Potion
Clone your face, voice, and gestures, then never record again
Potion is an AI video generation platform for sales outreach that builds a digital clone of your face, voice, and gestures from a short capture, then generates hundreds or thousands of personalized videos in which that clone delivers a different script to each prospect, complete with accurate lip sync, branded video pages on your own domain, viewer analytics, and native connections into HubSpot, Outreach, Apollo, Salesloft, and roughly 50 other sales tools.
Overview
Potion was founded in 2020 in San Francisco by Kanad Bahalkar and Andrew Boos and has raised around $3.5M in seed funding. Bahalkar came from Knit Health, a computer-vision sleep-monitoring startup acquired by Google, and stayed on the Google Health team before starting Potion. That background matters, because Potion is fundamentally a computer vision company that decided the best commercial application of generative video was making salespeople stop recording.
The premise is blunt. Personalized video works, but recording 300 of them a week does not scale, so Potion removes the recording. You capture yourself once, the model learns how you speak, emote, and gesture, and from then on you type a script or record a template and Potion generates videos of you saying whatever the prospect needs to hear. Lip sync is matched precisely to the audio and gestures are synchronized so the output does not have the uncanny stillness of a talking-head avatar.
The pricing is straightforward and expensive relative to the small end of this market. Starter is $99 a month per workspace and includes 750 dynamic AI videos and three full AI video templates per month. Professional is $299 and makes video generation unlimited while adding Zapier webhooks, multi-user workspaces, roles and permissions, custom branding and domains, and priority support. Enterprise is quote-only and is where whitelabel, SOC 2 documentation, and TLS 1.2 encryption commitments live. Every tier offers a seven-day free trial and there is no free plan. Videos are capped at five minutes.
The uncomfortable question with Potion is not whether the technology works. It does, and 29-plus language support means you can prospect into markets where you do not speak the language. The question is what happens when a prospect realizes the person who appeared to record a video just for them never existed in that moment. Potion's customers include SAP, Gorgias, and Orum, so plenty of sophisticated teams have decided the tradeoff is worth it. Buyers who have not thought that through should think it through before signing.
Best for
Sales teams running high-volume personalized video where recording individually is arithmetically impossible, particularly outbound organizations already on Outreach, Salesloft, or Apollo who want video variants generated per prospect and are comfortable sending a synthetic likeness.
Not the right fit for
- Anyone whose selling proposition is authenticity; the entire mechanism is a video of you that you did not record, and a prospect who works that out may trust you less than if you had sent plain text.
- Internal team communication; $99 a month per workspace buys face cloning nobody on your team needs, and Loom at $15 or Bubbles at $12 will do the job better.
- Low-volume senders; if you send fewer than fifty videos a month, recording them yourself in Bonjoro or Loom costs less money and produces a better result.
- Teams needing long-form video; Potion caps videos at five minutes, so demos, training, and walkthroughs are out of scope entirely.
- Regulated industries where synthetic representations of a named employee raise disclosure obligations; SOC 2 sits on the quote-only Enterprise tier and the compliance conversation around cloned likenesses is yours to have, not Potion's.
How it works
- 1
You capture yourself once, giving Potion enough footage to learn your face, voice, expressions, and how you gesture while speaking. This is the only recording session the product requires and everything afterwards is generated from it.
- 2
You define the video. That can be typing a script from scratch, recording a video template where only certain segments vary, or converting an existing screen recording into an AI-delivered video. Most sales use is the template path: a fixed body with a generated personalized opening.
- 3
You connect a data source. Potion integrates with HubSpot, Outreach, Apollo, Salesloft, and roughly 50 other sales, marketing, support, and CRM tools, and its export-less integrations mean the prospect list stays where it lives rather than passing through a CSV.
- 4
Potion generates one video per prospect. Your clone delivers each variant with lip sync matched to the generated audio and gestures synchronized to it, in any of 29-plus languages if you are prospecting internationally.
- 5
Each video is delivered on a custom branded page served from your own domain on Professional, with themeable player styling and optional stickers, emoji, GIFs, and images layered on. Viewer analytics track engagement and feed back into your sequencer or CRM.
Feature breakdown
25 features in 5 modulesCloning and generation
The computer vision core, and the entire reason the product exists.- Face cloning
- Capture yourself once and Potion builds a digital likeness that can deliver any script, so a salesperson appears in hundreds of videos they never sat down to record.
