Napkin reads text you have already written and offers diagrams, flowcharts, mind maps and infographics to match — no prompt engineering involved. Everything it generates stays editable, and it exports to PPT, PDF, PNG and SVG.
Used this tool? Rate it
Used this tool? Rate it
Napkin turns written text into visuals — diagrams, flowcharts, mind maps, infographics and data charts — and it does so from writing you already have rather than from a prompt you compose for the occasion. That distinction is the whole product. Most generative tools ask you to describe the picture you want; Napkin asks you to paste the paragraph you already wrote, then offers visuals that match what it says. Napkin Visuals handles the diagram side, while Napkin Slides, still marked Beta, turns an idea into a complete, ready-to-present deck.
The company behind it is not obvious from the branding: the terms of use name Second Layer, Inc. as the company that owns the platform, operating from Los Altos, California. Founders Pramod Sharma and Jerome Scholler previously built the AI gaming company Osmo, and the team is small, spread across the US, France and India. TechCrunch reported the company leaving stealth in August 2024 with $10 million in seed funding from Accel and CRV, at which point it had ten employees, and Inc. reported that Osmo, launched in 2013, was worth $120 million by the time it was acquired in 2019.
The origin story explains the design. The founders first tried to recreate two people sketching ideas on paper, digitally, and found that starting from drawing was too high a barrier for the people they wanted to help. Only when large language models matured did a text-first entry point become viable. Sharma described the philosophy as a generation-first experience where editing complements the generation — you get something immediately, then shape it, rather than facing a blank canvas.
A note on the name before you go looking for reviews: several unrelated products are also called Napkin. This page is about napkin.ai. napkin.one is an idea-collection app and napkin.io is a serverless API platform — neither is related to this product.
Napkin fits poorly for professional designers who need precise control over composition, for anyone whose brand demands a distinctive visual identity, for data analysis proper as opposed to illustrating figures already stated in text, and for mobile-first workflows.
Napkin runs on credits rather than seats alone, and the crucial mechanic is how those credits are consumed: roughly 1 credit is charged per word selected for generation. Cost therefore scales with how much text you highlight, not with how many visuals you try — a model that rewards selecting precisely and punishes illustrating whole pages indiscriminately. The company notes this may change for more complex outputs in future.
Three tiers are on offer. Free gives 500 AI credits per week, unlimited editing, unlimited file imports and unlimited PNG and PDF export, but stamps Napkin branding on the visuals. Plus at $9 per person per month raises that to 10,000 credits monthly, adds unlimited PPT and SVG export, three brand styles and removes the branding. Pro at $22 per person per month provides 30,000 credits along with exclusive designs, unlimited custom branding, uploaded fonts and optional credit top-ups. Annual billing saves 25 percent, and an enterprise plan is available through sales. There are currently no non-profit or student discounts, with the company pointing to the free plan instead — worth noting given how much of the testimonial base is educational.
On the administrative side, you can upgrade, downgrade or cancel at any time, and cancelling leaves paid features available until the end of the billing period. Upgrades are charged immediately and prorated; downgrades take effect at the end of the cycle. Failed payments are retried up to eight times over a month before the account drops to free. Reassuringly, your content remains yours regardless of subscription status, and cancelling still leaves you access to everything you created — you lose the paid features, not the work. Team seats are charged per member who accepts an invitation.
The honest framing is that Napkin occupies a narrow but real niche: the gap between "I have written something" and "I need a picture of it", where you would otherwise have made nothing at all.
Output quality depends heavily on how specific your text is, and independent testing found real failure modes. In TechCrunch's review at launch, the reviewer found that with vague input Napkin "grasps at straws", producing visuals not grounded in the text and in one case inventing pros and cons outright, and described the output as following a generic, homogenous design language. Two caveats belong in both directions: this is one reviewer's hands-on experience rather than a statistical assessment, and it dates from the August 2024 launch, so the product has had time to improve. But it remains the most substantive independent evaluation available, and the underlying risk — a diagram that confidently misrepresents your text — is inherent to the approach rather than a bug that gets fixed.
You own the visual but not its parts. The terms are precise about this: you retain ownership of the generated visual but you do not own the individual components inside it — the icons, illustrations and templates remain Napkin's property. Commercial use of the output as a whole is explicitly permitted, including commercial resale, marketing materials and business presentations. What you may not do is extract an icon or template and use it separately. For most users this is irrelevant; for anyone hoping to harvest design assets, it is a hard stop.
