What's new at ioZen

Four additions that work together: a way to total answers into a score, and a smarter Record Lookup field to feed it.

Calculated fields. A hidden number field can now total a group of other fields tagged with a shared score group: sum, average, min, max, or count. Unanswered items in the group are excluded from the total and flagged, so a routing rule can require a complete score before acting on it. Optional bands attach a label like “low” or “high” to ranges of the total.

Live score display. A content block can interpolate that running total, and its band, mid-conversation with {{field_name}} and {{field_name.band}}. The display isn’t saved; the server recomputes the canonical value at submit, on web and WhatsApp alike.

Lookup carry values. A Record Lookup field can now copy other columns from the matched row onto the submission, for example a numeric score sitting behind a Likert-scale label. The stored answer stays the label the respondent picked; the carried value resolves on the server and writes nothing if the label matches more than one row.

Lookup row filters. A Record Lookup field can be scoped to the rows matching a column value, so one shared table can back many different questions instead of needing a separate table per question. A label that repeats across the table stays unambiguous because it resolves within its own scope.

Available on all plans.

A cluster of fixes across three areas: Agent Blocks backed by decision tables, spreadsheet imports into Records, and WhatsApp billing clarity.

Agent Block, decision-table accuracy. A column set to fill silently from the match now writes correctly even when the table pins it to a single value before the respondent has answered anything; previously it was skipped unless an already-asked column held a value. A silently-filled column is now only written when the questions already answered determine a single row, so every respondent with the same answers gets the same value instead of an arbitrary first match. Table-backed questions now validate the respondent’s answer against the actual valid choices, on web and WhatsApp alike, instead of silently accepting anything or skipping ahead. When an agent block captures several fields in one reply, all of them now land in a single submission instead of splitting across several.

Agent Block, exports and the submissions table. Agent block answers used to sit at the very bottom of the export column picker, under the block’s full instruction text, easy to miss in a long form. Each answer now appears where its block sits in the flow, with its own icon and a short spreadsheet header. Large queued exports without a manual column selection now include agent answers too, and the submissions table preview shows what the respondent actually answered, agent-collected turns included.

Cleaner Records imports. A spreadsheet with two columns sharing the same name used to crash the import preview; repeated headers now get renamed automatically (V2, V2 (2)) and import as separate fields. A column containing only 0 and 1 used to import as a checkbox with no way to fix the guess; those columns now import as numbers unless the values are spelled out (true, false, sí, no), and every mapped field now has a type dropdown so you can correct it either way.

Clearer WhatsApp billing. A workspace with plan credits left but no purchased credits is now told, plainly, why WhatsApp won’t send: in the inbox composer, the usage page, and a low-balance email, in English and Spanish. Plan credits cover AI and integration tools; WhatsApp needs purchased credits specifically.

Agent Block fixes on Pro, Business, and Enterprise. Import fixes on all plans. WhatsApp billing clarity on Business and Enterprise.

Credit packs now price at a flat $0.01 per credit across the whole ladder. Growth is $50 for 5,000 credits, Scale is $100 for 10,000, and a new Peak pack adds 50,000 credits for $500 for workspaces running heavier volume.

Boost stays at $10 for 1,000 credits. Volume discounts on the larger packs are gone, so a credit spent on WhatsApp or any other vendor pass-through costs exactly what the invoice says, with nothing to reconcile.

Available on Pro, Business, and Enterprise plans.

Billing settings now shows a breakdown of the month’s credit usage by category: AI, integration tool calls, and WhatsApp messages. It’s still one balance and one top-up; nothing changes about how you buy or spend credits. The rows always add up to the total shown above them.

This also completes a naming change from “AI credits” to “credits,” since one wallet has funded more than AI for a while now, and the old name undersold what it actually covers.

Available on all plans.

Connecting a WhatsApp number now asks upfront whether you’re bringing a number you already have in Meta, or want a new US number provisioned for you. Bringing your own number now works in any country, and no longer fails at Meta’s authorization step the way it sometimes did before.

Each workspace is also isolated to its own provider account, so one workspace’s WhatsApp Business account can never block or interfere with another workspace’s numbers. Linking a number by ID is now verified against the provider first, so a mistyped ID reports an error immediately instead of leaving a connection that silently receives nothing.

WhatsApp connections can now be removed from the Integrations tab. A number with conversation history is disconnected, not deleted, so past threads stay in Team Inbox.

Available on Business and Enterprise plans.

Since Intake Bots started running on WhatsApp, the wizard had been shortening AI-generated select and boolean option labels to fit WhatsApp’s 24-character row title limit, on every bot, whether it would ever run on WhatsApp or not.

Options are now written for the web first, at full length. WhatsApp still renders them safely: the row title truncates while the full text carries in the list description, and very long option sets fall back to a numbered list. A long option label can no longer cause the questions behind it to get dropped during generation.

Available on all plans.

Two fixes to attribution accuracy.

ioZen now captures the utm_id campaign ID from landing URLs. Ad platforms substitute a numeric ID there (TikTok’s campaign ID, Meta’s campaign ID), and it’s the stable key to track by, since utm_campaign carries a name that changes whenever someone renames the campaign. It shows up on the submission attribution panel and as a CSV export column.