- Voice cloning
- Your voice is reproduced so generated scripts sound like you rather than like a text-to-speech reader dubbed over your face.
- Gesture and expression cloning
- Potion learns how you move while you speak and reproduces it, which is the difference between a convincing clip and the frozen-shoulders look that gives most avatar tools away.
- Accurate lip sync
- Lip movement is matched precisely to the generated audio, which is the single most scrutinized quality signal in synthetic video and where cheaper tools fail visibly.
- Script-to-video generation
- Type a script and get a finished video of your clone delivering it, with no recording session required for each new message.
- Template-based generation
- Record a base video and mark the segments that vary per prospect, so the majority of the runtime is genuinely you and only the personalized parts are generated.
- Screen recording conversion
- Turn an existing screen recording into an AI-delivered video, so a demo walkthrough can be re-narrated per audience without re-recording.
- 29-plus language support
- Generate the same message in more than 29 languages, which lets a single seller prospect into markets they do not personally speak. This is the capability no human-recorded tool can match.
Personalization and scale
Turning one clone into thousands of individual messages.- Dynamic AI videos
- 750 per month on Starter and unlimited on Professional, each generated as a distinct video for a distinct prospect rather than a merge field on a shared asset.
- AI video templates
- Three full templates per month on Starter, defining reusable structures for different campaigns, personas, or product lines.
- Bulk generation from a list
- Feed in a prospect list and generate the whole set in one run, which is the operational difference between video as a channel and video as a favor.
- Personalized variables
- Name, company, role, and other fields drive what the clone actually says, spoken in your cloned voice rather than inserted as on-screen text.
Video pages and branding
Where the prospect lands, and how much it looks like your company.- Custom branded video pages
- Each video sits on a page carrying your branding rather than a generic player, with custom domains available on Professional.
- Custom video player themes
- Style the player itself to match your brand rather than accepting a vendor default.
- Stickers, emoji, GIFs, and images
- Layer visual elements onto the video page to make a generated message feel less like a machine output and more like a designed touch.
- Custom domains
- Professional serves video links from your own domain, which keeps a cold email from advertising a third-party vendor in the link preview.
- Whitelabel video AI
- Enterprise removes Potion from the experience entirely, which matters for agencies reselling the capability.
Integrations and workflow
Built to disappear into an existing sales stack.- Fifty-plus sales and CRM integrations
- HubSpot, Outreach, Apollo, and Salesloft among roughly 50 sales, marketing, support, and CRM tools.
- Export-less integrations
- Prospect lists stay in the system of record rather than being exported to CSV and reimported, which removes the most common failure point in bulk video workflows.
- Zapier webhooks
- Available on Professional for anything not natively connected, including triggering generation from arbitrary events.
- Multi-user workspaces
- Professional adds shared workspaces so a team runs from common templates rather than each seller building their own clone workflow.
- Roles and permissions
- Control who can generate against a given clone, which matters when the clone is a specific named person's likeness.
Analytics and support
Adequate for measuring a channel; not a revenue intelligence platform.- Video analytics
- Track who watched, how far they got, and what they clicked, per video and per campaign.
- Priority support on Professional
- Faster response than the Starter tier, plus the strategy consultations and team coaching that come with Enterprise.
- Seven-day free trial on every tier
- Enough to build a clone and test output quality, though not enough to measure a reply-rate result on a real campaign.
Use cases
4 documentedOutbound team sending 2,000 touches a month
Text sequences have stopped working, video demonstrably lifts reply rates, and nobody has the hours to record two thousand individual videos.
One clone capture, a template with a generated personalized opening, and Professional's unlimited generation covers the whole volume with a distinct video per prospect at a flat $299.
Founder selling into three language markets
The product sells in Germany and Spain as well as the United States, and the founder speaks only English, so localized video is impossible.
Potion generates the same message in 29-plus languages delivered by the founder's own cloned face and voice, which no human-recorded tool can do at any price.
Sales leader standardizing rep messaging
Eight reps each record their own version of the pitch, quality varies wildly, and coaching the bad ones is a permanent tax.
Approved templates in a shared Professional workspace mean every generated video follows the same structure, with roles and permissions controlling who can generate against which clone.
Agency running video outreach for multiple clients
Each client wants video in their own brand, from their own domain, without a third-party vendor visible anywhere in the experience.
