Baseer GTM is a 24/7 multi-agent system that finds the right GCC enterprise accounts, researches and qualifies them, drafts personalized outreach across LinkedIn and email, and routes every outbound action through a human approval gate — while a dashboard gives the operator full visibility and control.
We have built a complete, automated sales-and-marketing machine for Baseer — not a prototype. It runs continuously on production infrastructure and covers the full GTM cycle:
| Area | Status | Detail |
|---|---|---|
| Production infrastructure | Live | API (Cloudflare Workers), dashboard (Cloudflare Pages), database (Neon Postgres), agent runtime (this machine, 24/7) |
| Sales pipeline | Running | Apollo sourcing unblocked; journey state machine, approvals, and execution rails all live |
| Qualified accounts | 43 loaded | Real GCC targets incl. Saudi ministries (Justice, Health, HR), ZATCA, GOSI — more flowing in |
| Verified contacts | Migrating | 434 verified contacts + 342 accounts from prior system being ICP-reviewed and imported |
| Marketing engine | Deployed | Signals, market analysis, content calendar, PR — all four dashboards live |
| Quality | 803 tests | Full automated test suite green; every change verified before deploy |
Everything flows through one disciplined spine — the same pattern for every channel, which is what makes the system safe and auditable:
Every step writes to Neon; nothing is implied. Malformed or unsafe outputs are quarantined, not shipped.
Eighteen specialized roles, each with its own instructions, allowed tools, and hard safety constraints. They never talk to each other directly — they collaborate through data and one referee (the Orchestrator).
| Agent | Job | Tools |
|---|---|---|
| Account Sourcing | Discover GCC enterprise accounts that fit the ICP | Apollo · web search · ScrapeGraphAI |
| Contact Discovery | Find reachable decision-makers at sourced accounts | Apollo · LinkedIn (active-only) · Prospeo |
| Enrichment | Complete + verify account/contact data with evidence | Apollo · Prospeo · web research |
| Account Intelligence | Deep account research: digitization signals, tech stack | Apollo · ScrapeGraphAI (Arabic-aware) |
| ICP Strategist | Refine targeting from performance data | Analytics reads |
| Outreach Copywriter | Draft personalized, evidence-grounded outreach | Contract-producer (no sends) |
| SDR Sequence | Plan multi-touch cadences | Sequence planning |
| LinkedIn — Sara | Propose LinkedIn actions from Sara's account (read-only) | LinkedIn search + activity gate |
| LinkedIn — Rawan | Propose LinkedIn actions from Rawan's account (read-only) | LinkedIn search + activity gate |
| Booking & Pipeline | Handle bookings, prep notes, pipeline health | Cal.com reads |
| Safety & Booking | Enforce booking-driven stop rules | Contract-producer |
| Learning & QA | Weekly performance review + learning loop | Analytics reads |
| GTM Orchestrator | The quality gate — reviews every output, decides approve / route / reject / spawn children | Evidence + policy context |
| Agent | Job | Tools |
|---|---|---|
| Social Listener | Scan Twitter/X, community, GitHub for Baseer-relevant signals | TwitterAPI · news RSS · read tools |
| Twitter Engagement (Norah) | Draft replies to relevant tweets (approval-gated, Arabic-aware) | TwitterAPI read → draft only |
| Market Intelligence | Monitor competitors, research papers, tenders, news | News RSS · arXiv · ScrapeGraphAI |
| Content Planner | Weekly content calendar for Sara/Rawan/Norah | Research + persona validators |
| PR Outreach | Draft per-outlet press pitches from the GCC media list | Contract-producer (approval-gated) |
Agents never call each other. A sourcing agent's approved accounts become child tasks the Orchestrator spawns for contact-discovery; approved contacts spawn enrichment; enriched contacts spawn outreach drafts. Depth is capped (≤3), duplicates are deduped, and only whitelisted task types can be spawned — so no runaway loops.
The runtime claims a queued task from Neon, gives the agent its data + tools, and records the typed output back to Neon — where it is schema-validated. A valid output moves the task to awaiting orchestrator review; an invalid one is quarantined. Heartbeats and a stale-task reaper keep the 24/7 loop crash-safe.
When an agent proposes something that sends — a message, reminder, tweet, or pitch — it becomes an approval item you see on the dashboard. You approve, reject, or request an edit. Only your approval, matched exactly to the drafted content, lets the execution layer send it. The dashboard also shows agent health, the task board, quarantine, relay status, and the full contact journey for every prospect.
Every finding, from any source, must clear the ICP rubric before it enters the pipeline.
| Control | What it guarantees |
|---|---|
| Exact-match approval | A send only fires if an approved record matches the drafted message, target, channel, and timing exactly. |
| Global kill-switch | One action halts all outbound instantly; resuming is owner-only. |
| Volume caps + warm-up | LinkedIn ≤100 invites/week per account; email ramps gradually and self-throttles if bounces rise. |
| Suppression & stop rules | Bounces, complaints, replies, and bookings automatically stop further outreach to a contact. |
| Role-based access | Sensitive actions (autonomy level, pause-clear) are owner-only; every change is audit-logged. |
| Read-only agents | Agents propose; they cannot send. Only the policy-gated execution layer talks to providers. |
Baseer now has an end-to-end GTM system that a solo operator can run: it discovers and qualifies the right GCC accounts, builds the outreach, and executes only what you approve — across LinkedIn, email, social, content, and PR — with full visibility and hard safety rails. The sales pipeline is live with real qualified accounts; the marketing engine is deployed; and the data foundation is being enriched with hundreds of previously-verified contacts.