What is the 'Internet of Agents' mentioned in the deck?

Lakera uses this term to describe the next evolution of AI, moving beyond simple chatbots to autonomous agents that interact with each other and various services. Slide 3 illustrates this transition from 'Conversational AI' to 'Agents' and finally to a complex web of interconnected agents, which introduces new security risks like agent containment and transactional governance.

How does Lakera acquire its threat intelligence data?

A significant portion of their data comes from 'Gandalf,' a cybersecurity game they developed. According to Slide 7, the game has had over 1 million players and thousands of organizations participating. This crowdsourced 'red-teaming' provides Lakera with millions of attack data points, with 100k+ new entries added daily to their database (Slide 5).

What are the key technical claims for Lakera Guard?

The deck emphasizes ease of use and performance. Slide 5 claims developers can 'secure LLMs in minutes' with one line of code. Slide 6 highlights 'industry-leading protection' based on the PINT benchmark, low latency for real-time applications, and an integration time of less than 5 minutes.

Is there any information about Lakera's revenue or valuation?

No. The deck focuses on 'active deals' and 'interest' rather than hard revenue figures. Slide 10 shows a bar chart of monthly deal creation growing 3x in four months, but the Y-axis is unlabeled. The $20M Series A figure and 2024 date are reported by external sources, not stated within the slides.

Who is the target customer for Lakera?

The deck specifically targets large enterprises. Slide 9 notes that 34% of Fortune 100 companies have shown interest, and Slide 6 includes testimonials from a 'Fortune 200 Enterprise Tech Co' and a 'Security Engineer at major tech firm,' indicating a focus on high-value B2B contracts.

Lakera Pitch Deck: All 10 Slides + Teardown

See all 10 slides of the Lakera pitch deck — a 2024 Series A deck — with a slide-by-slide teardown of what the deck does well and where it falls short.

Lakera's Series A deck is a masterclass in narrative-driven fundraising for a technical product. Rather than leading with dense architecture diagrams, the deck establishes a new category—the 'Internet of Agents'—and positions Lakera as the essential security layer for this evolution. The company highlights a significant data moat: their viral game, Gandalf, which has engaged over 1 million players to create a massive threat intelligence database. This community-led growth is paired with enterprise validation, claiming interest from 34% of the Fortune 100. While the deck is light on specific f…

Key takeaways

The deck introduces the 'Internet of Agents' (IoA) as the next major platform shift requiring a new security paradigm (Slide 1).
Lakera positions its product, Lakera Guard, as an 'AI Application Firewall' that can be integrated in under 5 minutes (Slide 5).
A proprietary threat intelligence database is fueled by 100k+ new entries every day (Slide 5).
The company uses 'Gandalf,' a viral red-teaming game, to claim the title of the world's largest AI red team with 1M+ players (Slide 7).
Lakera claims to be growing its community of security professionals faster than any competitor, supported by search traffic charts (Slide 8).
Enterprise traction is quantified by interest from over 34% of Fortune 100 companies (Slide 9).
Sales velocity is highlighted by a chart showing a 3x increase in active deals between January and April 2024 (Slide 10).
The deck completely omits a team slide, a detailed roadmap, and a specific use-of-funds breakdown.

The Narrative: From Chatbots to Autonomous Agents

Lakera’s Series A deck, used to raise $20M in 2024, is built on a clear evolutionary thesis. It doesn't just pitch a tool; it pitches a solution for an inevitable future. By defining the 'Internet of Agents,' Lakera creates a sense of urgency around a new class of security threats that traditional firewalls are not equipped to handle. The deck is visually sparse but strategically dense, focusing on three pillars: a massive data moat, rapid enterprise adoption, and developer-centric simplicity.

Slide 1: Title Slide

The deck opens with the company name and a bold sub-headline: "AI security for the Internet of Agents (IoA)." The date is noted as May 2024. The branding is clean, utilizing a high-contrast visual style that suggests a modern, technical product. The term 'Internet of Agents' is the first piece of category creation, signaling that the company is looking beyond the current hype of simple LLM wrappers.

Slide 2: The New Era of AI

Slide 2 uses a quadrant-style graph to map the evolution of software. The X-axis represents "Increasing interfaces" (from Code to Audio), and the Y-axis represents "Increasing capabilities." Traditional software is relegated to a small box in the bottom left. The graph shows a trajectory through Conversational AI and Agents, culminating in the "Internet of Agents" in the top right. This slide establishes the scale of the opportunity, claiming AI "transforms every aspect of our economy."