Enterprise whitelabel plus custom domains and per-client workspaces let the agency deliver generated video as its own capability rather than as a reseller.
Pricing
from $99 per month per workspace (Starter)Per-workspace monthly subscription with a generated video allowance on the entry tier and unlimited generation above it; no per-seat pricing on published tiers.
| Plan | Price | Includes |
|---|---|---|
| Starter | $99 per month per workspace |
750 videos a month is real volume, but three templates is tight for a team running more than a couple of campaigns. |
| Professional | $299 per month per workspace |
Unlimited generation at a flat $299 for a whole workspace is aggressive pricing against per-seat competitors once you have more than three sellers. |
| Enterprise | Custom quote only |
SOC 2 documentation only exists here, which is a problem for a mid-market buyer with a security review and a $299 budget. |
Billing notes
- Pricing is per workspace rather than per seat on the published tiers, which is unusually favorable for teams and means a five-person team on Professional pays about $60 a seat.
- Starter's 750 dynamic AI videos per month is the meter; exceeding it triggers upgrade prompts or usage management rather than automatic overage billing.
- Videos are capped at five minutes regardless of tier, which rules out demos, training, and anything long-form.
- The seven-day trial is the only free access; there is no permanent free plan on any tier.
- SOC 2 compliance and TLS 1.2 commitments are Enterprise-only, so a security questionnaire will push a self-serve buyer into a sales conversation.
Value assessment: Potion is priced for volume and the arithmetic only works if you have it. At $99 for 750 generated videos, each personalized video costs about thirteen cents, which is absurdly cheap compared with a seller's time. At $299 for unlimited generation across a multi-user workspace, a five-person outbound team is paying roughly $60 a seat for capability that Vidyard charges $59 plus a $24 add-on for and caps differently. That is good value. The problem is the floor. If you send eighty videos a month, you are paying $99 for something Bonjoro does for $15 and Loom does for $18, and you are sending synthetic video instead of real video to get there. Potion is excellent value above roughly 300 sends a month and poor value below it, and there is no free plan to help you find out which side of that line you are on.
Strengths & limitations
Strengths
- Gesture and expression cloning alongside face and voice is what separates Potion's output from the visibly synthetic avatars most competitors ship.
- Lip sync accuracy is the quality signal prospects scrutinize hardest and Potion invests specifically there, which is consistent with a founding team out of computer vision.
- Twenty-nine-plus language support lets one seller run localized video into markets they cannot speak, which no human-recorded tool can do at any price.
- Per-workspace rather than per-seat pricing means a five-person team on Professional pays roughly $60 a seat for unlimited generation, undercutting per-seat competitors badly.
- Export-less integrations across 50-plus tools including HubSpot, Outreach, Apollo, and Salesloft keep prospect lists in the system of record instead of round-tripping through CSVs.
- Multiple generation paths, script-to-video, template-based, and screen recording conversion, cover more workflows than a single avatar generator would.
- Customer references include SAP, Gorgias, and Orum, which is meaningful validation for a company with only about $3.5M raised.
Limitations
- The entire value proposition depends on prospects not minding a synthetic likeness, and there is real reputational downside when they do.
- No free plan and only a seven-day trial, which is not enough time to measure whether generated video actually lifts your reply rate.
- Five-minute maximum video length rules out demos, onboarding, training, and any long-form use.
- Starter's three AI video templates per month is restrictive for a team running several campaigns simultaneously.
- SOC 2 compliance and encryption commitments are Enterprise-only, so a mid-market security review pushes you off the self-serve path.
- Small company with roughly $3.5M raised since 2020, so vendor risk is materially higher than with Vidyard or Atlassian-owned Loom.
- Analytics are functional rather than deep; there is no equivalent of Vidyard's account-level intent aggregation or Marketo and Gong integration.
Head-to-head comparisons
3 alternativesPotion vs Sendspark
from $49 per month (Solo, one seat)Sendspark keeps a real recording of you and personalizes around it with a cloned voice and dynamic backgrounds from each prospect's website, at $99 a month for three seats. Potion generates the whole video from a clone of your face, voice, and gestures, at $99 for 750 videos in one workspace. Choose Sendspark if the video should still be genuinely you and the animated GIF thumbnail matters; choose Potion if volume has made real recording impossible or you need 29-plus languages.