Browser event IDs and the pixel cookies are now stored on the submission when the form completes, so the server-side conversion call shares an event ID with the browser pixel event. Ad platforms use that shared ID to deduplicate the pair instead of counting one conversion twice. Pixel events also now fire only after the submission is durably saved, so a failed save can no longer report a conversion with no matching record behind it.

Available on Pro, Business, and Enterprise plans.

Every Intake Bot has been a web-only conversation until now. Business and Enterprise workspaces can connect a WhatsApp number and run the exact same bot there.

Connect a number you already manage in Meta, in any country, or have ioZen provision a new US number for you. Business includes up to 3 numbers; Enterprise is unlimited, and each workspace runs on its own provider account.

Nothing about the bot changes to support this. Conditional logic, agent blocks, calculated fields, and the full submission pipeline (contact, record, Process Board card) all run on WhatsApp exactly as they do on the web. Select fields render as buttons or lists, and respondents can send images, PDFs, audio, and video straight from the chat.

Messages draw on purchased credits, not your monthly plan allotment.

Available on Business and Enterprise plans.

TikTok was quietly dropping our conversion events. The request body shape we were sending had been deprecated, and Events API 2.0 expects a different format. Events looked fine on our side and never showed up in TikTok.

We now send conversions with the current Events API 2.0 request format. ioZen also injects the TikTok Pixel on published Intake Bot pages and fires a browser-side SubmitForm event alongside the server-side Events API call. Both share the same event_id, so TikTok deduplicates them instead of counting twice or missing the browser signal.

Match quality improves with ttclid, the _ttp cookie, and hashed external IDs. A Test Event Code field lets you verify against TikTok’s sandbox in the Test Events tab without touching live data.

Available on all plans.

A Process Board gets noisy fast when everyone owns every card. You could drag cases between stages, but ownership lived in someone’s head or a side channel.

Cards can now be assigned to one or more team members. Pick assignees from the case view, see their avatars on every card, and filter the board by assignee from the toolbar. Newly assigned members get an email with a direct link to the case. Every assignment change lands in the case activity timeline, so you can see who owned it and when it moved.

Available on all plans.

You could already collect cases and score them. What you couldn’t do was look across a few hundred submissions and see where people finish, stall, or drop off, or change the bot’s voice after generation without rebuilding the flow. This release fixes both, and a couple of builder gaps that kept getting in the way.

Analytics on every Intake Bot

Each Intake Bot now has an Analytics tab. You get a completion view (finished, in progress, abandoned), answer distributions for select, number, date, boolean, and agent-collected fields, and a breakdown of where input came from: typed answers, voice, advisor questions, and agent turns.

Pick a schema version and a date range so totals stay explainable when the bot has changed. Turn on AI insights when you want recurring topics, top questions, and sentiment pulled from open-text answers and advisor Q&A. Insights run on demand and use credits. Analytics and insights are on Business and Enterprise.

Voice and Tone after generation

The Voice Pack that shaped an AI-generated Intake Bot used to freeze at create time. There’s now a Voice and Tone settings section where you can change tone, emoji use, voice examples, negotiation style, and celebration behavior after the fact. A scoped re-polish rewrites just the welcome and farewell to match, and leaves the rest of the flow alone.

Agent blocks inside branches

Agent blocks no longer have to sit at the top of a flow. Drop them inside branches and repeating groups. Their collected outputs flow through records, mappings, exports, PDFs, and validation the same way top-level fields do, so a goal-based conversation can live deep inside a conditional path.

Bulk editing in the builder

Select multiple questions and act on them from one toolbar: duplicate, move, apply a shared condition, or delete. The move menu shows a recursive tree, so a selection can land in a nested branch or repeating group in one click instead of climbing a level at a time.

Analytics and AI insights on Business and Enterprise. Voice and Tone and bulk editing on all plans. Nested agent blocks on Pro and up.

Content blocks could already carry rich text, embeds, and CTAs. They couldn’t carry a product photo, a badge, or a simple diagram without sending people somewhere else.

Type /image inside a content block or CTA modal and drop a file in. Uploads compress to responsive WebP in the browser and serve from a CDN, so they stay sharp on Retina screens without eating bandwidth. Pick small, medium, large, or full width. Image storage counts toward your workspace quota.

Available on Pro and Business plans.

Some paths through a bot shouldn’t reach the last question. A respondent might be disqualified halfway through, or they’ve confirmed everything you need by step three. Forcing them to a fixed ending creates friction and, in disqualification cases, a worse experience than just stopping cleanly where it makes sense.

Any content block can now be toggled as the intake’s endpoint. When a respondent reaches it, that block delivers the closing message and the submission is finalized right there. No redirect, no “please wait” screen.

The pipeline still runs

Ending early doesn’t skip anything on the back end. Contact creation, record mapping, the process board card, and any routing rules all fire exactly as they would for a standard submission. The difference is where the conversation stops for the respondent, not what happens with the data.

Each early-ended submission also carries an outcome label you define. It shows up on the submission detail view and is available as an export column, so you can filter and report on which path each case took.

Works well with conditional logic

Pair an early-ending block with a condition and you get branch-specific closings. One branch says “You’re all set, we’ll be in touch.” Another says “Based on your answers, this isn’t the right fit right now.” Each ends the intake at exactly the right point, with the right message.

