Release Notes

Documentation Index

Fetch the complete documentation index at: https://docs.emanate.ai/llms.txt

Use this file to discover all available pages before exploring further.

Version history and updates for Emanate.

---

​

v2.10.0 (2026-05-21)

​

Highlights

The Leads and Accounts pages are rebuilt around saved views, a refreshed layout, and a personal My Inbox scope. Sequences gain channel combos for parallel email + LinkedIn sends, smart reply detection that pauses cadences the moment a contact responds on any channel, and LLM-generated sequences drafted straight from an Expand job. A new Send to Sequence flow promotes Expand contacts into canonical leads, drafts a sequence, and enrolls them in one motion. Plus workflows can now route across multiple inboxes, and submitted quotes are editable with teach-back. The Win-Back engine moves beyond fully-dormant detection to the three-type framework — Type A (fully dormant), Type B (product lapse), and Type C (cherry-picking) — so you can recover share inside accounts that are still buying but quietly losing programs to competitors. Plus: campaigns get font sizes, PDF attachments, and managed sender signatures; every inbox thread is auto-paired with the right lead/account via a clickable Linked to badge; inbox configuration gets per-inbox forwarding and ignore-subject regex; and the AI Model picker now exposes GPT 5.5 and Claude Opus 4.7.

​

New Features

• Leads & Accounts overhaul: Both pages have been rebuilt with a unified layout, a ViewSwitcher dropdown for saved views, and a personal My Inbox scope that hides leads claimed by other reps. Saved views persist filters, sorts, and column selection per user.
• Sequences — Channel combos: A single sequence step can now fan out across email and LinkedIn in parallel, rather than picking one channel per step. Build cadences that hit a contact through multiple touchpoints on the same day without authoring a step for each.
• Sequences — Smart reply detection: Inbound email and LinkedIn replies are now wired directly into sequence conditions. When a contact responds on any channel, the sequence pauses or branches automatically — no manual cleanup required.
• Sequences — Generate from an Expand job: Open any saved Expand search and have Emanate draft a full multi-step sequence tailored to the persona and search intent. The draft lands in the sequence builder pre-populated and ready to review.
• Send to Sequence from Expand: A new Send to Sequence sheet promotes selected Expand contacts into canonical leads, drafts a sequence, and lands them in your My Inbox in a single action. The review screen uses the live sequence-builder preview, so what you see matches what will send.
• Workflow - Multi-inbox routing: Workflows can now route across multiple inboxes in a single deployment, and a new AI handoff toggle on each routing branch controls whether the agent should draft, reply, or hand off to a human.
• Quote line-editing with teach-back: Submitted quotes are now editable — adjust line items after submission and Emanate captures the change as a teach signal that improves future quoting on similar requests.
• Win-Back Three-Type Detection: The old per-account median-gap classifier (active / at-risk / dormant based on 1.5× / 6× the account’s median order gap) is replaced by a per-org configurable three-type signal model. Type A (fully dormant) is on by default with explicit gap-days + lifetime-revenue thresholds; Type B (product lapse) and Type C (cherry-picking) are opt-in with their own thresholds. Each detected signal carries detected_at / last_seen_at / recovered_at, and recovered signals re-emerging are reopened automatically. See Win-Back.
• Win-Back Enrollment Column: The Status column on the Accounts table has been renamed to Win-Back Enrollment with three values — Auto (driven by detection), Yes (force-include), and No (force-exclude). Auto rows show the currently-resolved state on the popover button so you can scan without opening the drawer.
• Per-Type Re-Engagement Drafts: Each digest now creates one draft sequence per active type, pre-loaded with the right outreach angle (reactivation for A, supply continuity for B, advantage pitch for C), pre-enrolled with primary contacts, and held in draft state. Accounts already in another active sequence are skipped to prevent double-enrollment.
• Add Account by LinkedIn URL: Paste a LinkedIn company URL into Create Account — Emanate looks up the company, pulls firmographic data, and pre-fills the dialog. Falls back to a stable contact-lookup ID when no canonical domain can be resolved. Plus an unlimited-results name search bar inside the dialog. See Accounts Overview.
• Bulk Enroll Account Contacts into a Sequence: Multi-select contacts inside the account drawer’s People section and bulk-enroll them into any active sequence. Contacts already in the chosen sequence are skipped automatically. See Account Contacts.
• Linked to Lead / Account Badge on Inbox Threads: Every email thread is automatically paired with the lead or account whose contacts are on it. The Linked to badge on the thread row deep-links straight into that drawer with the thread already scrolled into view. See Agent Inboxes — Linked to.
• Inbox-Level Forward Setting: Forward a copy of every incoming message on an agent inbox to a stakeholder address — configured per inbox under Email → Domains → Inbox Settings. Distinct from broadcast-copy forwarding (which forwards every campaign send). See Forwarding Out of the Agent Inbox.
• Ignore-Subject Regex Patterns: Filter out automated invoices, calendar invites, support-bot replies, and other noise before the agent ever sees them. Configure one regex per line under Inbox Settings → Ignore subject patterns. See Inbox Setup.
• GPT 5.5 + Claude Opus 4.7 in AI Model Tab: Both new models are now selectable per inbox under AI Settings → AI Model. Claude Opus 4.7 is particularly strong on long-context threads and nuanced procurement / RFQ replies. See AI Model Selection.
• Campaigns: Font Size + PDF Attachments + Sender Signature: The Classic email editor now supports font-size selection from a curated scale, multiple PDF attachments per broadcast, and an auto-applied sender signature. See Email Campaigns — Formatting and Attachments.
• Managed Email Signatures for Sequences: Sequences and campaigns now share a single managed signature per inbox — edit once on the Email Settings page and every active and draft sequence picks up the change on the next send. Personalized blocks are signed before being wrapped in the campaign HTML shell, so the signature always ends in the right place. See Sequences — Managed Email Signatures.