Full Potion vs Sendspark comparisonPotion vs Hippo Video
from $0 (Free), then $20 per month billed annually (Pro)Hippo Video also ships AI digital twins and avatar video, bundled with text-to-video, sales pages, and CRM integrations, at $20 to $60 a month. Potion's cloning is more focused and noticeably better on lip sync and gesture, but it does one thing where Hippo Video does five. Take Hippo Video for breadth on a smaller budget; take Potion when the realism of the clone is what determines whether the campaign works.
Full Potion vs Hippo Video comparisonPotion vs Bonjoro
from $0 (Free), then $15 per month billed annually (Starter)Bonjoro is the philosophical opposite: it fires a CRM trigger telling a human to actually record a thirty-second video for one person, unlimited, from $15 a month. Potion exists so nobody records anything. Pick Bonjoro when the fact that a real person took real time is the persuasive mechanism, typically in onboarding and retention. Pick Potion when the volume is high enough that no human could have recorded them anyway.
Full Potion vs Bonjoro comparisonImplementation & onboarding
- Setup time
- A day or two to a first campaign, most of which is waiting. The clone capture session is short but processing takes time, and you should record several test scripts and watch them critically before sending anything to a prospect.
- Learning curve
- Moderate and mostly about writing. The mechanics are simple; the skill is writing scripts that a clone can deliver convincingly, which means shorter sentences, fewer subordinate clauses, and personalization variables placed where a natural speaker would pause. Bad scripts make good cloning look bad.
- Onboarding
- Self-serve on Starter and Professional through a seven-day trial. Enterprise adds strategy consultations, team coaching, and the highest support tier, and is the only route to SOC 2 documentation.
- Migration notes
- Nothing meaningful migrates in, because the asset is your clone rather than a video library. Leaving is cleaner than with a hosting platform for the same reason, though generated video pages already sent to prospects will break on cancellation. If your clone was trained on capture you supplied, confirm deletion terms for that biometric data before signing.
Platform, API & security
- Platforms
- Web applicationBrowser-based recording and generation
- API
- Zapier webhooks on Professional plus roughly 50 native export-less integrations; a general-purpose public API is not published on self-serve tiers.
- Compliance
- SOC 2 on the Enterprise tierTLS 1.2 encryption commitments on Enterprise
- Data residency
- Not published as a self-serve option.
- SSO
- Not published on self-serve tiers; roles and permissions are available on Professional.
- Security notes
- Custom domains and branded pages on Professional. SOC 2 compliance and encryption commitments sit on the quote-only Enterprise tier. Because Potion stores a biometric likeness of a named employee, deletion and consent terms deserve more scrutiny here than with any other tool in this category.
Support & resources
- Channels
- Email support on StarterPriority support on ProfessionalHighest priority support, strategy consultations, and team coaching on Enterprise
- Documentation
- Product documentation covering clone setup, template creation, personalization variables, integrations, and analytics.
- Community
- Active founder-led content presence around AI video in sales; no large standalone user community.
Company
- Founded
- 2020
- Headquarters
- San Francisco, California, United States
- Ownership
- Venture-backed, private
- Founders
- Kanad Bahalkar, Andrew Boos
- Employees
- Small team, not disclosed
- Funding
- Approximately $3.5M in seed funding. Customers cited by the company include SAP, Gorgias, and Orum.
Funding history
| Round | Amount | Year | Notes |
|---|---|---|---|
| Seed | Approximately $3.5M | 2022 | Raised to build generative AI video for personalized sales outreach. |
Timeline
- 2017Kanad Bahalkar works on computer-vision health monitoring at Knit Health, which is acquired by Google; he stays on the Google Health team.
- 2020Potion is founded in San Francisco by Kanad Bahalkar and Andrew Boos to apply generative video to the scaling problem in personalized sales outreach.
- 2022Launches publicly with generative AI mass-creation of personalized videos and raises approximately $3.5M in seed funding.
- 2024Adds gesture and expression cloning alongside face and voice, and expands to 29-plus languages for international prospecting.
- 2025Ships screen recording conversion and template-based generation, and grows the integration surface to roughly 50 sales, marketing, support, and CRM tools.
- 2026Prices Starter at $99 per month for 750 dynamic AI videos and Professional at $299 for unlimited generation with multi-user workspaces, custom domains, and Zapier webhooks.