Available on Free, Pro, and Business plans.

When an Instruct Mode advisor block is running, the bot answers questions. That’s the point. But respondents don’t always want to keep asking — sometimes they just need one answer and want to move on.

Previously, that required waiting for the bot to decide the conversation had reached a natural end. For some respondents that wait felt like being stuck.

What changed

Advisor blocks now show a Continue or Finish button on every single turn, regardless of where the conversation is in the flow. The label adapts to context: “Continue” if there are more steps ahead, “Finish” if the advisor is the last block. Builders can customize the label if the default doesn’t fit their flow.

No change to how the advisor itself behaves. It still answers from knowledge, still grounds every response in what you attached. The button just ensures respondents are never trapped waiting for the bot to wrap things up on its own.

Some intake flows have a natural point where the respondent needs to ask a question before they can answer one. What does this coverage actually include? What happens after I submit? Do I qualify if I work part-time?

An FAQ block can cover common questions, but it breaks the conversation. Instruct Mode keeps the flow intact.

What Instruct Mode does

Drop an Instruct Mode agent block onto an Intake Bot and attach markdown knowledge sources. When a respondent asks a question, the agent answers it using only what you attached. Nothing from the open web. Nothing it made up. If the answer isn’t in the attached knowledge, it says so.

Two configurations:

Standalone advisor. The block shows up as a dedicated Q&A step. Respondents ask questions and get answers. A persistent Continue or Finish button is always visible so they can move on whenever they’re ready.

Mid-collect Q&A. Inside a Collect Mode flow, the agent can answer a respondent’s question and then redirect back to gathering the fields it still needs. The conversation stays on track.

Why this matters

Previously there was no good answer for a respondent who had a question mid-flow. They could abandon, look things up elsewhere, or submit guesses. Instruct Mode gives them answers without breaking the intake.

Available on Pro plans and above.

If you use Records-backed decision tables in an Agent Block, you may have a column that should be filled automatically once a row is matched rather than asked as a question. A product code, a tier identifier, an internal category. This release adds that.

What derived columns do

Mark any mapped column in a decision table as “fill silently from match.” Once the respondent’s answers identify a row, that column’s value is pulled from the matched row and saved to the submission. It never shows up in the conversation.

The derived value behaves like any other field from that point: it appears in record mapping, process board cards, exports, and action block field mappings. Nothing downstream needs to know it came from a table lookup rather than a respondent answer.

Why this matters

Before this, you could only output columns the agent explicitly asked about. If you needed a derived value (like a plan code that depends on the combination of size and coverage type), you either asked the respondent a question they couldn’t meaningfully answer, or left the derivation to post-processing elsewhere.

Now you define it once in the table. The agent handles the logic.

Building the same FlowApp twice for different clients, offices, or environments has always been manual work. Copy each IntakeBot field by field, rebuild the Process Board stages, map the routing rules again. This release eliminates that.

How the copy works

A three-step wizard handles the whole thing. Pick the destination workspace, review a preflight checklist that shows every Intake Bot, Process Board, phase, and Record Type about to be installed, then run the copy. The operation is atomic — either everything lands cleanly or nothing changes.

Copied Intake Bots arrive as drafts so you can review them and reconnect any dependencies before publishing. The source workspace is untouched throughout.

Handling missing integrations and contact fields

If the destination workspace is missing integrations or custom contact fields that exist in the source, the result step surfaces them with deep links. You can jump straight to reconnect an integration or create the missing field without leaving the flow.

Structure and configuration only — no submissions or cases are copied.

Available on all plans.

When a respondent needs to pick from an existing list, typing a free-text answer creates problems downstream. You get inconsistent spellings, values that don’t match your records, and manual cleanup before anything can route correctly.

Record lookup fields fix this at the source.

How it works

Add a record lookup field to any Intake Bot and point it at a Record Type. Respondents see a search box (for large datasets) or a list of selectable options (for smaller ones), sorted A to Z. Whatever they pick is stored as a reference to the actual record row, not a copied string.

That reference travels through the rest of the submission: record mapping can resolve the linked row automatically, process boards associate the card with the right entity, and exports include the resolved values rather than a raw text answer.

No extra setup for large catalogs

For Record Types with many rows, the lookup field loads matches on demand as the respondent types. No pagination, no loading the full list upfront. Builders define the field once and it scales with the record set.

Available on all plans.

Fixed question sequences work fine for simple bots. They start to feel awkward when the path depends on a combination of answers, or when you need to collect something open-ended that takes a few rounds to pin down. The new Collect Mode agent block handles that kind of intake.

What it does

Instead of asking questions one by one, a Collect Mode block runs a goal-driven conversation. You define the fields you want, attach knowledge sources (markdown context or Records-backed decision tables), and the agent figures out how to gather them. The conversation adapts to what the respondent has already said rather than marching through a fixed list.

The outputs are ordinary submission fields. Process boards, records, and action blocks pick them up the same way they would any other field.

Decision tables keep things grounded

For cases where you need the respondent to pick from a valid set of combinations, you can back the agent with a Records decision table. The agent enforces valid combinations server-side once the respondent’s answers pin a row. Large option sets use search-as-you-type so respondents never type something invalid.