​

Improvements

• Sequences — 50-lead cap removed: The previous per-enrollment ceiling has been lifted, with a sweep cron and staggered bulk-defer keeping send rates within provider limits on large enrollments.
• Email Insights — Participant-side counting: Engagement counts are now derived from the participant side of each message and source pagination loads the full window, so highlights reflect the complete date range instead of a truncated slice.
• Always-Initials Avatars: Contacts without a photo show their initials instead of a generic avatar across the contacts list, drawer, and decision-maker results — easier scanning at a glance.
• Remove a Contact from an Account: The Contact Drawer now has a Remove from account action that detaches the contact without deleting them from your workspace. The contact’s record stays available for re-attach or move to another account.
• Stable Contact-Lookup ID Persisted End-to-End on Account Contacts: Contact enrichment now persists the lookup-provider’s stable contact ID on the contact record, so re-running Contact Enrich on the same person uses the existing enrichment payload instead of triggering a fresh search.
• Workflow Squads — Instructions Limit Raised to 5000 Characters: Long agent instructions no longer hit the truncation boundary when pasting from prior versions or external docs.
• Sequence Signature Stability on Empty Orgs: Sequences in workspaces that haven’t configured a signature yet now fall back to the inbox display name + reply-to so empty-state sequences still send something usable.
• All Inboxes Sent Tab + Multi-Grant Coverage: The Sent view across all connected inboxes now correctly aggregates messages from orgs with multiple OAuth grants on the same Google / Microsoft account.
• All Inboxes Prefetch Cap Raised (30 → 100): Larger orgs with many inboxes load full inbox previews without an extra round-trip.
• Quoting Auto-Linkage to Accounts: Submitted Oracle quotes now stamp a lead_id on creation and auto-link to the matching account via the customer-number ↔ email-domain registry, so quote activity feeds the right account’s interaction history.
• Oracle ERP Retry Stuck Quotes: Failed or stuck quote submissions can now be retried from the UI. Duplicate-retry protection prevents approved quotes from being resubmitted accidentally.
• Submitter on Quote Detail Panel: The quote detail panel now shows who submitted the quote, including org-member display names.
• Lead Website Domain Normalization: Lead website URLs now flow through the same canonical-domain normalizer that powers account matching — https://www.example.com/, http://example.com, and bare example.com all resolve to the same key, so dedup, enrichment, and segment membership stay consistent.

​

Bug Fixes

• Win-Back Migration Hardening: Win-back migration timestamps were bumped to avoid main-collision and the signals RLS policy was made idempotent so re-running migrations on staging branches doesn’t error.
• Skip Win-Back Digest for Accounts in Active Sequence: Accounts that are already enrolled in an active re-engagement (or any other active sequence) are no longer re-enrolled on the next digest cycle, preventing duplicate touches.
• Person Columns Route to Contact Fields When Target = Accounts: Imports targeting accounts with sheets that have person-shaped columns now correctly write to contact_* fields on the auto-created primary contact, instead of leaving person values stranded on the account row.
• Padded Empty Rows in XLSX Streaming: Empty leading rows in XLSX exports no longer break headerRowIndex detection during streaming parse.
• Dedup Prefetch Pagination (P0): Dedup prefetch was silently capped at 1000 rows for orgs with larger workspaces — now paginates through the full set so fuzzy-dedup decisions are made against the full account base.
• Dropped Dead account_orders.deal_id Column: A vestigial column on account_orders that wasn’t referenced by any code path has been removed, with cascade indexes added on the columns that matter.
• Single-Deal Delete Wired End-to-End: Deleting a single deal from the account drawer now refreshes the deals list and the account stats row correctly.
• Onboarding Redirect After OAuth Return: First-run users completing the LinkedIn / Email OAuth handoff no longer land on a stale onboarding step — the gate logic now reads completion state from the OAuth callback before deciding where to send the user.

---

​

v2.10.2 (2026-05-27)

​

Highlights

A pre-release iteration that tightens the Monitor experience around how it shows up across the product, restores live web search on the enrichment fallback path, and lets you see where enriched values came from. Plus a sweep that scrubs the underlying enrichment provider’s brand name from every user-facing surface — the data still flows the same way, the copy just stops naming the vendor.

