuses on recruiting speed and sales-team growth; Heininger on AI enablement, agents and software procurement. Becker suits readers exploring hiring-process design, while Heininger addresses how companies should train people and deploy AI systems.
Moritz Heininger’s posts support a LinkedIn placement for a German-speaking business audience.
Read more
For a B2B SaaS campaign on AI enablement and agent adoption for mid-market clients, their content addresses the right subject in the right language, and profile-level reach is substantial though mid-scale: roughly 21,600 LinkedIn followers as of 9 October 2026.
The posts publish no data on followers’ geography, age, gender, employer, job function or seniority, so ask for an audience survey or LinkedIn follower demographics before treating this as a verified fit against a specific target list.
The content speaks to people who run, sponsor or buy AI programmes inside companies, rather than practitioners looking for tool tips.
The seven LinkedIn posts were published between 8 April and 5 October 2026 and are written in German.
They argue that companies fail with AI because teams were never taught to use it; that an AI agent should be onboarded like a new hire, with a job description, access rights and a probation period; that headcount cuts attributed to AI are often reversed later; that SaaS interfaces will move behind APIs and MCPs as agents take over the front end; and that security capabilities arrive as a by-product of better coding models.
One post announces an eight-week Agent Builder course aimed at people already using AI daily but stuck in the chat window.
The posts also promote a Berlin summit with 60 curated seats described as Sales, RevOps and Founder, and describe a working day that included an event concept for a leadership dinner with 25 executives.
These examples point to commercial and leadership readers in the DACH market.
Account-level reach indicates how many people could see a placement, but not whether they are mid-market software buyers specifically.
Ask for recent post-level results for comparable topics and formats — impressions, reactions, comments, reshares and follower growth over the last three to six months, on posts about agent adoption and enablement rather than event promotion.
Platform demographics can speak to geography, age and gender; only a survey or another documented source can speak to employer, job function and seniority.
If neither exists, plan the engagement as a reach-and-topic test and measure it against the campaign’s own qualification criteria before committing to a larger programme.
LinkedIn followers
21,588
LinkedIn profile snapshot accompanying the reviewed posts (https://www.linkedin.com/in/moritzheininger) · 2026-10-09
Publication dates of posts
7 posts between 8 April 2026 and 5 October 2026
Publication dates recorded for the LinkedIn posts · 2026-10-09
Moritz Heininger’s posts do not document a brand partnership or sponsorship offer.
Read more
The seven LinkedIn posts cover AI enablement, agent adoption, sales operations and AI-related job changes; none includes a media kit, rate card, disclosed sponsored placement or brand inquiry route.
They do show the commercial work Heininger runs directly.
One post (15 July 2026) describes an eight-week Agent Builder cohort with weekly input sessions, Friday consulting and one-to-one coaching on a participant’s own project, starting with a paid pilot group.
Another (22 June 2026) shows them taking a speaker slot at the Scalantec KI Summit in Berlin, a curated 60-seat event for sales, revenue operations and founders, alongside operators presenting real numbers and use cases.
Both are Heininger’s own programs and appearances, not published sponsor packages.
For a buyer planning a talk or workshop on practical AI-agent adoption, the posts show the subject matter and delivery rhythm of their cohort, but not the price, availability, usage rights or booking route.
Ask Heininger directly to confirm commercial terms.
A B2B SaaS company selling to mid-market clients and building AI enablement programmes or agent strategies around its own product is a good match for Moritz Heininger.
Read more
Their posts cover the campaign's subject and already contain the structure a client would want delivered.
The case rests on four points.
First, the enablement diagnosis: in one post, they challenge the recurring finding that AI produces hardly any measurable efficiency gains by asking which companies were actually measured.
Their counter-example is a nine-hour working day covering ten projects — an event concept for a leadership dinner with 25 executives, a personalised CRM-matched email campaign, a landing page in corporate design, a fix between two tools, a competitive-intelligence analysis, a consultant profile, a podcast launch plan, and a study across 12,477 companies with an interactive report, five posts and 25 graphics — plus preparation for three AI training sessions.
The bicycle comparison follows: putting bikes in an office and concluding after three months that nobody cycles faster, when nobody was taught to ride.
Their conclusion is that companies do not have an AI problem; they have an enablement problem.
That is the same problem the client's programme is meant to solve, making the framing practical rather than abstract.
Second, the agent-onboarding framework.
One post treats a first AI agent as a new hire and works through the missing pieces in order: a job description naming the one task, the accountable department and a three-month success measure; accounts with correct permissions for ERP, CRM and Drive; a ramp from suggestions to approved actions to independent work; probation tested on 50 real cases from the previous year, including hard ones; rules covering data protection, works council and the EU AI Act; placement in the tools the team already uses; feedback reviews; and a true cost per completed task, including control and maintenance.