Fallback and guardrails

If AI is unavailable or credits run out, the block falls back to static questions. A credit budget and turn cap prevent the conversation from running away. The authoring UI shows each slot’s resolved behavior — free text or constrained by a table — as you build, so there are no surprises at runtime.

Available on Pro plans and above.

For a while now an Intake Bot would collect the case and hand it off, and the rest was on you. Push it into your CRM by hand, or wire up a webhook and hope it held. This release closes that gap and a few others.

Create CRM records straight from the conversation

Drop an action block onto an Intake Bot and it creates the contact, deal, or lead in HubSpot or Zoho the moment a case reaches a point in the flow or finishes. You map each field once with a tool picker. Every run is logged on the submission with its timing, credit cost, and any errors, so you can see exactly what fired and when.

Assisted and agentic execution

Some handoffs are simple: one record, fixed fields. Others need judgment. Action blocks now run in three modes. Deterministic fires a single tool you preconfigure. Assisted lets the AI fill in any parameters you didn’t map, following a short instruction you write. Agentic turns the block loose as a multi-step agent across any toolkit you allow, with hard caps on steps and credit budget so it never runs away. The submission’s Actions tab shows every step it took.

Integrations now scale with your plan

We sorted every integration into Standard and Premium. Standard connections work on any plan. Premium ones, like the deeper CRM and productivity tools, open up on Business and Enterprise. Free workspaces can connect up to two integrations; every paid plan is unlimited. Downgrade later and your existing connections and data stay put. The only thing that pauses is new runs on locked tools.

Build a question set from a paragraph

In the Intake Bot editor you can paste a block of text or a list of questions and let AI turn it into real fields, with types, required flags, and select options already filled in. Review each draft, edit anything that’s off, then drop them into the bot. Want to skip the AI? One-per-line mode adds them straight across and spends no credits.

A quieter pass on security, privacy, and reliability

Plenty changed under the hood. Plan and feature changes now apply the instant they happen, so an upgrade unlocks features without a page refresh, and live updates run over authenticated private channels. Each workspace’s data retention window is now tracked, and expired submissions, along with their files and stored secrets, are being cleared automatically as that rolls out over the coming days. We also tightened database access, outbound request handling, content security policy, and input validation, and added redaction so personal data never reaches the logs.

Available across plans, with assisted and agentic action blocks on Pro and up.

When you’re setting up a Meta integration, you need a way to confirm the conversion triggers are firing correctly without polluting your live dataset. Previously that required running a live test submission and then cleaning up the data.

Now, any test submission run inside ioZen (bot preview or single-field test view) automatically routes to Meta’s Test Events tab when a Test Event Code is set. You can verify your On Submission and Field-Level conversion triggers end to end, with real Pixel + CAPI pairs showing up in the sandbox exactly as they would in production.

Real submissions are isolated. They always go to your live dataset and never carry the test code, so it’s safe to leave the Test Event Code set permanently after initial setup.

Available on all plans.

Meta’s Conversions API has been part of ioZen for a while. The problem with server-side-only attribution is that Meta needs both a browser signal and a server signal to deduplicate correctly. Without the Pixel, CAPI events can end up competing with browser events fired by other tools on the same page, or just missing the browser signal entirely.

Starting today, ioZen injects the Meta Pixel automatically on every published Intake Bot page. When a submission comes in, a browser-side Lead event fires alongside the CAPI event. Both carry the same event_id, so Meta knows they’re the same conversion and counts it once.

Match quality also improves. The _fbp and _fbc cookies are now passed through alongside the raw IP via CAPI, which means Meta has more signals to match the conversion to an ad click. In practice this shows up as better attribution in Events Manager, not higher conversion counts.

The Test Connection button opens the bot in a new tab to fire a real Pixel event, then links you to the Events Manager Test Events tab so you can confirm the deduplicated pair arrived. A Test Event Code field lets you use Meta’s official sandbox without touching live data.

Available on all plans.

Two improvements shipping together today.

Generation recovery. When an Intake Bot generation job times out, the spinner card used to stay there permanently. The only fix was to know that deleting the bot would clear it, which most people didn’t know. Now stale jobs are treated as recoverable. Delete actions mark a timed-out job as failed before cleanup, so the card resolves cleanly. Runtime AI-generated content blocks also have a fallback path: if the AI call fails at runtime, the block falls back to the authored copy, or to a localized message if no copy is present. No more blank blocks.

QR code downloads. The Intake Bot Sharing Hub now includes a QR Code option. Preview the code, copy the linked public URL, or download a PNG for use in printed materials, event signage, or offline distribution. The QR code updates automatically when you change the bot’s public slug.

Both updates are available on all plans.

The new Ludicrous tone is for intake bots that are supposed to be fun. Demo bots. Showcase bots. Bots where the point is to show off what ioZen can do.

Ludicrous unlocks all 9 Magic Moment kinds with no budget cap. Other tones limit how many effects the AI can place and which kinds are available. Ludicrous removes both restrictions. The AI also picks intensity (subtle, standard, or epic) per content block based on context, rather than locking everything to a single level.