​

New Features

• Account-Page Monitor Sidebar + Bulk Attach (EMA-305): A right-side Monitors sidebar lives on the Accounts page and lists the org’s account-scoped Monitors with member preview and status. Create one with signal preset chips (CEO, Funding, Hiring, News, Leadership changes, Product launches, New contracts) plus an optional free-form prompt, then bulk-select rows in the table and Attach to Monitor from the toolbar. Caps: 20 monitors per org, 10 members per monitor, one monitor per account (moves are surfaced via confirm). Membership changes auto re-anchor the underlying watch with last-write-wins + a 10-second debounce. Empty membership reverts a monitor to draft rather than soft-deleting so you can reattach later. Opt-in Monitor column on the accounts table (off by default) shows which monitor an account belongs to. See Account-Page Monitor Sidebar.
• Monitor Outcome-Based Type Labels (EMA-330): The Monitor create/edit dialog now offers two outcome-framed choices instead of the abstract Event Stream / Snapshot terminology — Find New Events (“notified whenever something new happens or appears”) and Track Changes (“you already have a specific company, project, list, or page and want to know when something changes”). The Lite tier picker has been removed from the form; new monitors default to the base compute path. See Monitor Types.
• Edit Cadence + Signals on a Running Monitor (EMA-305): An Edit button on the Monitor detail view lets you change cadence and (for account-page monitors) signal preset chips after a monitor is running. Saving an edit re-runs a baseline scan (~5–10 minutes) so the new configuration has a fresh anchor. The natural-language query itself remains locked once registered. See Editing a Running Monitor.
• Source Citations on Hover (EMA-326, EMA-283): Enriched Revenue, Employees, and Industry cells now reveal a hover tooltip with the citation URL, page title, and confidence the lookup used to ground the value. Rows enriched before the citations layer shipped keep the tooltip hidden so the cells stay clean. See Hover for Source Citations.
• AI Score Docs — Heat Tiers + UI Walkthrough (EMA-185 follow-up): The ICP Scoring page now documents the full AI Score → Grade (A / B / C / D) → Heat (Burning / Hot / Warm / Cold) mapping plus what you see in the cell (grade badge + heat icon + heat label) and the popover (score breakdown + why-this-account + why-now).

​

Improvements

• Restore Live Web Search on the Enrichment Fallback Path (EMA-314): When the primary enrichment provider has nothing for a row, the fallback path now uses an OpenAI Responses-API call with tools: [{ type: "web_search" }], letting the model look up live web data again. The primary path was unaffected — it still handles the bulk of enrichment traffic with its own grounded-search support.
• Backend LLM Migration to GPT + Full Cost Instrumentation (EMA-310, EMA-311): Every backend claude-* runtime call site has been swapped to a GPT equivalent, and every LLM call across model families (chat completions / Gemini / embeddings / image-gen) now lands an observability trace with a non-zero costUsd. This is a cost-control + observability change — the user-facing AI Model picker for inbox agents still exposes Claude Opus 4.7 and GPT 5.5, since those are managed by the voice/email-agent runtime separately. Coverage at HEAD: 75 / 75 LLM caller files, 12 / 12 model literals.
• Brand Scrub Across Monitors, Snapshot UI, Enrichment Errors, and LLM Prompts (EMA-298, EMA-299): User-facing copy no longer names the underlying enrichment provider — strings on the Monitor info popover, the snapshot setup layout, the public llms.txt, health-check errors, the lib-level error surface, and the Expand LLM prompts have all been genericized. Internal field-source identifiers are unchanged.
• Swap Orders and Deals Tab Order in the Account Drawer (EMA-301): New tab order is Details → Contacts → Orders → Deals → Notes. Driven by a prod check showing account_orders carries ~3.4× the rows of account_deals for industrial workspaces, so the higher-signal tab gets the closer position to Contacts. Lead drawer tab order is unchanged. See Account Details.
• Unify Account Deletion via a Single delete_account RPC (EMA-292): Deleting an account from the UI now removes the entire row and lets FK cascades clean up children, replacing the older session-scoped unlink semantics. The remove_account_from_session RPC stays in the DB for non-UI callers (session-revert flows).
• Remove the enrich-pending-leads-cron Runaway Job (EMA-302): A scheduled cron that was double-firing enrichment for already-queued leads has been removed. Single-source-of-truth enqueue paths now drive every enrichment run.
• LLM Observability — Runtime Context + Per-User Attribution (EMA-303): Every LLM trace now carries the calling org’s slug and the requesting user’s ID, so cost dashboards can split spend by org and by individual rep.

​

Bug Fixes

• Source Hover Was Silently Disabled on Enriched Cells (EMA-326): A defensive Tailwind selector added during an earlier table refactor was inadvertently disabling pointer-events on every nested hover-card trigger, including the FieldSourceTooltip. Selector removed — Revenue / Employees / Industry hover works again.
• Backup-Key Fallback on Per-Column Primary Contact (EMA-325): The per-column contact-lookup path now uses the same backup-key rotation as the bulk enrichment pipeline when the primary key is rate-limited or out of quota, so a single quota event no longer leaves a row stuck with — on the primaryContact column.
• Demote Phone-Only Primary Contact on Discovery (EMA-325): When per-column contact-lookup discovers a real contact for an account that previously only had a stash-fallback phone number, the older phone-only row is demoted automatically rather than left as a stale “primary.”

---

​

v2.10.1 (2026-05-20)

​

Highlights

A point release focused on the Enrichment 2.0 rebuild — every enrichable field on Leads and Accounts now has a dedicated per-column Fill action with prerequisite checks that disable the button when inputs are missing. The bulk Enrich button retargets based on what’s checked, revenue is normalized to $XK / $XM / $XB everywhere, the Accounts table gets a sortable Timestamp column, and Email Insights metrics move to a participant-side counting model so CC/BCC threads contribute to the right accounts and reps.