The post cites a Gartner expectation that more than 40 percent of agent projects will be discontinued by the end of 2027 and argues that model choice matters about as much as a new colleague's laptop.
Third, an agent-readiness buying test.
In the headless-SaaS argument, dashboards move behind the scenes because agents need clean APIs and MCP connections to structured, machine-readable data.
Power shifts to whoever becomes the central data store for a function or company; once a vendor opens up as a data store, well-funded agent startups can pull out the context and bundle a free version.
The practical advice for mid-market buyers is to check how agent-ready any new software is and buy something else if the answer is only that work is in progress.
For a SaaS vendor whose product is part of that conversation, this is on-topic and uncomfortable in the right way.
Fourth, formats that already exist.
They run an eight-week Agent Builder cohort built in six layers — set up an agent, give it context about how you work, connect it through MCP to mail, calendar and CRM, turn recurring tasks into skills, add automations that run without a human trigger, and build an individual app — with Monday input, Friday consulting and one-to-one coaching.
It is aimed at people already using AI daily and stuck in the chat window.
They also took a speaker slot at the Scalantec KI Summit in Berlin, a curated 60-seat event for sales, revenue operations and founders, where operators presented measured cases, including a lead-scoring rebuild that produced 50 percent more meetings at the same volume and a three-person team at $3M ARR.
The fit is strongest for German-language campaigns aimed at DACH commercial and leadership readers.
The LinkedIn audience is mid-scale — about 21,600 followers as of 9 October 2026 — but its geography, job functions and seniority are unverified.
Ask for audience data before buying against a specific profile, and agree commercial terms directly: their posts state no sponsorship terms, rate card or booking route.
This is a subject-and-format fit, not a verified audience fit.
When to choose a specialist
A campaign that needs verified audience delivery in English or a published sponsorship product is a poor fit based on Moritz Heininger's posts.
Read more
The clearest example is a US or UK consumer brand, or a developer-tools company marketing in English to an engineering audience in the United States, seeking a paid sponsored-content series with guaranteed impressions against documented demographics, a media kit and a rate card.
The posts are all in German and aimed at a DACH business audience; none addresses an English-language or US audience.
They include no media kit, rate card, published sponsored placement, sponsorship policy or brand inquiry route, so there is no commercial product to book.
Audience information stops at about 21,600 LinkedIn followers as of 9 October 2026, with no engagement figures, follower demographics or survey.
A count tells a buyer how many people could see a placement, not whether they match a consumer or developer target list.
The formats shown are text posts only: no video, podcast recordings, long-form articles or transcripts appear, leaving a campaign built around produced video or a recurring audio show without a format to assess.
The content also works against the consumer campaign.
Their posts address enablement inside companies, agent onboarding, software procurement, headcount sequencing, security patch cycles and sales pipeline — commercial and operations material with no consumer-facing angle.
A brand can still test the channel with a small placement, but should treat the first engagement as a measurement exercise, not buy against a promised audience profile.
Ask for recent post-level results on comparable topics and an audience survey before committing.
A worked campaign example
The client is a B2B SaaS company that wants to build AI enablement programmes and agent strategies for its mid-market clients, and needs a talk or workshop on practical AI agent adoption. A workable format: a 60- to 90-minute leadership talk, followed by a half-day working session for the client's enablement, RevOps or customer-success team, using the client's own mid-market accounts as material.
Open with the enablement diagnosis.
The nine-hour, ten-project day and bicycle comparison reframe a stalled AI programme: the constraint is taught skill, context and workflow knowledge, not model choice.
This addresses the client's most common customer objection — that AI has not delivered measurable gains — without dismissing it.
Second, deliver the agent-onboarding checklist as a list the client's customers can apply before a pilot starts.
Define one task, one accountable owner and a three-month measure.
Set accounts and permissions for CRM, ERP and Drive, noting that outdated documents make even a strong performer produce errors.
Ramp from suggestions to approved actions to independent work.
Test probation on 50 real cases from the previous year, including difficult ones, benchmarked against what people already achieve.
Clarify data protection, works council and EU AI Act questions in advance, and name who notices a manipulated email.
Place the agent in Outlook, Teams and CRM, with an honest conversation for anyone worried about their job.
Review what the agent decided, why and where it underperforms.
Measure cost per completed task, including control and maintenance, before the start.
Close with the Gartner expectation that more than 40 percent of these projects will be discontinued by the end of 2027, and why that makes the first five items non-optional.
Third, assess agent-readiness across the client's own stack.
Turn the headless-SaaS argument into a procurement test the client can run with customers: for each tool, how machine-readable and structured is the data, who owns it, and does the provider's pricing resist a shift away from seats?
The advice to mid-market buyers is to walk away from software whose only answer on agent-readiness is that work is in progress.