Uniqueness is not guaranteed. Napkin does not guarantee that your output will be unique to you, and states plainly that similar or identical inputs may produce similar or identical outputs for other users. Given that the visuals are assembled from a shared library of design elements, two people writing about the same topic may well end up with near-identical diagrams — a genuine consideration if visual distinctiveness matters to your brand.
No warranty on results. The terms state that all results are provided "as is" and with "all faults", with no warranty as to accuracy, usefulness, completeness or intellectual property rights, and disclaim liability if a third-party AI service the product depends on becomes unavailable.
Data is used for training unless you opt out. The company's own wording is that Napkin will not use your data to train its models if you opt out of data sharing — which means the default is that it may. It does state that contracts with its AI subprocessors prohibit those third parties from training on customer data, which is a meaningful protection, but the first-party default is opt-out rather than opt-in. Anyone pasting confidential material should change this setting before starting.
Mobile cannot generate visuals. This is a genuine functional gap rather than a minor inconvenience, since generation is the entire point of the product. Presentations can be built on a phone; diagrams cannot.
Disputes go to arbitration, not court. Under the terms, disputes are resolved by binding arbitration rather than in court, with carve-outs for qualifying small claims and for intellectual property relief, and both sides must first meet by phone or video for an informal resolution conference before arbitration can begin. For context on where any dispute would be handled: Second Layer, Inc. is based at 626 Jay Street in Los Altos, California, the privacy policy took effect on 7 August 2025, data is processed in the United States, and the service is not intended for anyone under 13.
Credit costs scale with text length, which is easy to underestimate. Because charging is per word selected, a user accustomed to per-generation pricing may burn through an allowance faster than expected when illustrating long documents.
There is almost no public user-review data. Its Trustpilot profile was claimed in August 2026 but carries zero reviews, so there is no rating there to cite, and Product Hunt and G2 were both inaccessible behind bot protection during research. This page therefore cites no user rating at all, because none could be verified. Most search results for "Napkin AI review" are affiliate pages or AI-generated content-farm articles, several of which circulate user and funding figures that do not appear in any first-hand source.
Watch out for the name. Because several unrelated products share the name Napkin, reviews and comparisons found by searching the word alone frequently describe a different tool. Always check the domain.
The company is small and the funding is early-stage. Ten employees at launch and a single $10 million seed round is a normal profile for a startup at this stage, but it is worth weighing if you are considering standardising a team workflow on it.
Yes. The free plan gives 500 AI credits per week, unlimited editing, unlimited file imports and unlimited PNG and PDF export. Its main constraints are Napkin branding on the visuals, no PPT or SVG export, and standard icons only. For occasional use it is genuinely workable rather than a token trial.
Yes. The terms explicitly permit using generated output for commercial resale, marketing materials, scientific research, educational purposes and business presentations. The one restriction is that you cannot extract the individual components — icons, illustrations, templates — and use them separately outside the visual they appear in, since those remain Napkin's property.
No, and this is the product's central design choice. You supply text you have already written — typed, pasted or imported — and click the Autospark icon on the passage you want illustrated. The system reads the text and proposes matching visuals for you to choose among.
At roughly one credit per word selected for generation, so cost tracks the length of the text you highlight rather than the number of visuals you generate. Selecting a single paragraph rather than a full page is the main way to control spending. The company notes this may change for more complex features later.
PNG and PDF on every plan including free, plus PPT and SVG on Plus and Pro. SVG matters if you need the visual to stay vector for later resizing or editing in design software.
Only partly. You can view existing work and create or edit presentations on mobile, but visual generation is not available there, so the main feature requires a desktop browser.
By default it may be. The company states it will not use your data to train its models if you opt out of data sharing, so the opt-out is available but not the default. Separately, it says contractual agreements prevent its AI subprocessors from training their models on customer data.
Visuals and presentations can be produced in more than 60 languages regardless of your input language. The interface itself is available in six: English, French, Spanish, Italian, Korean and Portuguese.
The terms of use identify Second Layer, Inc. as the owner of the platform, based at 626 Jay Street, Los Altos, California. It was founded by Pramod Sharma and Jerome Scholler, previously of the AI gaming company Osmo, and raised $10 million in seed funding from Accel and CRV in August 2024.
No. At least three unrelated products use the name. This one is napkin.ai, the text-to-visuals tool. napkin.one is an app for collecting ideas, and napkin.io is a platform for building and deploying serverless APIs. Reviews of those products say nothing about this one.