​

New Features

• Per-Column Fill (EMA-217): Each enrichable field has its own icon-only Fill action on the column header (with per-row variants where useful). Four pipelines share the registry — web research (industry, revenue, employees, founded, segment, funding), company resolver with Claude 4.6 fallback (domain, website, companyName, address), GPT (whyThisAccount, whyNow, priorityAction), and contact lookup (primaryContact: name/email/phone bundled). Each cell tracks its own lifecycle (queued → filling → idle) so partial refreshes never block the row. See Account Enrichment — Per-Column Fill.
• Per-Field Prerequisite Check (EMA-257): Fill buttons are now disabled with a tooltip when any visible row is missing what the pipeline needs to ground the lookup — company name and/or website/domain for the web-research and contact-lookup paths, company name for the resolver and GPT paths. The tooltip names the specific missing fields so you know what to fix. See Prerequisite Check.
• Selection-Driven Enrich Button (EMA-260): The top-of-table Enrich button on Leads and Accounts now retargets based on what’s checked — no selection runs the legacy first-50-of-filtered scope, individual checkboxes enrich exactly those rows, and select-all-across-pages enriches the full set (minus exclusions) capped at 50. The button label updates live with the count. See Bulk — Selection-Driven Enrich Button.
• Revenue Normalization (EMA-263): Revenue is stored as a raw integer and rendered everywhere through a single formatRevenue helper — $1.5B, $50M, $250K, or — for empty/zero/negative. The enrichment prompt is now constrained to return a clean integer (no $, ranges, or “approx”), and pre-existing messy strings are normalized on read. See Revenue Formatting.
• Accounts Timestamp Column + Sort Toggle (EMA-249, EMA-210): The Accounts table now ships with a Timestamp column sourced from each row’s earliest linked import session (falling back to created_at), rendered to the second. Default sort is newest-first; click the header arrow to toggle ASC/DESC. Merged accounts inherit the survivor’s earliest session, so merges don’t reshuffle the top of the table. See Timestamp Column and Default Sort.
• Quote-Account Row Creation Time (EMA-244): Accounts auto-created from inbound quote submissions now carry an explicit row-creation timestamp that feeds the Timestamp column correctly, instead of inheriting the quote’s submission time.

​

Improvements

• Email Insights — Participant-Side Counting Model (EMA-255): All four headline stats (Avg Response Time, Emails Sent, Emails Received, Unique Recipients), the highlights, and Account Coverage now decide sent-vs-received from From / To / Cc / Bcc participants rather than the provider’s stored direction column. CC and BCC threads now correctly count toward the right rep and account. See How Sent vs Received Is Decided.
• Email Insights — Most Active Account Cleanup (EMA-255): The Most Active Account highlight drops the “across N reps” suffix when N is 0 or 1, so single-rep orgs and accounts touched by a single rep get a clean headline number.
• Email Insights — Postgres-Aggregated Stats Cards (EMA-246): The stats-card row aggregates entirely in Postgres with paginated source fetches, so the metrics row loads independent of the larger insights query and stays fast on multi-thousand-message windows.
• Lead/Account Contact Model Unified (EMA-242): Leads and Accounts are both companies with one or more contacts, with every UI surface, server action, and write path routed through the lead_contacts table. The legacy top-level agent_leads.contact_* columns are no longer read or written anywhere, eliminating an entire class of “primary contact didn’t update” bugs. See Accounts vs Leads.
• Order Aggregates Recompute on Merge (EMA-262): Merging two accounts now triggers a recompute of total_order_count, lifetime_revenue, lifetime_tons, yearly revenue, first/last order date, peak year, and median order gap on the survivor — UI no longer shows stale or empty Orders tabs after a merge.
• Fit Score Documentation (EMA-185): The AI Score is now explained as an LLM synthesis over five weighted insights (Firmographic Fit, Recent Activity, Pain Points / Needs, Decision-Maker Access, Competitive Landscape) rather than a hand-tuned linear formula, including the score-to-grade and score-to-heat mappings the Leads and Accounts tables use. See ICP Scoring.
• Win-Back Signal Naming (EMA-185): Win-back signal variables in templates and digests use generic product-shape names, with explicit sender alias support so multiple reps can share the same outbound mailbox without losing per-sender attribution.

​

Bug Fixes

• Most Active Account “across 0 reps” Suffix Hidden: The Most Active Account highlight no longer renders an “across 0 reps” / “across 1 reps” suffix on small orgs.
• Accounts Timestamp Migration Ordering: EMA-249 / EMA-244 migrations were renumbered past 20260519010000 to clear collisions with the EMA-217 chain.
• Email Insights direction Column Ignored: The stored direction field on msgraph_messages (historically unreliable for forwarded sends) is no longer consulted; participant-side computation is the single source of truth.
• Per-Column Fill Lifecycle Reliability: Industry-fill lifecycle status now persists filling across background-task retry attempts so the cell badge doesn’t flicker between queued and filling.

---

​

v2.9.1 (2026-05-03)

​

Highlights

A new Order History capability for accounts: import historical shipments, see lifetime revenue and buying cadence in the drawer, and have Emanate watch for accounts that have gone quiet. When an account drifts past its normal order gap, you get a digest email plus a pre-built win-back sequence ready to launch. Plus tighter email auto-response controls down to the lead, account, or thread level, and a one-click path from monitor events to Expand drafts.