Slide 3: The Security Gap

This slide identifies the problem. It contrasts "Today" (Conversational AI) with the "Future" (Internet of Agents). As AI moves from a simple human-to-app interface to a complex web of autonomous agents interacting with multiple services, the threat surface expands. The slide lists three critical needs: Agent and infrastructure security, Transactional governance, and Agent containment. It also includes news clippings of AI failures (e.g., DPD chatbot errors, Bing Chat prompt injections) to ground the theoretical risk in current reality.

Slide 4: The Platform Opportunity

Slide 4 is a transition slide that reinforces the core message: "Securing AI is the next big platform opportunity." It repeats the timeline from Conversational AI to the Internet of Agents, positioning Lakera as the enabler that allows this transition to happen securely. This is a classic 'picks and shovels' play for the AI gold rush.

Slide 5: Product and Data Moat

Slide 5 introduces "Lakera Guard," described as the "fastest and most accurate AI Application Firewall for enterprises today." The slide highlights three value propositions: securing LLMs in minutes, ease of deployment (cloud or on-prem), and a growing threat intelligence database. Crucially, it quotes a figure: the database contains "tens of millions of attack data points" and grows by "100k+ entries every day." A code snippet on the right demonstrates the simplicity of the API, showing how a few lines of Python can intercept a prompt injection.

Slide 6: Social Proof and Benchmarks

This slide focuses on validation. It mentions a "PINT benchmark score" and claims "Best prompt defense," "Strongest in operation," and "Lightning-fast" latency. While the specific percentages and millisecond counts are blurred in this version, the slide emphasizes an "integration time" of "<5 mins." Four testimonials from unnamed but high-profile sources (e.g., "Fortune 200 Enterprise Tech Co") reinforce the product's market fit.

Slide 7: The Gandalf Moat

Slide 7 reveals the source of Lakera's data advantage: a game called "Gandalf." The slide calls it the "world’s largest AI red team." It claims the game is used by "1M+ people and 1000s of organizations." By gamifying the process of breaking LLMs, Lakera has crowdsourced its threat intelligence. The slide shows a T-SNE plot of attack vectors, visually representing the diversity of the data they have collected. This is a powerful 'flywheel' argument: more users play the game, the database gets stronger, and the product becomes more effective.

Slide 8: Community Growth

Slide 8 claims Lakera is growing its community of security professionals "faster than any of our competitors." It shows a dashboard with search traffic and engagement metrics. While the specific numbers are blurred, the trend lines show a sharp upward trajectory in late 2023 and early 2024, suggesting that Lakera is winning the mindshare battle in the AI security space.

Slide 9: Fortune 100 Traction

This is a high-impact 'momentum' slide. It states: "And, we have had overwhelming interest from over 34% of Fortune 100 companies." This specific figure (34%) is used to demonstrate that the problem Lakera solves is a top priority for the world's largest organizations. It shifts the narrative from 'cool startup' to 'enterprise-grade necessity.'

Slide 10: Sales Velocity

The final slide shows a bar chart of "active deals" by month. The headline claims "3x more active deals since January" and that they are "closing leading brands in record time." The chart shows consistent growth from August 2023, with a significant spike in April 2024. This provides the 'why now' for investors, showing that the company is at an inflection point in its commercial journey.

What Lakera Does Well

Category Creation: By coining and centering the deck around the "Internet of Agents," Lakera moves the conversation away from being a commodity security tool to being a foundational platform for the next decade of computing.

Data Flywheel: The inclusion of the Gandalf game is brilliant. It explains how they can be better than incumbents (proprietary, crowdsourced data) rather than just claiming they are better. It turns a marketing asset into a core product advantage.

Developer Friction: The emphasis on "one line of code" and "<5 mins integration" directly addresses the primary objection in enterprise software sales: the difficulty of implementation.

What is Missing from the Deck

Team Slide: There is no mention of the founders or the engineering team. In a Series A for a highly technical cybersecurity product, the pedigree of the team is usually a major selling point. Its absence suggests this version of the deck was focused purely on market momentum and product-market fit.

Business Model: The deck does not explain how Lakera makes money. Is it per-seat, per-query, or a flat enterprise license? While investors at this stage often prioritize growth over immediate profitability, the lack of a pricing framework is a notable omission.

Competition: The deck claims to grow faster than competitors but never names them or explains why their technical approach (e.g., an 'AI Firewall') is superior to existing WAFs (Web Application Firewalls) or other AI security startups.

