Back to vendors
E

Epiminds

Also known as: Lucy, Epiminds Lucy

Visit site
Entry priceContact sales (pricing not yet public)Full pricing detail

Multi agent AI marketing team fronted by Lucy, orchestrating 20+ specialized agents that report, optimize, pace budgets, and execute across Google, Meta, and TikTok for agencies, with humans in control.

Epiminds is a Stockholm company building a multi-agent AI marketing team for agencies and in-house teams, fronted by Lucy, an AI marketing manager coordinating more than twenty specialized agents across reporting, optimization, budget pacing, bidding, creative analysis, audits, competitive insight and strategic planning. Founded in 2025 by Elias Malm, formerly of Google, and Mo Elkhidir, who led machine learning teams at Spotify and Kry, it emerged from the EWOR incubator and raised a $6.6 million seed led by Lightspeed in October 2025, with agencies managing more than 240 brands on the platform.

The argument is that legacy dashboards are siloed and manual while point AI tools solve narrow problems without coordination. Rather than surfacing insights for someone else to act on, Lucy executes: agents connect directly to client ad accounts on Google, Meta and TikTok, pull live performance data, analyze it, and take action across platforms, adjusting budgets, optimizing ads and launching updates. Each agent knows what the others are doing, and the founders describe the result as a living system rather than a dashboard.

Agencies upload their playbooks, training documents and client context so Lucy adapts to their methods and, over time, mirrors how the team thinks and operates. She runs assigned tasks and workflows on autopilot in the background, monitors accounts proactively and flags risks such as budget pacing problems before they hurt performance, and is reachable conversationally in Slack or by email. Onboarding a new client takes under thirty seconds, and the platform integrates across CRMs, ecommerce, analytics, data warehouses and trend platforms alongside the ad channels.

Epiminds is early. It was founded in 2025, is in closed beta available to selected agencies, and publishes no security page, no data residency options and no developer surface. The product runs on commercial models from OpenAI, Anthropic and Google with proprietary layers specialized for marketing, chosen by Epiminds rather than by the customer. For an agency that wants execution rather than another dashboard, it is a distinctive bet; a buyer needing published security posture, model choice or an API to build against should expect to wait.

Vendor details

Canonical URL

https://epiminds.com

Category

GTM / revenue agent

Subcategory

Agentic marketing automation

Funding status

Six point six million dollar seed led by Lightspeed Venture Partners (Oct 2025), with EWOR, Entourage, and angels; 240+ brands using the platform; Stockholm based

Company status

independent

Use cases & customers

Primary use cases

Multi agent campaign management for agenciesAutomated reporting and budget pacingBid and creative optimization across ad channelsProactive account monitoring and risk flaggingCross channel marketing analysis and execution

Target customers

Digital marketing agenciesIn house marketing and performance teamsTeams running Google, Meta, and TikTok campaigns

Deployment options

SaaS

Integrations

Lucy connects directly to client ad accounts on Google, Meta and TikTok to read live metrics and execute changes, adjusting budgets, optimizing ads and launching updates, and integrates across the wider marketing ecosystem including CRMs, ecommerce, analytics, data warehouses and trend platforms for one view of performance. Agencies reach Lucy conversationally in Slack or by email. No public API, SDK or MCP interface is documented; VoltAgent, the framework Epiminds builds on, is a development dependency rather than a customer surface.

In practice

An agency onboards a new client in under thirty seconds and Lucy's agents immediately begin pulling live campaign data, generating reports, and recommending optimizations across channels

A pacing agent detects a budget overrun risk mid flight and flags it before it hurts performance, while a creative agent proposes new variants

The agency embeds its playbooks and checklists so Lucy's agents execute optimizations that match how the in house team already works, with humans approving strategic calls

Agentic Index coverage score

7.0 / 14 capabilities · 50%

Integrations & Tool Calling Full

Epiminds states Lucy takes action across every platform, adjusting budgets, optimizing ads and launching updates, connecting directly to client ad accounts on Google, Meta and TikTok to both read live metrics and execute changes. The catalog spans advertising channels, CRMs, ecommerce, analytics, data warehouses and trend platforms, delivering one view of performance, and Lucy is reachable in Slack and by email. Changing spend in a live ad account is direct action, not a suggestion handed back.