​

New Features

• Account Orders & Activity Status: Import historical sales-order or shipment data through the standard Import Wizard — the AI classifier recognizes ERP exports automatically and rolls line items up to order headers. Each account gets a new Orders drawer tab with a status badge (Active / At Risk / Dormant), a stats row (lifetime revenue, peak year, typical gap), a switchable Revenue/Tons chart with time-range dropdown, and a recent-orders list. See Orders.
• Inactive-Account Digest Email: A scheduled digest emails the current list of at-risk and dormant accounts to anyone in your org. Frequencies range from every 5 minutes to weekly. Configure under the Accounts page ⋮ menu → Email notification. Includes a heartbeat email when nothing is inactive, so you always know the monitor is alive.
• Auto-Generated Win-Back Sequence: Every digest creates a fresh draft sequence pre-loaded with the Re-engagement template (7 steps), pre-enrolled with the inactive accounts’ primary contacts, and held in draft state. Click straight from the email to review and launch.
• Per-Lead and Per-Account AI Auto-Response: The Per-Record AI Controls introduced in v2.9.0 are now configurable from the Interactions tab on lead and account contact drawers, plus from the thread view itself. Override the agent, mode (auto-draft vs auto-reply), Knowledge Base, and inbox-history toggles for an individual record. Precedence is Thread > Lead > Account > Inbox default — see Inboxes — Per-Record AI Overrides.
• Monitors documentation: Monitors now have their own docs section covering creating monitors, tag management, the events timeline, and the new one-click Expand Drafts flow.
• Monitor → Expand Drafts (one-click): When a monitor surfaces an event, hit Generate Expand draft on the event card and Emanate writes a search prompt and deep-links you into the Expand editor with the prompt pre-filled. Idempotent per user — clicking again on the same event opens the existing draft. A new refresh icon regenerates the prompt while preserving the search ID. Plus a calendar date picker on the timeline header with per-day event counts. See Expand Drafts.
• lead_contacts as a First-Class Import Target: The Import Wizard now recognizes lead_contacts sheets directly — useful when a sheet is structured as one row per person (rather than one row per company). Parent leads are auto-created from the company column, contacts are inserted, and the first contact per lead is marked primary. The flat agent_leads import path also writes a primary contact inline.

​

Improvements

• Larger imports backed by async parsing: The 25 MB → 200 MB import limit shipped in v2.9.0 is now backed by a dedicated parsing task that runs off the request path. Status badges in the wizard show uploading → parsing → classifying → … so progress is visible end-to-end on multi-hundred-thousand-row sheets.
• Streaming XLSX parser: Spreadsheet parsing has been rewritten to stream rows rather than load the entire workbook into memory. Real-world performance: a 44 MB / 118k-row XLSX completes in ~48s end-to-end, where the previous parser would have stalled the heartbeat.
• Email threading reliability: Replies sent through a connected Gmail or Outlook inbox now consistently land in the same conversation on the recipient’s side, including for the second and subsequent replies in a thread. RFC In-Reply-To and References headers are captured on inbound and propagated on outbound.
• Contact drawer email coverage: The Interactions tab on a contact drawer now shows every Gmail/Outlook thread the contact is on (sender or recipient), with thread-level grouping, inbox badges per row, an inbox filter dropdown, and a refresh button that backfills missing messages on demand.
• Pending AI drafts in the contact drawer: Auto-drafted replies now appear inside the thread view of the contact drawer, not just in the unified inbox — review, send, or dismiss without leaving the lead/account view.
• Hidden provider column on the Email Integrations table: Cleaner default view; the data is still available in row details.

​

Bug Fixes

• Orphaned needs_review self-heals: When a duplicate was deleted while its surviving sibling was still in needs_review, the surviving lead/account is now automatically reset to pending so the badge clears and the row becomes editable. A one-time backfill repaired existing affected rows.
• Inline lead contacts on flat imports: Flat agent_leads sheets now reliably create a primary lead_contacts row per lead — previously the contact was lost in some classifier paths.
• Switching import targets retains data: Toggling the wizard target from accounts to agent_leads (or vice versa) now correctly remaps Company and Contact Name columns instead of leaving company values in person fields.
• Production import crash: Fixed a background-task bundling issue where xlsx and exceljs were not externalized correctly, causing prod imports to fail on launch.

---

​

v2.9.0 (2026-04-30)

​

Highlights

A new Map page brings geographic intelligence to accounts and leads, autonomous email agents can now read, draft, and reply on their own with a long-range memory layer, the Knowledge Base is rebuilt as a unified table with side drawers, and a new Quote Generation pipeline turns RFQs into reviewable drafts.

​

New Features

• Geographic Map Intelligence: A new Map page under Intelligence → Map plots accounts and leads on an interactive map. Cluster, zoom, and filter by segment, and click any pin to open the account or contact drawer
• Autonomous Email Agents: Configure agents that read, draft, and reply to email threads on their own. Each agent connects to a dedicated inbox, follows your sequence rules, and can be set to draft-only or fully autonomous on a per-thread basis
• Long-Range Memory for Email Agents: Agents now have a long-term memory layer that retrieves relevant historical conversations when drafting — so replies reference prior commitments, past objections, and the full account relationship rather than just the current thread. Configurable history window from 1 to 15 years, scoped per agent and per inbox, with a background backfill that keeps memory fresh without slowing the UI
• Per-Record AI Controls: Toggle AI assistance independently for each lead, account, or thread. Disable AI on sensitive accounts while keeping it on elsewhere, or override at the thread level when a conversation needs human-only handling
• Sequences Template Library v4: Modernized starter sequences with cleaner copy, better defaults, and a refreshed header layout
• LinkedIn Media Link in Sequence Steps: Attach a LinkedIn URL to a sequence step and have it render as a rich media card in the recipient’s inbox
• Knowledge Base Redesign: Rebuilt as a unified table with side drawers and visual search. Single table view across documents, files, and notes; inline drawers replace full-page navigations so you can preview and edit without losing context; visual search surfaces results by content, not just filename
• Quote Generation: New automated quote pipeline for industrial and configure-to-order customers. The agent ingests a request for quote, produces a structured draft, and routes it to a review queue where a human can approve, edit, or reject before it goes out. Connection setup (wallet + connection test) is built into the Integrations page
• Forward Broadcast Copy: Automatically forward a copy of every campaign broadcast to a stakeholder address
• Larger Imports (25 MB → 200 MB): Import larger CRM exports without splitting files