Integrations
- HubSpot
- Outreach
- Apollo
- Salesloft
- Salesforce
- Gmail and Outlook
- Zapier webhooks on Professional
- Roughly 50 sales, marketing, support, and CRM tools with export-less connections
Frequently asked questions
10 questionsWhat is Potion?
Potion is an AI video generation platform for sales outreach. It builds a digital clone of your face, voice, and gestures from a single capture session, then generates hundreds or thousands of personalized videos in which that clone delivers a different script to each prospect. Videos are delivered on branded pages, tracked for engagement, and connected into HubSpot, Outreach, Apollo, Salesloft, and roughly 50 other tools.
How much does Potion cost?
Starter is $99 a month per workspace and includes 750 dynamic AI videos and three full AI video templates. Professional is $299 a month per workspace with unlimited AI videos plus Zapier webhooks, multi-user workspaces, roles and permissions, custom branding and domains, and priority support. Enterprise is quote-only. Every tier has a seven-day free trial and there is no free plan.
How realistic are Potion's AI videos?
Better than most, because Potion clones gestures and expressions as well as face and voice, and invests specifically in lip sync accuracy. That combination is what separates convincing output from the stiff, frozen-shoulders look most avatar tools produce. It is still generated video, and a prospect paying close attention may notice, particularly on long or complex sentences.
Should I tell prospects the video is AI generated?
That is your call and your legal team's, not Potion's, but it is the most important question in this purchase. The entire mechanism is a video of you saying something you never said. Some teams disclose it and find prospects impressed by the technology; some do not and take the reputational risk. In regulated industries, synthetic representations of a named employee may carry disclosure obligations, so check before you send.
How long can Potion videos be?
Five minutes maximum, on every tier. That is fine for prospecting, follow-ups, and short explanations and rules out demos, onboarding walkthroughs, and training. If you need long-form video you need a different tool alongside Potion, typically Loom for internal and explanatory recording.
What languages does Potion support?
More than 29. This is arguably the single most valuable capability in the product, because it lets one seller prospect into markets they do not personally speak, delivered in their own cloned face and voice. No human-recorded video tool can do this at any price, which makes Potion a genuinely differentiated purchase for companies selling internationally from a single location.
Does Potion have a free plan?
No. There is a seven-day free trial on Starter and Professional and nothing beyond that. Seven days is enough to build a clone, generate test videos, and judge output quality. It is not enough to run a campaign and measure whether generated video actually lifts your reply rate, so plan the trial around quality assessment rather than performance measurement.
Is Potion worth it for a small team?
It depends entirely on volume. At $99 for 750 videos, each personalized video costs about thirteen cents, which is trivially cheap compared with a seller's time. Below roughly 300 sends a month the arithmetic inverts: Bonjoro at $15 or Loom at $18 will let you record real videos for less money and produce a more trustworthy result. Potion earns its price above that line, not below it.
What integrations does Potion have?
Roughly 50 sales, marketing, support, and CRM tools, including HubSpot, Outreach, Apollo, and Salesloft, using what Potion calls export-less integrations so prospect lists stay in the system of record rather than round-tripping through CSVs. Professional adds Zapier webhooks for anything not natively connected. There is no general-purpose public API published on self-serve tiers.
Who owns Potion and how stable is it?
Potion is a private venture-backed company founded in 2020 in San Francisco by Kanad Bahalkar and Andrew Boos, with approximately $3.5M in seed funding. Bahalkar previously worked on computer vision at Knit Health, acquired by Google. Customers cited by the company include SAP, Gorgias, and Orum. With that funding level, vendor risk is materially higher than with Vidyard or Atlassian-owned Loom.
Editorial verdict
Potion is the most technically impressive product in this category and the one that most deserves a deliberate decision rather than a trial signup. The cloning is good, gesture and expression synthesis puts it ahead of the avatar features bolted onto broader platforms, and 29-plus language support is a capability no human-recorded tool can match at any price. Per-workspace pricing at $299 for unlimited generation makes it cheaper per seller than Vidyard for any team of four or more. Buy it if you are sending several hundred videos a month, if you are selling into languages you do not speak, and if you have consciously decided that a synthetic likeness of your team is acceptable to your buyers and your compliance function. Do not buy it below about 300 sends a month, where recording real video in Bonjoro or Loom is cheaper and more credible, and do not buy it if authenticity is your differentiator, because you would be paying $99 a month to undermine it.
Written by the SaaSTracker editorial team. Awards, when shown, are judged against the published criteria in our methodology.