Actively building
Follow the product areas we are shipping across personalized video campaigns, presenter-led generation, outreach, integrations, and team workflows.
The create-video flow now starts with one Source Video step instead of a scattered method and tutorial sequence. Users choose Record in BHuman or Upload a video, then stay in the same workspace for preview, transcription, variables, and continue.
Recording now starts with a practical choice: speak naturally, or paste a script and read it while recording. The teleprompter supports editable notes, manual scroll controls, speed, and text size while the recorder still handles camera, mic, preview, and save.
Uploading a source video now uses the actual selected file, shows upload and transcription progress, and gives retryable errors when something fails. Users no longer land on a video setup screen that looks ready before the file is really processed.
The AI Studio homepage now separates the reusable video templates users create from the campaigns they launch. It also shows the right New template and New campaign actions plus video usage against the customer's plan limit.
New campaign now takes users to the saved templates list and highlights the source video they should pick first. New template remains the separate path for recording, uploading, importing from Speakeasy, or browsing community templates.
Teams with many saved videos can now search the template area by name before choosing what to launch. Search filters the full template list before pagination, so older but reusable videos are easier to find.
Generated-video pages now keep checking for completed rows every few seconds and merge live updates when they arrive. Users see ready and failed counts, last checked time, and a manual Refresh option instead of guessing whether a campaign finished.
CSV and spreadsheet uploads for campaign recipients now keep upload status and generated-video rows in sync more reliably. The campaign setup is clearer while the list is processing, especially when the file has many recipients.
Generating a sample now creates a visible pending row immediately, then updates progress through live events and polling fallback. Finished samples appear as preview tiles and open in an in-app player, with an external link only as a fallback.
Transcript tools were tightened so users can mark the words that should change per recipient, like first name, company, title, location, or offer. The toolbar now makes variable naming clearer and avoids the extra picker step that slowed people down.
Saved templates now restore variable segments, transcript words, source-video preview data, and recent samples when reopened. Users can edit or generate another sample from an old template without re-transcribing or re-marking personalization words.
Sample and generated-video failures now use customer-facing retry messages instead of exposing provider names, request IDs, or transport details. The goal is simple: tell the user what happened and whether to retry, reconnect, or refresh.
Settings now has a Connect inbox flow for Gmail and other SMTP/IMAP providers. Connected sender accounts can be used by generated-video email sending, and sender credentials are stored securely rather than exposed back in the UI.
The settings modal now keeps account, workspace, connected inboxes, API keys, Zapier, product settings, and billing in clearer places. Product billing lives inside the product tabs, and integrations show real states instead of fake or hard-coded setup instructions.
LeadR now shows connected LinkedIn accounts, account health, sender slot usage, and clearer Connect, Add slots, or Retry states. Users can tell whether they are ready to launch outreach or need to fix an account first.
A new LeadR user now sees one obvious first move: connect a LinkedIn account. The empty campaigns panel stays hidden until there is an account or campaign to manage, so the first-run screen no longer looks like two different setup states at once.
LeadR moved into the BHuman workspace with account connection, campaign surfaces, entitlement checks, and navigation integrated into the app. Users can manage LinkedIn outreach from the same workspace as their video tools.
AI Studio can now load finished Speakeasy projects and import a selected render as a campaign video. The handoff keeps useful script and transcript context so users do not have to rebuild the source video from scratch.
Speakeasy auth now routes through BHuman, reducing the separate-account friction when a BHuman user opens Speakeasy or imports a finished render. This is the groundwork for cleaner cross-product workflows.
The updates flow now starts from real product material: visual uploads, large image uploads that resize automatically, and walkthrough recordings. Users can create product update videos from the assets they already have instead of starting from a blank presenter-only flow.
Speakeasy added structured b-roll planning, visual references before timeline generation, preview failure reasons, and real b-roll rendering. Generated videos can now include supporting product visuals and motion graphics instead of only presenter footage.
Speakeasy added controls for deleting saved presenters, choosing presenter assets more reliably, and authoring guided templates with structured motion-graphics options. This gives teams more control over the people, visuals, and reusable formats behind generated videos.
The public site now has real BHuman example videos grouped by sales, lifecycle, and support use cases. Each example has a watch page and links into related workflows so buyers can see what a real campaign could look like.
The website now has practical workflow pages, industry pages, comparison pages, real examples, product screenshots, and proof assets for buyers evaluating BHuman. The goal is to show concrete use cases, not just say personalized video works.
We are tightening LeadR launch around the choices users actually need before sending: healthy sender account, audience, selected video, and final launch settings. This work also covers reconnecting accounts, refreshing account state after connection, and preventing stale account health from blocking launch.
We are adding campaign batch tooling so larger AI video jobs are easier to queue, monitor, and recover when generation takes longer than expected. The user goal is clear batch visibility instead of wondering which videos are still waiting, processing, or ready.
We are connecting the Demos creation path to Speakeasy's generation system so a rough screen recording can become a cleaner presenter-led demo. The intended flow is record or upload a rough walkthrough, choose presenter and voice, review the script direction, then generate the polished demo.
We are making the LinkedIn invite note setting explicit, including the case where a user intentionally wants no note at all. LeadR should respect that choice instead of treating a blank note like an accidental missing field.
Recover videos are being shaped into reusable templates that can personalize at least the recipient's first name. That should make win-back, follow-up, and recovery campaigns faster to create without rebuilding the video structure each time.
Upcoming generation reliability work is focused on keeping mouth movement aligned with the generated voice when render queues are slow or delayed. Users should see fewer failed videos and fewer results where the voice and lips drift apart.
Upcoming UI cleanup will make generated-video processing cards stay readable when campaigns have long names, long recipient fields, or detailed status messages. The page should keep its layout while users monitor bigger batches.
We plan to unblock Speakeasy email OTP signup so new users can create an account, verify email, and start generating presenter-led videos without a manual handoff.
Customer priorities shape this list
Tell us what you are trying to automate, which tools are involved, and where the current process breaks. The roadmap is strongest when it is grounded in real campaigns.