​

Improvements

• Bulk Discovery Match Limit (25 → 100): Bulk searches return up to 4× more results per run
• Website Enrichment: Improves match quality on bulk discovery
• Stuck Job Recovery: For long-running discovery runs
• Workspace & UI Polish: New shared Storybook for UI primitives (internal, but powers a more consistent look), refreshed loaders, empty states, and dialog/sheet padding, tighter vertical spacing and surfaced status badges across Leads and Accounts, and smaller-viewport optimizations for laptops and split-screen workflows

​

Bug Fixes

• Fixed dialog race conditions on the leads three-dot menu
• Resolved several layout-shift issues across the product

---

​

v2.8.1 (2026-04-27)

​

Highlights

UX polish on the lead and account drawers — every contact is now clickable and openable, Find Decision Makers returns far more results with proper ordering, and old import sessions show up reliably in the session card list.

​

Improvements

• Find Decision Makers — bigger, better-ordered results: The Decision Makers search in the account/lead drawer now returns up to 100 contacts (was 20), ordered from most senior to most junior. Leaving the title filter blank now searches all roles instead of silently filtering to executives. Works on both leads and accounts, and resolves a domain automatically when only the company name is on file
• Click any contact in the drawer: Non-primary related contacts in the lead drawer’s People section are now clickable — they open the Contact Drawer with full inline edit, enrich, discrepancy-resolve, and delete capabilities. Matches the existing account behavior

​

Bug Fixes

• Old import sessions visible again: The session card list on the Leads and Accounts pages was capping at the latest 20 sessions (and including dismissed/reverted ones at the top), which hid older XLSX imports for orgs with many sessions. The list now filters out dismissed/reverted sessions by default and surfaces the full set of active imports. The Import History drawer can still see hidden sessions so its dismiss controls keep working

---

​

v2.8.0 (2026-04-25)

​

Highlights

A revamped enrichment pipeline with verified addresses and duplicate detection, Email Insights for team email analytics, custom fields on accounts, session-scoped custom columns, and a richer broadcast workflow with duplicate and view-sent actions.

​

New Features

• Email Insights: A new Insights tab on the Email page aggregates outbound and inbound email across your connected inboxes and broadcast sends. See average response time, daily volume, an hour-by-day heatmap, top responders, unanswered threads, stale accounts, per-rep performance, and per-account coverage with sentiment indicators. Filter by date range and rep
• Verified Addresses with Alternate Review: Enrichment now returns verified billing addresses (street, city, state, postal, country) plus a specific business segment label. When the new address differs from an existing one, it’s stored as an alternate with Accept / Keep Existing buttons rather than silently overwriting
• needs_review Status: A dedicated amber-pill state for rows that require human attention — surfaced in the column picker, the status filter, and an N need review count badge next to the enrichment ETA. Triggered by suspected duplicates, alternate addresses, or low-confidence fields
• Duplicate Detection for Leads & Accounts: When enrichment resolves to a domain already claimed by another row, the Duplicate Review dialog offers Keep Separate, Merge, or Delete. Auto-enrichment on import skips matched/duplicate rows (you can still enrich them manually)
• Enrichment Throughput: Enrichment runs in batches of up to 100 leads or accounts at a time, with batch size adjusted in real time based on available resources. Applies to all enrichment — auto, manual, and bulk
• Reliable Phone Reveal: Direct mobile numbers populate automatically during enrichment, with a built-in fallback to a secondary source if the primary lookup is slow. The Mobile column now fills in reliably for both leads and accounts
• Custom Fields on Accounts: Full parity with leads — add custom columns of any type, filter and sort the table on them, edit inline in the drawer, and use them in audience segments
• Session-Scoped Custom Columns: Custom columns now belong to the import session that created them (cap of 10 per session). The “All Leads” / “All Accounts” views show the union of every session’s columns read-only, while session views allow full read/write. Import wizard auto-registers all mapped columns
• Editable Session Names: Hover the page title on a session view and click the pencil to rename — useful for distinguishing imports like “Q2 NorCal manufacturers” from “Apr 18 trade-show list”
• Account Drawer: Company / Intelligence Sub-Tabs: The Details tab is split into a Company sub-tab (basic info, custom fields, attachments, firmographic enrichment, metadata) and an Intelligence sub-tab (Why This Account, Why Now, Priority Action, AI fit score and grade)
• Account Attachments: Upload, download, and delete files on accounts — proposals, signed agreements, decks — mirroring the attachments experience on leads
• Duplicate & View Sent Broadcasts: The ⋮ menu on any broadcast row clones it as a new draft. Sent broadcasts are now clickable and open a full-screen detail dialog with the rendered HTML preview and metadata. Works across Classic, AI Personalized, lead, and account broadcasts
• Audience Tab Restructure: Audiences are now organized into General contacts & audience (with nested Contacts | Segments inner tabs), Lead audience, and Account audience. A unified Add to audience dialog on the Leads and Accounts tables handles both create-new and add-to-existing in one place
• Monitor Tag Management: The sidebar tag filter is now a popover dropdown with multi-select. Monitor detail views have an inline + button to add or remove tags without opening the edit dialog. The monitor form replaces free-text tag entry with a combobox over your org’s existing tags