Founder Takeaways

Quantify your 'Unfair Advantage': Lakera didn't just say they have data; they cited "100k+ entries every day" and "1M+ players." Specificity builds credibility. · Use a 'Wedge' Strategy: Slide 5 explicitly identifies Lakera Guard as their "wedge into the market." Showing investors that you have a narrow, high-velocity entry point that leads to a larger platform play is highly effective. · Visualizing Momentum: If you don't have massive revenue yet, show the velocity of interest. Slide 10's chart of active deals and Slide 9's percentage of Fortune 100 interest are excellent proxies for future revenue. · Category First, Product Second: Spend the first few slides convincing the investor that the world is changing in a way that makes your existence mandatory. Lakera does this by framing the shift from chatbots to agents as an inevitable infrastructure change.

Frequently asked questions

What is the 'Internet of Agents' mentioned in the deck?: Lakera uses this term to describe the next evolution of AI, moving beyond simple chatbots to autonomous agents that interact with each other and various services. Slide 3 illustrates this transition from 'Conversational AI' to 'Agents' and finally to a complex web of interconnected agents, which introduces new security risks like agent containment and transactional governance.
How does Lakera acquire its threat intelligence data?: A significant portion of their data comes from 'Gandalf,' a cybersecurity game they developed. According to Slide 7, the game has had over 1 million players and thousands of organizations participating. This crowdsourced 'red-teaming' provides Lakera with millions of attack data points, with 100k+ new entries added daily to their database (Slide 5).
What are the key technical claims for Lakera Guard?: The deck emphasizes ease of use and performance. Slide 5 claims developers can 'secure LLMs in minutes' with one line of code. Slide 6 highlights 'industry-leading protection' based on the PINT benchmark, low latency for real-time applications, and an integration time of less than 5 minutes.
Is there any information about Lakera's revenue or valuation?: No. The deck focuses on 'active deals' and 'interest' rather than hard revenue figures. Slide 10 shows a bar chart of monthly deal creation growing 3x in four months, but the Y-axis is unlabeled. The $20M Series A figure and 2024 date are reported by external sources, not stated within the slides.
Who is the target customer for Lakera?: The deck specifically targets large enterprises. Slide 9 notes that 34% of Fortune 100 companies have shown interest, and Slide 6 includes testimonials from a 'Fortune 200 Enterprise Tech Co' and a 'Security Engineer at major tech firm,' indicating a focus on high-value B2B contracts.

Cover slide of the Lakera pitch deck — Series A 2024 — Lakera pitch deck, slide 1 (2024)

Lakera pitch deck: the facts

Company: Lakera
Year: 2024
Stage: Series A
Slides: 10
Sector: Cybersecurity
Deck type: Fundraising
Outcome: $20M raised
Headquarters: Europe (Switzerland)

Lakera pitch deck PDF

The full Lakera deck is embedded on this page and can be read slide by slide in the browser — no download or account required. Each slide is covered in the breakdown above.

What the Lakera pitch deck was used for

This deck is Lakera’s **Series A fundraising presentation in 2024**, used to secure a **$20M round** for its generative AI security platform.[1][2][3] It focuses on the shift from conversational AI to an “Internet of Agents” and positions Lakera as a real-time security layer for LLM and agentic applications backed by data from its Gandalf red‑teaming game.[1][3][11][14] The company operates in the cybersecurity / AI security sector, protecting enterprise generative AI apps from malicious prompts and other threats.[1][2][3] The lean 10‑slide deck was aimed at investors such as Atomico and existing backers to fund product scaling and market expansion in the US and Europe.[1][2][5]

Business model: Lakera provides **real-time generative AI security** for enterprises, screening prompts, model outputs, and agent tool calls to detect and block AI-specific attacks such as prompt injection, data exfiltration, and policy bypasses.[1][2][11]

Round: Series A.[1][2][3][4][5][8][11]
Raised: $20M Series A.[1][2][3][4][5][8][11]
Lead investor: Atomico.[1][2][3][4][5][11]
Investors: Atomico (lead).[1][2][3][5][4][11], Citi Ventures.[1][2][3][5][4], Dropbox Ventures.[1][2][3][5][4], Redalpine (existing investor participating).*[1][2][3][5][4][13][11]
Founded: 2021[11]
Headquarters: Zurich, Switzerland (with presence in San Francisco).[1][5][3]
Industry: Cybersecurity / AI Security (Generative AI and LLM application security).[1][2][3][11]

Year: 2024.[1][2][3][4][5][8][11]