Sourceepiminds.comread 2026-09-01

Workflow Orchestration Full

Multi-agent execution is the product's whole architecture. Lucy is an AI marketing manager coordinating more than twenty specialized agents across reporting, optimization, budget pacing, bidding, creative analysis, audits, competitive insight and strategic planning, with each agent aware of what the others are doing and the chain running from fetching data through analysis and recommendation to execution. The vendor's own framing is a living system rather than a dashboard, and work runs continuously in the background rather than as one-shot generations.

Sourceepiminds.comread 2026-09-01

Knowledge Grounding & RAG Full

Customers are told to upload playbooks, training docs and client context, and Epiminds states Lucy adapts to the agency's methods and preferences and over time mirrors how the team thinks and operates. That material persists per agency and is read when agents analyze and act, alongside the live campaign data pulled from connected accounts. The vendor and its coverage say agents are trained on this material, but what is documented is upload and adaptation rather than fine-tuning.

Sourceepiminds.comread 2026-09-01

Human Oversight & Guardrails Partial

The boundary is what gets assigned, not a gate on what runs. Lucy runs assigned tasks and workflows on autopilot, managing them continuously in the background, so an agency bounds the agent by choosing what to delegate, and that is a real if coarse control given the actions involved: adjusting budgets, optimizing ads and launching updates in live ad accounts.

No approval step before a spend change, no review queue, no autonomy setting and no per-action permission scope is documented by the vendor. Press coverage quotes an agency saying agents execute approved actions while humans keep strategic control, but that describes the agency's own working practice, and the vendor's own homepage says autopilot.

Sourceepiminds.comread 2026-09-01

Security, Identity & Governance Not documented

No security posture is published. No security, trust or compliance page exists on the first-party site, and no attestation, certification, single sign-on, role-based access control, audit logging, encryption statement or sub-processor list is documented.

The privacy policy carries a no-training commitment, GDPR compliance and a reference to appropriate organizational and technical measures, which is language rather than security controls.

The company was founded in 2025, raised its seed in October 2025 and is in closed beta available to selected agencies, so an unpublished posture is expected at this stage.

Sourceepiminds.com and its privacy policyread 2026-09-01

Observability & Auditability Not documented

The agency gets marketing reporting, not a record of what the agents did. Agents pull live performance metrics, surface insights and flag risks, which describes campaign outcomes rather than agent behavior, and no run history, activity log, action trail or failure record is offered to the agency; the homepage's no black box is a claim rather than a feature. VoltOps, the console of the VoltAgent framework Epiminds builds on, traces agents for Epiminds' own engineers and is not offered to customers.

Sourceepiminds.com and voltagent.dev case studyread 2026-09-01

Memory & State Persistence Partial

Continuity by necessity rather than by design. The agents run continuously against the same accounts, so campaign history and pacing state carry between runs. No memory feature is named, scoped or given a lifetime. The uploaded playbooks, training docs and client context that persist per agency are grounding material rather than agent memory, and the claim that Lucy learns each agency's methods and improves with every campaign is model adaptation rather than a store an agent writes and reads.

Sourceepiminds.comread 2026-09-01

Deployment & Data Residency Not documented

Cloud only, with nothing the customer selects. No region choice, data center disclosure, VPC, private tenancy, self-hosted or on-premises option is published. GDPR compliance for EU and UK users is legal cover rather than residency, and Stripe as payment processor is not a deployment property. Onboarding a client in under thirty seconds describes provisioning speed, not topology. Where agency and client data is hosted is not stated.