The plan AI now selects Ludicrous automatically when the prompt asks for a showcase, demo, or “all the magic moments” experience. If you’re building something for your own portfolio or trying to show a client what ioZen is capable of, describing that in the prompt is enough to activate it.

Non-Ludicrous bots are not affected. Their intensity policy continues to follow the tone’s configured level.

Available on all plans.

Content blocks can now do something besides show text. Add a Magic Moment and the block triggers an ambient effect when the respondent reaches it.

Nine effects are available: confetti, fireworks, thunderstorm, deep-breath, gong, dramatic-explosion, drama-curtain, rainbow-portal, and flying-unicorn. Each runs in three intensities: subtle, standard, and epic. A kind picker in the block editor replaces the old toggle so you can see and choose exactly what you’re adding.

Performance. Each effect is lazy-loaded independently. If a bot has no Magic Moments, zero bytes are shipped for them. A bot using only confetti loads only the confetti code. Effects respect prefers-reduced-motion and Save-Data. There’s also a bot-level kill switch in settings to disable all effects at once.

AI selection. The AI skeleton generator now picks Magic Moment kinds based on the bot’s tone. Playful bots can receive up to two effects from the full library. Warm bots get one effect from a curated set (deep-breath, gong, confetti, drama-curtain, fireworks). Professional and formal bots receive none. Celebration effects you configure manually are preserved when you regenerate.

Available on all plans.

When AI Document Extraction launched, it only worked on top-level file questions. Any file field inside a conditional branch was excluded.

That limit is gone. Extraction can now be enabled on file questions at any nesting depth. The runtime extractor finds the uploaded document regardless of where in the conversation tree it lives, and fills the child answers correctly.

Two other improvements shipped at the same time. Generated child questions no longer include a redundant photo upload or location field below the file you just uploaded. Those were noise in the conversation and they’re removed. And the AI extraction panel in the field editor now shows which child questions are tied to the extraction, so builders can see exactly what will be pre-filled at runtime.

Available on Pro, Business, and Enterprise.

File upload questions now do more than collect a document. Point ioZen at any file field, enable AI extraction, and the bot reads the uploaded file and fills in the answers for the respondent.

Upload a driver license and the name, address, and date of birth fields complete on their own. Upload an invoice and the account number, total, and date pull out cleanly. Upload a utility bill and the service address is already there when the respondent gets to that question.

How it works. Builders configure the feature in the field panel: toggle AI extraction on, add a document hint describing what kind of document to expect (for example, “US driver license” or “electricity bill”), and optionally generate child questions automatically from a sample document. The generated child questions use the same conditional logic editor as any other field and support nested branching.

At runtime, respondents see a “Reading your document…” indicator while ioZen processes the upload with vision AI. Fields that meet the 60% confidence threshold fill automatically. Fields that don’t stay blank for the respondent to fill in manually. The respondent then reviews every pre-filled answer before the conversation continues.

Where it lives. AI Document Extraction is available on file upload questions at any position in the conversation. Builders on Pro and above can enable it. The child questions it generates behave like any other field.

Available on Pro, Business, and Enterprise.

The biggest release since Submission Intelligence, all shipping at once: voice input across every field type, a rebuilt mobile chat, an AI that pauses to explain its plan before it builds anything, an Integrations directory, in-product feedback bots, and a four-tier pricing structure with a real Enterprise option.

Voice on every field. Respondents can dictate answers anywhere now. Text and textarea fields stream live transcription with a light AI polish for capitalization and punctuation. Email, phone, URL, number, date, boolean, and select fields go through a typed extraction pipeline that turns speech into structured values. “Next Friday at 3pm” becomes an ISO date. “Five five five one two one two” becomes a phone number. “All except YouTube” selects every option but YouTube. Voice ships on by default for new bots, in EN and ES, and audio never leaves the device.

Mobile-first intake chat. The public chat was rebuilt around a single persistent composer that stays mounted across every field transition. The iOS virtual keyboard no longer dismisses between questions. Advancing through text, email, date, select, geolocation, and file fields feels continuous. Zero keyboard pops, zero viewport jumps, zero flicker. The advance button sits in the same pixel position on every field, whether it shows Continue, Skip, or Send.

An AI that pauses to explain itself. The Smart FlowApp Wizard and onboarding conversation now draft a short conversation plan (goal, audience, beats, branches to honor, priority signal) and pause for your approval before generating the schema. Approve to build, or describe what you want changed and the AI refines the plan. You stay in the loop on branching logic before any schema is written, which catches obvious problems early. Like asking someone to upload a file before checking whether they have one.

Generation that’s faster and calmer. Bot generation now runs as a multi-pass pipeline: a creative model writes the conversation, then a separate reasoning model enriches each field with intelligent extraction, follow-up behaviors, and validation. Bots arrive with content blocks personalized via template variables and a single Voice Pack that keeps tone consistent from welcome to farewell. We upgraded every AI surface to the latest model generations with automatic fallback, fixed a truncation bug that was silently cutting off long generations, and tuned the failure UI so transient provider hiccups no longer flash a red “Generation failed” message on bots that ultimately succeed.