Total funding: $30M total funding (including the $20M Series A in 2024 and prior $10M round in 2023).[1][3][5][13][11]

Use of funds as presented: Advance Lakera’s real‑time GenAI security platform, accelerate product development, and expand market presence particularly in the US and Europe, including strengthening its San Francisco and Zurich operations.[1][5][8]

What happened after the Lakera deck

Following its $20M Series A in 2024 and previous $10M round, Lakera grew its generative AI security business and was subsequently acquired by Check Point Software Technologies, with reports indicating a $300M purchase price.[7][11][13]

Raised **$10M** in earlier funding in 2023 led by Redalpine to develop its AI security platform.[13]
Raised **$20M Series A** in July 2024 led by Atomico with Citi Ventures, Dropbox Ventures, and Redalpine, bringing total
Expanded operations with presence in **San Francisco and Zurich**, focusing on US and European market penetration.[1][5]
Later **acquired by Check Point Software Technologies**, with media reports citing a purchase price around **$300M**, fo

What the Lakera deck got right

Compelling and timely problem statement. Lakera addresses a **high‑urgency, clearly defined problem**: securing generative AI and LLM applications against prompt injection and other AI‑specific threats, which is widely recognized in press and industry coverage.[1][2][3][11]
Distinctive data moat via Gandalf community. The company leverages **Gandalf**, described as the world’s largest AI red team and a viral prompt‑injection game with over one million users and tens of thousands of new attacks logged daily, to build a continuously updated threat database.[1][3][5][11][14]
High‑caliber investor syndicate. Lakera’s Series A round is **led by top‑tier VC Atomico** with participation from Citi Ventures, Dropbox Ventures, and Redalpine, providing strong external validation of both the market and execution.[1][2][3][4][5][11]
Forward‑looking positioning around agentic AI. The narrative around moving from conversational AI to an **Internet of Agents** positions Lakera as core infrastructure for agentic systems, differentiating it from basic chatbot security tools.[1][3][11][14]
Runtime enterprise‑grade protection focus. Public reporting highlights **real‑time, in‑line protection** for enterprise GenAI apps, indicating a product capable of operating at runtime rather than purely as a consulting or assessment service.[1][2][3][11]
Demonstrated fundraising momentum and strategic exit. Lakera’s prior **$10M funding round in 2023**, plus the $20M Series A, shows consistent investor support and runway to build out a robust platform before its later acquisition.[13][1][3][11][7]

What could have been stronger

Limited visibility on founding team in accessible materials. Public sources do not clearly state the founders’ names or detailed founding team backgrounds for Lakera’s Series A materials, limiting an investor’s ability to assess founding experience and prior exits from this deck alone.[11][13]
Sparse quantitative business traction in publicly described deck. None of the accessible funding announcements or profiles provide detailed **financial or usage traction metrics** (e.g., ARR, paying customers, retention), focusing instead on strategic narrative (Internet of Agents) and Gandalf engagement numbers, which may leave investors with unanswered questions about commercial traction.[1][2][3][5][8][11]
Go‑to‑market strategy not deeply detailed in public descriptions. Although external articles mention Lakera’s presence in both Zurich and San Francisco, publicly available deck‑related coverage offers limited clarity on its go‑to‑market motion (direct vs. channel, pricing, customer segments), which investors might find underdeveloped.[2][3][5][11]
Compliance and assurance credentials not emphasized. Security buyers often demand detailed compliance posture and certifications (e.g., SOC 2, ISO 27001); these are not referenced in any Series A coverage, suggesting the deck did not foreground such proof points even though they matter in enterprise security sales.[1][2][3][11]
Limited explicit competitive positioning visible externally. While Lakera’s platform is framed as uniquely data‑driven via Gandalf, public analyses of the Series A do not show a clear competitive landscape slide comparing Lakera to other AI security vendors, leaving differentiation somewhat narrative‑driven.[1][2][3][11]

How an investor would read this deck

What draws attention

Clear articulation of a **fast‑emerging problem**: prompt injection and other LLM vulnerabilities in enterprise AI applications, which investors recognize as a critical risk area.[2][3][11]
Strong **lead investor syndicate** (Atomico, Citi Ventures, Dropbox Ventures, Redalpine) validating market importance and Lakera’s position.[1][2][3][5][4]
Compelling **data advantage via Gandalf**, with over one million users and tens of thousands of attacks per day feeding models, creating a defensible moat.[1][3][5][11][14]
Positioning around the **Internet of Agents**, suggesting Lakera is building security infrastructure for agentic AI rather than only chatbots, aligning with forward‑looking investor theses.[1][3][11][14]
Evidence of **enterprise adoption and ecosystem ties** (e.g., Gandalf used in Microsoft security training, collaborations like NVIDIA NeMo agents red‑teaming).[1][3][5][11]

