For every promise we make, here are the features that actually ship today, mapped to the value they unlock. Five capability areas, CRM in beta. One personal AI per SDR. One workspace.
Per-prospect research and email in your SDR's voice. No templates. Dual-channel email + LinkedIn.
02 · RepliesIntent classification, auto-response on routine, escalation in your voice. Market intelligence by default.
03 · LinkedInAI-drafted posts and replies, sent manually in one click. No automation risk to a personal account.
04 · Personal AIA personal AI per SDR. Produces on day one, sharpens with every edit, travels with them when they leave.
05 · CRM BetaA positive reply becomes a deal automatically. Native accounts and deals, inbound lead capture with optional auto-engage, and one-way or two-way sync to your existing CRM (HubSpot or Microsoft Dynamics 365).
The bulk of an SDR's day is research and drafting. Haiman does both, on every single prospect, in your SDR's voice. No templates. No merge fields.
AI-driven target-company sourcing from your structured ICP criteria: company names, websites and relevance signals across industries, geographies and topics.
A per-company brief from 12–18 web searches: overview, decision-makers, recent news, pain points, outreach angles and a 0–100 fit score. Cached in the lead view.
Automated email-address discovery and LinkedIn lookup, each result verified and tagged high / medium / low confidence. Bulk via CSV import.
Every lead scored against your ICP. Sub-threshold accounts never reach the queue, so the team works the highest-fit prospects, not the longest list.
Every email written fresh from real-time research and the SDR's own voice. A different message for every prospect, every time.
Six templates per account, email-only and email + LinkedIn. Working-hours-aware, send-capped, reply-aware, with auto-pause if mailbox health degrades.
Automated sector monitoring: ICP-relevant companies and contacts surface into the pipeline as news breaks.
Time-bound campaigns with custom sequences, CSV import and per-campaign reporting, for events, launches or one-off pushes.
The two biggest SDR bottlenecks are bad leads in and manual everything. Both are solved at intake: ICP fit scored on every prospect before a draft is written, and per-prospect research replacing six tabs and a coffee. The team spends its time building pipeline, not triaging lists.
An inbound lead that isn't contacted in minutes is mostly gone. Haiman classifies every reply by intent, drafts routine responses automatically, escalates the rest, and surfaces the intelligence buried in the inbox.
Every inbound reply tagged on arrival: interested, objection, not now, unsubscribe, out of office, meeting request. Routing follows automatically.
Routine intents drafted automatically in your SDR's voice; ambiguous or high-stakes replies escalated with full thread context.
Gmail and Outlook replies stitch to the same conversation, so a lead who answers from another address is still recognised. No orphaned reply, no double-outreach.
Interested replies can include the SDR's calendar link automatically, with handoff back to their calendar for confirmation.
A dashboard button forces an out-of-band poll instead of waiting for the scheduler, useful when watching a hot lead.
Objections, competitor mentions and pricing pushback tagged at classification. A weekly digest surfaces what's actually being said in-market, with no extra work for the SDR.
Slow lead response time is one of the top bottlenecks in B2B sales: the probability of booking drops sharply if an inbound isn't contacted within minutes. Haiman closes that window, and turns a side effect of reply handling into a structured feedback loop back to product and marketing.
Posts and replies in your SDR's voice, threaded through the same approval queue as outbound, on the same voice profile. DMs are drafted by Haiman, sent by the SDR, on purpose.
Each SDR connects their personal account; posts and replies publish as the user, from their own identity. No third-party-sent watermarks.
A one-line intent (“share something on the new sustainability report”) becomes a draft post in your SDR's voice. Queue, edit, post, done.
Paste a comment or post URL plus the text; Haiman threads it, drafts a reply in their voice and queues it. LinkedIn stops being a thing they “should do”.
Every social action (post, threaded reply, sales handoff) flows through the same approval surface as outbound. One place to review, one voice.
Social posts and replies use the same voice profile as outbound email, so brand, outbound and LinkedIn presence all sound like the same person.
When a commenter signals sales interest, route them to the SDR queue for research and outreach in one click. The post becomes the lead becomes the meeting.
Haiman drafts every DM in your SDR's voice; the SDR sends manually. Five seconds per send: a working channel instead of a banned account.
Tools that “AI-send LinkedIn DMs at scale” use cookie-session automation that violates LinkedIn's terms of use, so accounts get warned, restricted or banned. Draft-and-present gives SDRs personalised, voiced DMs without risking the personal account they take to their next job. Same voice across email and LinkedIn means brand and outbound finally stop sounding like different companies.