A new home for everything else. The sidebar got a refresh with a workspace switcher and a FlowApp switcher you can use from anywhere. There’s a new Integrations directory listing 25+ destinations across marketing, automation, CMS, CRM, email, and productivity. Each card deep-links to the right setup path. Two in-product bots collect feedback and integration requests with full context attached as hidden fields. Wizards persist their state to localStorage so a refresh or browser restart no longer loses your place. Every published IntakeBot gets a dynamic Open Graph image rendered from its design tokens, so links shared on WhatsApp, iMessage, Slack, and the rest render a branded preview instead of a favicon. And bot share URLs are now customizable: pick a memorable slug from Settings and the public link, embed code, and QR code all update together.

Four tiers, with a real Enterprise option. Pricing has been restructured. Unlimited submissions are now included on every paid plan, forever. Pro is 3 FlowApps, 5 IntakeBots, and 5,000 contacts. Business is 10 FlowApps, 20 IntakeBots, 25,000 contacts, and 50 GB of storage. Enterprise starts at $500/mo and includes unlimited everything, SSO/SAML, a custom domain, white-label, 50,000 AI credits per month, and a dedicated CSM. Enterprise is custom-billed. Inquiries route through an in-product intake bot, not Stripe. Drafts and archived bots no longer count toward your published-bot limit either, so you can iterate or archive freely.

When an IntakeBot field is optional, the Send button used to sit there disabled until the visitor typed something. Nothing told them they could move on. Some people probably sat there wondering if the bot was broken.

Now optional fields show a Skip button. Tap it to move to the next question without typing anything. Start typing and it switches to Send with a smooth animation so the change feels intentional, not glitchy.

Works in English and Spanish. Available on all plans.

Two quality-of-life improvements shipping together.

Draft session resume. When a visitor starts an IntakeBot conversation and leaves before finishing, they can now come back and pick up where they left off. The next time they open the same bot, a prompt gives them the choice to continue their previous session or start fresh. Sessions expire after 24 hours, and the system detects schema version changes so a stale draft doesn’t confuse anyone. Works across all embed modes: inline, popup, and floating widget.

This should help with completion rates on longer intake flows where people get interrupted mid-conversation.

AI Value Extraction. This is the third AI mode in AI Field Intelligence, alongside Socratic Negotiation and Strict Validation. Value Extraction runs silently when a respondent’s answer contains conversational noise. If someone types “Hi, my name is Pedro Navajas” into a Name field, the AI extracts “Pedro Navajas” and saves that without showing the full raw response back to the user. When no value can be found, the bot generates a natural re-prompt instead of repeating the original question verbatim.

The number of re-prompt retries is configurable per field, from 0 to 3, with a default of 1. The AI wizard automatically applies extraction on name and identity fields when generating new bots.

Both features are available on all plans.

Fixed a bug where filtering the submissions list by status (In Progress, Completed, or Abandoned) returned no results. The filter option values were lowercase strings but the stored data uses uppercase enum values, so the comparison always failed. Values are now aligned and the filter works correctly.

Starting today, every ioZen plan includes the full feature set. Submission Intelligence, hidden fields, rich content blocks, document extraction, ad platform integrations, outbound webhooks, and intelligence-aware routing are all available on Free, Pro, and Business.

What changes between plans is capacity. How many FlowApps you can run. How many AI credits you get each month. How many webhook endpoints and API keys you can manage. Your team size. Your storage. The features are the same; the headroom is not.

This matters in practice. A freelancer on Free can now build an IntakeBot with Submission Intelligence, see AI-scored tiers on every card, and route leads based on those scores. A growing team on Pro gets the same tools plus more FlowApps, more contacts, renewable credits, and API access. Business adds the highest limits and removes ioZen branding.

We also added the AI credit balance to the sidebar so you always know where you stand without going into settings.

The pricing page reflects all of this now.

A reliability and quality pass focused on billing accuracy and AI output.

Credit grant fix. A bug was causing admin-granted bonus credits to carry over permanently across billing cycles instead of resetting monthly. This has been corrected and affected accounts have been updated. Bonus credits now consume after plan-included credits and before purchased credits, and reset cleanly on each billing cycle.

AI generation improvements. IntakeBot generation prompts have been refined to produce better field selection and higher-quality schemas. Submission Intelligence scoring prompts now assign tiers more accurately. Template variable interpolation is more reliable across edge cases.

Billing architecture. All plan limit enforcement now runs through a single resolution path with consistent defaults. Admin-enabled feature flags now reflect correctly without requiring a page refresh. Credit purchase webhook writes are atomic, preventing partial failures from leaving accounts in an inconsistent state.

The biggest release since public beta. Three things that matter:

Submission Intelligence scores every submission the moment it arrives. AI analyzes answers against configurable dimensions and assigns a tier, so your team sees what’s worth acting on before opening a single record. Sales teams see Hot / Warm / Cold. HR teams see Strong / Moderate / Review. Support teams see Urgent / Normal / Low. You configure the dimensions and tiers to match your process, or let AI generate them during bot creation. Tier badges show directly on Process Board cards. Business plans can route submissions automatically based on scores. Costs 2 AI credits per submission.

Geolocation capture is a new field type for IntakeBots. Respondents tap “Share Location,” grant browser permission, and their GPS coordinates are recorded with a static map preview (OpenStreetMap tiles, no API keys, no dependencies). Every display surface shows a “Open in Google Maps” link. Built for inspection reports, field service, incident tracking, delivery confirmation, and property management. Available on all plans.