Risks that stand out

The **AI security market is nascent and rapidly evolving**, with uncertainty around which solutions will become standard and how spending will scale in downturns.[2][3][11]
Dependence on **Gandalf’s community engagement** for threat data could pose a risk if user growth slows or if regulatory changes constrain data collection.[1][3][5][11][14]
Lakera’s success is linked to the **adoption trajectory of agentic and generative AI in enterprises**; slower or more regulated adoption could delay revenue growth.[2][3][11]
The company operates in a **crowded emerging category** (AI firewalls, prompt security, red‑teaming platforms), which may lead to pricing pressure and feature parity over time.[2][3][11]
Enterprise buyers may demand rigorous **certifications, audits, and integrations**; failing to meet these expectations could limit penetration into highly regulated sectors.[1][2][3][11]

Questions this deck invites

What are Lakera’s current **revenue levels, growth rate, and number of paying enterprise customers**, and how are contracts structured (usage‑based, per‑application, seats)?[1][2][3][8][11]
How does Lakera technically ensure **low‑latency, high‑reliability inline protection** for LLM and agent calls at enterprise scale, and what are its SLAs and performance benchmarks?[1][2][3][11]
How sustainable is Lakera’s **Gandalf‑driven data moat**—what are the legal, privacy, and consent frameworks governing use of attack data from public users and enterprises?[1][3][5][11][14]
What does the **competitive landscape** look like for GenAI security (other AI firewalls, red‑teaming platforms), and what unique capabilities or IP give Lakera durable differentiation?[1][2][3][11]
What are the key **go‑to‑market channels and unit economics** (sales cycles, CAC, churn) for selling GenAI security into large enterprises and regulated industries?[2][3][5][11]

What founders can take from the Lakera deck

A **lean 10‑slide deck** can successfully raise a large Series A when the problem is urgent, clearly articulated, and backed by recognizable proof points like user counts and enterprise references.[1][2][3]
Framing the company around a **structural shift** (from conversational AI to an Internet of Agents) helps investors see the opportunity as infrastructure‑level rather than a point solution.[1][3][14]
Demonstrating a **data moat**—in this case Gandalf’s massive red‑teaming corpus—can be more persuasive than generic claims of better models or features.[1][3][5][11][14]
Aligning with **top‑tier investors** (Atomico, Citi Ventures, Dropbox Ventures, Redalpine) provides strong external validation that can reinforce the deck’s narrative about market importance.[1][2][3][5][4]
Connecting a **community product** (Gandalf game) directly to the core enterprise offering (security detectors) shows how virality can be converted into defensible enterprise value.[1][3][5][11][14]
Clear articulation of **use of funds**—accelerating product development and international expansion—helps investors understand how capital will convert into scale.[1][5][8]

Lakera pitch deck: common questions

What does Lakera do?

Lakera is a Zurich‑based AI security startup that builds **real‑time security for generative AI applications**, protecting LLMs and agentic systems from prompt injection, data exfiltration, and other AI‑specific threats.[1][2][11]

How much did Lakera raise in its Series A and who invested?

Lakera raised **$20M Series A funding in July 2024**, led by European VC Atomico with participation from Citi Ventures, Dropbox Ventures, and existing investor Redalpine, bringing total funding to **$30M**.[1][2][3][5][4]

What is the funding from this deck being used for?

The **Series A funding** is used to advance Lakera’s real‑time GenAI security platform, accelerate product development, and expand its go‑to‑market presence, especially in the US and Europe (including its San Francisco presence).[1][5][8]

What is Gandalf and how does it relate to Lakera’s product?

Gandalf is Lakera’s **viral AI security game and red‑teaming community**, described as the world’s largest AI red team, with over one million users and tens of thousands of new attacks logged daily.[1][3][5][11][14] The data from Gandalf feeds Lakera’s detectors, keeping its security models up to date against emerging prompt attacks.[1][3][11]

What happened to Lakera after the Series A round?

According to later reporting, Lakera was **acquired by Check Point Software Technologies**, with media reports citing a purchase price around **$300M** and noting Lakera had raised $30M across its 2023 and 2024 rounds before the acquisition.[7][11]