Sourceepiminds.comread 2026-09-01

Prebuilt Agents, Templates & Packs Full

A named set of prebuilt agents an agency receives ready-made. Lucy fronts more than twenty specialized agents covering reporting, optimization, budget pacing, bidding, creative analysis, audits, competitive insight and strategic planning, and an agency onboards a client in under thirty seconds and has the team working immediately rather than assembling it. The agency's own playbooks and templates shape how those agents behave, but the agents themselves arrive prebuilt.

Sourceepiminds.comread 2026-09-01

Triggers & Channel Coverage Full

Work comes to Lucy and her agents by three routes. In the background, Lucy runs assigned tasks and workflows on autopilot and manages them continuously, monitoring accounts across every channel without being asked. Signals in the data start work too, since agents proactively flag risks such as budget pacing problems before they hurt performance. And in conversation, an agency chats with Lucy in Slack or emails her with questions and tasks, so she works where the team works.

Sourceepiminds.comread 2026-09-01

Model Flexibility & Routing Partial

Vendor-internal multi-provider use, not customer choice. The company states its product is built on commercial large language models from OpenAI, Anthropic and Google with proprietary layers specializing them for marketing. That is three providers behind one product, selected and combined by Epiminds. Nothing documents an agency picking a model, bringing its own key, or routing work to a chosen provider. The company also plans an in-house marketing-specialized model, which has not shipped.

Sourceepiminds.com and Forbes coverageread 2026-09-01

APIs, SDKs & MCP Extensibility Not documented

Nothing callable from outside is documented. Epiminds builds on VoltAgent, which lets its own engineers prototype and deploy new agents quickly but gives an agency no surface to build against. The first-party site has no developer, documentation, API or status entry, and there is no docs subdomain. The integrations that do exist point the other way, with Epiminds reaching into ad platforms and the marketing stack.

Sourceepiminds.com and voltagent.dev case studyread 2026-09-01

Testing, Debugging & Optimization Partial

Epiminds states that over time Lucy mirrors how the agency's team thinks and operates, an asserted improvement loop on the agent's own work. Nothing says what feeds it, what it tunes or how a change is measured, and VoltOps traces are the vendor's own engineering console.

Sourceepiminds.comread 2026-10-01

Browser & Computer Use Not documented

No browser or computer use capability documented; agents act via direct integrations with ad platforms and the marketing stack.

Sourceepiminds.com and coverage pages retrievedread 2026-07-08

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Contact sales (pricing not yet public)

What is public

That pricing is processed via Stripe and not yet public is documented; figures are not.

Billing mechanics

Not publicly documented; payments processed through Stripe. Sold to agencies and in house teams with rapid client onboarding.

Cost watchouts

Cost likely scales with the number of clients or brands managed and ad spend or data volume processed; the platform processes more than sixty million marketing data points daily.

Variable cost rationale

Cost plausibly scales with the number of client accounts and brands managed and the volume of ad data processed, but pricing is not yet public so exposure is inferred.

Additional watchouts

Early stage; pricing not yet public, so ask how cost scales with the number of clients or brands managed.

Sales call required

Yes, required for paid access

Free / trial

Closed beta on some surfaces; no public free tier documented

Key ambiguities

Pricing is not published and the billing unit is not stated.

Missing data

All pricing figures, billing axis, trial terms.

Agentic Index verified 2026-10-01

Alternatives to Epiminds

The closest documented capability profiles to Epiminds among GTM and revenue agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • TextYess6.5 / 14Adds documented Observability & Auditability
  • FlashLabs5.0 / 14A lighter documented profile than Epiminds
  • Korl6.0 / 14Adds documented Testing, Debugging & Optimization
  • Swiftly6.0 / 14Adds documented Testing, Debugging & Optimization
  • BambooBox5.5 / 14Adds documented Security, Identity & Governance
  • Kana6.5 / 14Adds documented Testing, Debugging & OptimizationEpiminds vs Kana →

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Head to head

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.