A personal AI for each of your SDRs, learned from how they actually write. Starts producing on day one, sharpens every week, travels with them when they leave.
With permission, Haiman reads up to your SDR's last 500 sent emails (personal info removed before any model sees them) as live reference. Drafts sound like the SDR from message one.
Once a week, a structured profile is extracted: greeting patterns, sign-offs, average length, phrases favoured and avoided. Injected into every draft from then on.
After ~4–8 weeks, a model personal to your SDR trains on the gap between drafts and sends. Closer to their voice on the first try, with measurably less editing.
UK or US spelling set per user, applied across every draft, so cross-Atlantic teams don't fight over “colour” every Tuesday.
The profile follows the SDR, not the company seat. When they move on, their voice doesn't get reset onto the new hire.
Writing samples, style profile, edit history and the trained model export in standard formats, no Haiman dependency. Lead lists and sent-email rows stay with the company.
An SDR's writing trains only their own personal AI, never shared with other tenants, never used to train any base model.
Ramp is the most expensive thing about an SDR team: three months in, twelve months out. The personal AI compresses ramp, because drafts sound like the team within weeks; and it solves the turnover tax, because the next hire is effective in the prior hire's voice from day one, while the leaver takes their own profile with them. Both sides win, and you don't start from zero.
The meeting you book shouldn't fall out of the system the moment it's won. A positive reply becomes a deal automatically, on the right account, in the same workspace as outbound and social. Leads from your own website and backend can post straight in too, through a simple API, and your AI can answer the moment one arrives, in your own voice. In beta, open to early-access teams.
The same classifier that tags a reply interested creates the deal, on the right account, the moment it lands. No copy-paste, no booked meeting left in an inbox.
Push leads straight from your own website forms, landing pages or backend. One authenticated endpoint turns a submission into a contact, account and open deal. Per-company API keys, idempotent retries and a dry-run mode make it safe to wire up.
Switch it on and an inbound lead doesn't just become a deal: your AI sends the first-touch and starts a follow-up sequence the moment it lands, in your own voice, from your own mailbox. Leads are shared round-robin across the team, and it stays off until you turn it on.
Accounts, deals and contacts live in the workspace, company-scoped, on the same multi-tenant isolation as everything else in Haiman.
Move deals through stages on a board. Every stage change is recorded, so the pipeline carries history, not just its current state.
Connectors for HubSpot and Microsoft Dynamics 365 sync accounts and deals into your system of record, one-way or full two-way: make your CRM the source of truth and Haiman follows it.
Prefer your own plumbing? Every CRM event fires a signed (HMAC) outbound webhook, and deals export on demand as CSV. No connector required.
CRM shares the identity and approval queue with outbound and social. The deal and the email that created it are one record, not two systems to reconcile.
An SDR books a meeting, and then it falls out of the system: a sticky note, a spreadsheet, a CRM nobody updates. Haiman closes that gap at the source. The reply that reads as interested becomes the deal, automatically, the moment it happens. Leads can arrive the other way too: your own forms and backend post directly into the CRM through the inbound API, so a deal exists the moment someone raises their hand, wherever they raised it. Turn on auto-engage and that raised hand gets a first-touch in your own voice within minutes, instead of waiting for someone to notice the queue. The native store is the default home; if your team already lives in HubSpot or Microsoft Dynamics 365, the same events sync there, one-way or as full two-way sync with your CRM as the source of truth. Either way, the pipeline you earned doesn't get lost between tools.
The architecture assumes EU data residency rather than retrofitting it. The OAuth posture is verified by Google after independent assessment; the personal AI is portable by contract, not by promise.
OAuth client and restricted Gmail scopes approved after independent assessment against OWASP ASVS L2, the bar most consumer AI tools never clear.
Outlook OAuth approved across tenants; your team connects Outlook mailboxes without per-app review from IT.
Application and primary database in Ireland; personal-AI storage in Frankfurt; training and serving in Finland. Business data stays in the EU.
EU data residency from day one. Right-to-erasure at the substrate level: one-click hard-delete cascades through writing samples, style profile and personal model.
TLS in transit, AES-256 at rest. OAuth mailbox access for Gmail and Outlook, no passwords stored. Strict CSP, admin MFA and strong sessions throughout.
Every query scoped by company and agent. Cross-tenant exposure is prevented at the API layer, not by convention.
Connect Gmail or Outlook, set your ICP, and your SDRs are shipping outreach inside the first afternoon. Monthly billing, no annual lock-in.
See pricing & startYour ICP, your stack, your questions. A procurement and security review pack is available on request.
carl@haiman.ai