For a SaaS vendor, this positions its product against weaker competitors and gives the sales team a concrete question to ask.
Fourth, use the workshop for a hands-on version of the Agent Builder curriculum, compressed to half a day and applied to real work.
Set up an agent, give it context about how the participant works, connect it via MCP to mail, calendar and CRM, turn one recurring task into a reusable skill, add one automation that runs without a human trigger, and finish with a sketch of a small custom app.
Retain the cohort rhythm described in Moritz Heininger's posts — input session, consulting slot, one-to-one coaching — so participants leave with a working layer rather than a demo.
For executive attendees, close with the security example.
The numbers they report from Anthropic's announcement — 181 successful exploits for the new model against two in several hundred attempts for its predecessor, a 27-year-old OpenBSD bug, a 17-year-old FreeBSD remote code execution bug, a 16-year-old FFmpeg bug, and working exploits produced overnight by engineers without security training — make the case that patch cycles measured in weeks instead of hours are a management decision, not a technical detail.
It gives executives a reason to stay in the room.
Participants can leave with a filled agent job description, a written access and permissions list, a 50-case test plan, a cost-per-task model and a ranked shortlist of workflows by effort against payoff for the first pilot.
Their posts state no booking terms, availability or usage rights; agree those directly before the campaign goes out.
Since audience composition is unverified, measure the first engagement against the client's own qualification criteria rather than planning a scaled programme.
Marcel Eberhardt is a better choice when the client's bottleneck is the sales proposal itself, not AI adoption.
Read more
The posts document a shift from a 10% to an 87% offer-to-win rate and from €10k to €68k revenue per proposal by treating the proposal as the main selling tool, and list the assumptions that got in the way.
A B2B professional-services firm that wins work through proposals and needs a workshop on proposal structure, project understanding, approach, success factors and commercial offer would get a directly applicable framework.
The second post adds a cost model for mis-hires and cultural debt, useful when the client's problem is team composition rather than agent onboarding.
Marcel Eberhardt's posts focus on sales proposals and hiring culture, not AI enablement, AI agents, agent onboarding, agent-readiness, or any AI adoption framework. For an AI enablement campaign, ask for practical agent-adoption material before booking.
Melissa Schröder is a better choice when the client's programme needs a leadership or retention component rather than practical AI agent adoption.
Read more
The posts give a ten-sentence framework for everyday leadership — acknowledging good work, asking for input, admitting mistakes, inviting disagreement, asking what someone needs, trusting decisions without taking them back, and protecting time off — alongside the argument that workplace conditions, rather than a lack of motivation among employees, usually damage motivation. A B2B SaaS company whose mid-market clients are struggling with manager behaviour, team motivation, or retention after an AI-driven reorganisation could use this material in a complementary leadership workshop.
Melissa Schröder's posts contain no AI enablement, agent strategy, or technical adoption content.
They do not address building agent workflows, assessing agent-readiness, or running an AI pilot.
For a talk or workshop on practical AI agent adoption, ask for an agent-onboarding framework and an agent-readiness buying test before booking.
Johannes Becker is a better choice when the bottleneck is hiring for an AI or sales team, not adopting AI.
Read more
The posts detail a structured recruiting process: 108 applications in three days, 15 candidates selected, a six-question video task, structured interviews, a final exercise where candidates present the business model, and a signed contract 23 days after the application arrived.
A SaaS company recruiting AI-literate sales or customer-success staff and seeking a fast, video-filtered process could adapt it directly.
The second post explains why a high application volume is not the same as a good shortlist, useful for any hiring manager facing an AI-driven rush of applicants.
Johannes Becker's posts cover recruiting-process design for a sales team, not AI-agent deployment. For a campaign on practical AI agent adoption, ask for an agent-onboarding checklist, an agent-readiness procurement test, or an Agent Builder curriculum before booking.
This page covers seven LinkedIn posts by Moritz Heininger, all written in German.
They cover AI enablement and productivity measurement, onboarding a first AI agent, agent-ready software and data strategy, AI-attributed job cuts and the rehiring that follows, an autonomous exploit-finding model release, an eight-week agent-building course, and a Berlin sales-AI summit.
The posts do not show publication dates, so they are organised by theme rather than as a timeline.
A profile dated 9 October 2026 shows about 21,600 LinkedIn followers.
No engagement figures, follower demographics or audience survey are published, so statements about who reads the work rest on the content itself and are marked as unverified.
The posts do not document a media kit, rate card, sponsorship policy or booking route, and no information is included here about any other channel, long-form article, video or podcast.
Comparisons with Melissa Schröder, Marcel Eberhardt and Johannes Becker draw on those creators' own published posts.
Where the posts are silent — audience composition, commercial terms, career history — this page says so rather than filling the gap.