​

Improvements

• Domain Resolution Pre-Stage: Before company and contact lookups run, an AI step resolves messy company names — abbreviations, misspellings, internal customer-list short codes — into canonical domains. Significantly improves match rates on rows that previously couldn’t be enriched
• Customer-List Acronym Resolution: Internal short codes like “DOFASC” auto-expand to full company names before enrichment runs
• Enrichment Review Dialog: After each batch, a snapshot-diff view tags every changed field as match, auto_filled, kept_existing, mismatch, or uncertain, with synthetic alternate-address cards inline so discrepancies can be resolved in a single pass
• Primary Contacts Protected: Newly discovered decision-makers land as secondary contacts when a primary already exists. Phone-only primaries created during import are preserved instead of being overwritten
• Broader Decision-Maker Discovery: Enrichment now searches more thoroughly for decision-makers at the target company and selects the most senior contact available
• Lead Contacts Table: Replaces the previous embedded JSON with a relational contacts table; primary tracking and the drawer Contacts tab now read and write the proper table
• Enrichment ETA: Updated to ~12 minutes per batch to reflect the deeper pipeline

​

Bug Fixes

• Fixed a black-overlay flash on the leads three-dot menu the first time it was opened
• Fixed campaign segmentation UI alignment issues
• Fixed an issue where stopping a research run could leave the affordance hanging
• Resolved a tag-list scroll regression in the monitors sidebar

---

​

v2.7.0 (2026-03-20)

​

Highlights

Emanate now supports live transcript streaming, full account management with semantic search, Microsoft shared inbox routing, multi-channel outreach sequences, a unified email inbox, lead & account broadcasts, an images knowledge base, AI typeahead for search, email delta tracking, and a unified Logs page.

​

New Features

• Live Transcript Streaming: Real-time transcript streaming for voice and chat agents. See conversations unfold live with a pulsing “Live” indicator, auto-scrolling messages, and same-role message merging
• Account Management: Full account management with a filterable/sortable table, source-based routing, bulk operations, CSV export, and tag management. The enrichment pipeline auto-discovers contacts, scores accounts with fit/timing intelligence, and enables natural language semantic search across your account base
• Microsoft Shared Inbox Email Routing: AI-powered email classification routes incoming shared inbox emails to the correct workflow agent. Includes auto-draft/auto-reply modes, handoff attachments with case briefs, multi-turn conversation support, and a routing dashboard with confidence scores
• Lead & Account Broadcast: Send personalized emails to lead and account segments using merge tags. Create segments directly from the Leads or Accounts page, manage audiences in Campaigns, and send broadcasts with per-recipient personalization
• Images Knowledge Base: Upload images and PDFs to a visual knowledge base with AI-powered analysis. Automatic region detection, detail extraction, semantic search, PDF layer controls, and inline metadata editing in a gallery view with detail drawer
• AI Typeahead for Search: Inline AI-powered typeahead suggestions as you type search queries. Context-aware completions factor in your current data so suggestions stay relevant. Tab to accept, Esc to dismiss
• Email Delta Tracking: Tracks the difference between AI-generated email drafts and what you actually send. Classifies edits by category (tone, length, content, structure) with word-level diff highlighting. New “AI Learning” tab on the Logs page surfaces quality metrics
• Microsoft Graph Email Attachments: Full attachment support for Microsoft Graph emails — metadata sync, clickable attachment chips, full-screen preview (images inline, PDFs native, download fallback), and paperclip indicators on thread rows
• Logs Page: Unified Logs page combining Calls, Chats, and LLM traces into a single tabbed view. Click any trace row to inspect details in a side sheet
• Sequences: Multi-Channel Outreach Engine: Build automated outreach cadences spanning email and LinkedIn from a single conversational AI builder. Describe your goals and Emanate generates and executes the full sequence — enroll leads individually or from segments, track progress step-by-step, and let the system handle timing and delivery
• Connect Your Email Inbox: Link your Gmail, Outlook, or IMAP account directly to Emanate. All your email threads appear in a unified inbox with search, auto-refresh, and the ability to compose and reply without leaving the platform. New messages sync in real time with a 30-day backfill of recent history
• Unified AI Import Wizard: A single AI-powered import wizard now handles leads, accounts, and knowledge base files. Upload spreadsheets or CSVs and the system automatically classifies columns, detects header rows, and maps data to the right fields. Import history now shows per-file details with the ability to delete records from individual files
• Email-Powered Workflows: Workflows can now be deployed to email inboxes. Choose an inbox, set the mode to auto-reply or auto-draft, and your AI workflows process incoming emails automatically. The AI builder generates email classification and routing workflows out of the box
• Draft Review in Email Threads: When a workflow generates an email draft, it appears directly in the thread view with Send, Edit, and Dismiss actions — review AI-generated replies in context and send with one click
• Two-Way Draft Sync: Drafts created in Emanate now appear as real drafts in Gmail and Outlook. Edits, sends, or deletes from your email client stay in sync. Per-inbox toggles control whether AI agents use inbox history and knowledge base as context
• AI Case Brief Attachments: When an AI agent drafts an email reply, a Case Brief document is automatically attached with a structured summary — who the customer is, issue type, urgency, sentiment, key facts, and actions the AI has taken
• Email Open Tracking: Broadcast emails now track opens automatically with per-recipient open data on campaigns, enabled by default on new domains

​

Improvements

• Sequences: Test Mode & Step Preview: Test sequences before activating by sending preview emails to a sandbox address. Preview any step (email, LinkedIn, wait) with rendered content. New activation confirmation dialog shows enrollment count and send limits. Redesigned sequence cards with visible status indicators
• Smarter Segments with AI Reasoning: The segment builder now streams AI thinking in real time as it discovers audience patterns. Edit patterns through natural language chat, see AI suggestions as you build, and watch the reasoning unfold step by step

---

​

v2.6.0 (2026-03-06)

​

Highlights

Your knowledge base now uses smarter search across files and CRM data, the Campaigns and Metrics pages have been completely redesigned, and chat agents can send SMS.