Deploy-first onboarding replaces the old “confetti on creation” flow. After AI generates your first FlowApp, you now pick how to deploy it: direct link with QR code, embed code for your website, email instructions to your webmaster, or a step-by-step guide for your CMS platform (14 platforms supported, including WordPress, Squarespace, Wix, Webflow, and Shopify). Confetti now fires on your first real submission, not on setup completion. Session state is preserved, so navigating away no longer resets your progress.

We also tightened security across the credit system and submission API: atomic billing-period resets, rate limiting on public credit checks, Zod validation on all IDs, cross-bot update guards, and a fix for purchased credit packs being ignored in AI availability checks.

Three new AI-powered creation tools ship today. All three work the same way: you talk to ioZen in plain language, and it builds what you need.

Conversational Onboarding is the first thing new users see. Paste your website URL and AI figures out what your organization does, then suggests use cases matched to your role. Pick one (or describe your own) and ioZen generates your first FlowApp automatically. No clicking through setup wizards or reading docs first.

Smart FlowApp Wizard brings that same conversational approach to creating new FlowApps after onboarding. AI suggests use cases based on your organization’s profile, or you can describe what you need from scratch. You get a full proposal with questions, boards, phases, and routing rules. Review it, request up to two rounds of adjustments, and hit build. The whole thing takes about a minute.

Smart IntakeBot Create works inside any existing FlowApp. Open the creation panel and you’ll see AI-generated bot suggestions tailored to that FlowApp’s context. You can also describe what you need in plain language or upload a document and let AI extract the fields for you (Pro and Business plans). The result is a ready-to-use IntakeBot with all fields configured.

Under the hood, we also improved AI gateway reliability. Operations now have configurable timeouts (60 seconds by default, 300 seconds hard limit) and usage metrics are tracked independently from records. Deleting a record no longer wipes out the associated AI usage data.

Three changes today, all focused on tightening up how hidden fields and content blocks work in practice.

Hidden fields are now a paid feature. The visibility toggle, publish panel, and embed code generation all check your plan before letting you use hidden fields. Free plan users see the option but can’t activate it. On the server side, submissions from free plans have hidden field values stripped out, so there’s no way around it. If you’re already on Pro or Business, nothing changes for you.

Content blocks auto-save every edit. The Save button is gone. Every change you make to a content block persists the moment you make it. This matches how the rest of the IntakeBot editor works and removes a step that tripped people up (forgetting to save, losing edits on navigation).

The template variable picker moved. Instead of typing / and scrolling through the slash menu to find field variables, there’s now a dedicated button in the rich editor toolbar. Click it, pick a field, and {{field_name}} drops in. Faster, especially when you have a lot of fields.

IntakeBots can do more than ask questions now. Content blocks let you insert guided text, calls to action, and personalized messages between fields. Every plan gets plain-text blocks. Pro and Business unlock the full rich editor, embeds, and AI-generated content. This takes conversational forms to a whole new level.

The rich editor works like Notion. Type / to open the slash menu and drop in headings, callouts, code blocks, or video embeds. You can embed YouTube, Vimeo, Loom, Cal.com, Calendly, Stripe Payment Links, and Google Maps directly inside the intake flow. Up to three CTA buttons per block can open modal overlays or link to external URLs.

Template variables make all of this personal. Use {{field_name}} anywhere in question labels, helper text, content blocks, or CTA URLs to reference what someone already answered. Arrays, AI-processed values, and all field types work. Private fields resolve to [Private] so nothing leaks.

AI content generation takes personalization further. Write a prompt, and AI generates unique content for each respondent at runtime using their previous answers as context. Good for mid-flow recommendations, summaries, or tailored next steps. Private fields stay out of AI context automatically.

We also published step-by-step integration guides for Meta Conversions API, Google Ads, LinkedIn, TikTok, and custom webhooks. Each covers credential setup, configuration, testing, and troubleshooting. Find them at app.iozen.ai/docs/integrations.

Testing IntakeBots used to mean clicking through every field by hand. Test Autopilot changes that. Hit Auto-fill on any field to get realistic sample data, or hit Auto-play to walk through the entire bot start to finish. It generates locale-aware answers, handles AI follow-up questions with progressively detailed responses, and pauses on file uploads so you stay in control. Normal and fast speed modes let you watch the conversation or skip ahead.

Field duplication is here too. Select any field, click duplicate, and a copy drops in right below the original with a fresh variable name. Validation rules, AI settings, and conditional branches all carry over.

On the i18n front, Account Settings, Email Preferences, and the IntakeBot field editor are now fully translated into English and Spanish. Boolean fields default to Sí/No for Spanish bots, dates follow locale formatting, and test data matches the bot’s language. Under the hood, contact extraction switched to a schema-aware resolution map that eliminates guesswork on nested fields. Email notifications pull from contact mapping data directly, and inactive Postmark recipients get skipped instead of failing the whole job.

Any IntakeBot field can now be marked as hidden. Hidden fields never appear in the conversation, but their values are saved with the submission and can drive conditional logic just like visible fields.