​

New Features

• Smarter Knowledge Base Search: Your chat agents now use smarter search across your uploaded files and CRM data, files are automatically available when uploaded and removed when deleted
• Multi-File Import Wizard: You can now upload mixed file types (XLSX, CSV, PDF) in a single import session with a step-by-step wizard and import history tracking
• Lead Dynamic Intelligence: The leads table now shows industry-relevant columns (Employees, Revenue, Location, Founded, Industry, Tags) from automatic enrichment, plus you can create custom columns and toggle enrichment stages on or off
• Dual-Mode Broadcast Editor: You can switch between a Classic rich text editor and the AI Personalized flow when creating broadcasts, with template selection, scheduling, and preview text
• SMS for Chat Agents: Chat agents can now send SMS messages when assigned a phone number
• AI-Generated Campaign Creatives: You can now auto-generate campaign creatives based on your brand, complete with AI-generated images based on your brand assets

​

Improvements

• Campaigns Tab Redesign: Rebuilt with nested sub-tabs (Voice, Email/Broadcasts, Templates, Audience), with a streamlined interface for managing broadcasts and voice campaigns
• Metrics Page Redesign: New period-over-period trend indicators, conversation volume charts, peak activity heatmap, and assistant performance table with voice/chat filtering

​

Bug Fixes

• Fixed a security-related logging issue
• Fixed errors during import revert not being surfaced properly
• Fixed edge cases in file operations
• Fixed performance issue with large data queries

---

​

v2.4.1 (2026-02-20)

​

Highlights

This release empowers users with the ability to create custom structured outputs, allowing enhanced flexibility for data extraction tailored to specific business needs.

​

New Features

• Custom Structured Outputs: Users can now define their own data extraction formats using a visual field builder. This feature supports various templates like Quote Requests and Appointment Bookings, enabling personalized data collection.

​

v2.4.0 (2026-02-06)

​

Highlights

This release focuses on reliability improvements for lead enrichment and Business DNA extraction, along with several quality-of-life enhancements.

​

Updated Lead Enrichment

What’s New:

• Improved Email Matching: Better handling of email variations and aliases
• Company Domain Resolution: More accurate company identification from email domains
• Fallback Strategies: Multiple matching strategies to find the right contact
• Higher Match Rates: 15% improvement in successful enrichment
• Better Error Messages: Clear feedback when enrichment fails

​

More Reliable Business DNA Extraction

Improvements:

• Robust Logo Detection: Better detection of logos across a wider variety of websites and formats
• Enhanced Color Extraction: More accurate brand color detection from your website
• Improved Content Scraping: Better handling of modern websites, dynamic content, and protected sites
• Smarter AI Synthesis: More consistent and reliable results
• Faster Extraction: 30% reduction in average extraction time

​

Other Improvements

Agent Management:

• Fixed issues when updating voice settings
• Improved cleanup when deleting assistants
• Better error handling across the platform

Analytics:

• New conversation duration distribution chart
• Improved real-time campaign monitoring
• Better export formatting for CSV reports

Integrations:

• Improved CRM sync reliability
• Better handling of rate limits
• Enhanced field mapping options

​

Bug Fixes

• Fixed issue where voice agent setting changes could reset connected tools
• Resolved issue where leads could be missed during high call volume
• Fixed timezone handling in campaign scheduling
• Corrected ICP score calculation for missing enrichment fields
• Fixed chat widget display issues on some websites

---

​

v2.3.0 (2026-01-23)

​

New Features

• Workflow Squads: Coordinate multiple agents with a visual workflow builder
• Campaign Monitoring: Real-time dashboard for outbound campaigns
• Evaluation System: AI-powered quality scoring for conversations

​

Improvements

• Enhanced knowledge base processing speed
• Faster voice response times
• Better handling of long conversations

---

​

v2.2.0 (2026-01-09)

​

New Features

• ICP Scoring: AI-powered lead qualification
• Intent Signals: Automatic detection of buying signals
• Custom Fields: Support for custom lead capture fields

​

Improvements

• Expanded enrichment fields
• Enhanced CRM integration
• Better transcript search

---

​

v2.1.0 (2025-12-26)

​

New Features

• Chat Agents: Website chat widget support
• Lead Enrichment: Automatic lead data enrichment
• CRM Sync: Two-way sync with popular CRMs

​

Improvements

• Improved voice quality
• Better first-message latency
• Enhanced knowledge base queries

---

​

v2.0.0 (2025-12-12)

​

Major Release

Complete platform rebuild with:

• Faster Performance: Rebuilt from the ground up for speed and reliability
• Voice Agents: Full-featured AI voice agent support
• New Dashboard: Redesigned interface with improved navigation
• Scalable Infrastructure: Built to grow with your business

---

​

Version History

​

Getting Help

If you encounter issues after an update:

1. Check this page for known issues
2. Review the documentation
3. Contact support@emanate.ai