Values come from three places. Embed code attributes (data-fields) work for inline, popup, and widget embeds. URL parameters (?f.key=value) work for direct link sharing. And you can set static defaults in the field editor for values that stay the same across all submissions.

This opens up use cases that were impossible before. Pass a CRM record ID from your app into the IntakeBot so submissions link back automatically. Forward UTM parameters from your landing page into the submission for end-to-end attribution. Set an A/B test variant as a hidden field and branch conditional logic based on it. Track referral sources without asking the respondent anything.

Hidden fields are available on Pro and Business plans.

Three updates to how you manage data inside ioZen. The Contact Import Wizard walks you through uploading a CSV or Excel file and mapping columns to ioZen fields. Duplicate detection runs automatically with skip, update, or fill-empty strategies so your list stays clean.

Workspaces now have plan-tiered storage quotas. Free gets 100 MB, Pro gets 5 GB, Business gets 25 GB. Email alerts fire at 80%, 95%, and 100% usage so you’re never caught off guard. See the pricing page for full storage limits.

File upload size limits also scale by plan: 5 MB on Free, 10 MB on Pro, 25 MB on Business. Records that include file fields respect these limits at submission time.

The ioZen developer platform is here. The REST API lets you read and write submissions, manage contacts, and trigger actions programmatically. API key authentication with scoped permissions keeps your integrations secure. See the Developer Tools guide for an overview.

Interactive docs at app.iozen.ai/docs cover every endpoint with request/response examples you can try from the browser. No guessing at payload shapes.

Outbound webhooks deliver real-time events to your endpoints with HMAC-signed payloads and automatic retries via QStash. When something happens in ioZen, your systems know about it immediately.

API keys are managed from Workspace Settings. Create keys with scoped permissions, rotate them when needed, and stay within per-plan rate limits. Pro gets 3 keys at 100 req/min. Business gets 10 keys at 500 req/min.

The ioZen team can now deploy promotion codes through Stripe. Use them to reward beta testers, support early adopters, or extend special pricing to specific customers. Codes apply to both monthly and annual billing plans.

Multi-select fields in the public IntakeBot chat were showing duplicate chips when picking multiple values. Validation was also rejecting valid selections in some cases. Both issues are fixed. Multi-select fields in conversational forms now render and validate correctly regardless of how many options you pick.

Three billing fixes in this release. A race condition during plan changes could cause subscriptions to enter an inconsistent state. Credit-status polling was firing too aggressively under load. The upgrade dialog was redirecting to the billing page instead of straight to Stripe checkout. All resolved. View current pricing plans.

A round of infrastructure and UX polish. The email system moved to a scalable architecture with separate transactional and broadcast streams via Postmark. Deliverability is better and onboarding sequences are now possible.

Workspace routes show a mobile access gate on small screens, directing users to the desktop experience. A quick email link lets you open the app on your computer without hunting for the URL.

The AI FlowApp generation wizard now shows parallel progress indicators and smoother motion transitions. Generation status is tracked on the FlowApp model so state stays reliable even if you navigate away. IntakeBots created by the generator include all field types and conditional logic from the template.

ioZen moved from private alpha to public beta. Anyone can sign up, pick a plan, and start building. Stripe billing went live alongside this release.

Row-Level Security is now enabled on all remaining public tables. Function search paths are locked down and RLS policies are tightened to prevent cross-workspace data access.

Process Board limits got a bump too. Free goes from 3 to 5 boards, Pro from 30 to 50. This prevents hitting the ceiling during AI-powered FlowApp creation, which generates boards as part of the setup flow.

Three tiers let you add AI credits when you need them. Boost (1,000 for $10), Growth (5,000 for $40 with 20% savings), Scale (10,000 for $70 with 30% savings). Purchased credits never expire, so you use them at your own pace. View all pricing options and credit packs.

A full responsive overhaul across the app. Navigation, panels, boards, and settings all work on mobile viewports now. Legacy layout code is cleaned up.

The iOS keyboard bug that overlapped the chat input in embedded IntakeBots is fixed. Viewport management and iframe height adjust correctly across inline, popup, and widget embed modes when the keyboard appears.

Marketing Attribution Tracking moves to Pro and above. Field-Level Conversions, Webhooks, and Ad Platform Integrations (Meta, Google, LinkedIn, TikTok) are now Business-only. Clearer value separation between pricing tiers.

The core infrastructure for public launch. Stripe billing lets you subscribe to Pro or Business, manage your subscription from workspace settings, and switch between monthly and annual billing with prorated charges. See current pricing plans.

Process Board cards, contact changes, and record updates now sync across all connected users instantly via Supabase Broadcast. Multiple team members can work on the same board without refreshing.

Drag-and-drop on Process Boards was rebuilt with @dnd-kit for smoother card movement, better touch support, and more reliable phase transitions. The embedded IntakeBot widget also adapts to mobile viewports now. Floating button, popup, and inline modes all resize correctly on phones and tablets.

Controlled access for the beta phase. New users need an invitation code to sign up, and workspace collaborators can be added directly. Beta applications are tracked through a dedicated IntakeBot.

IntakeBot field schemas now stay in sync with their linked RecordType automatically. Changes to one propagate to the other, reducing manual work and preventing data mismatches between what gets collected and what gets stored.