How much does Tofu cost?

Custom (mid-market/enterprise; sales-led)

What are the best alternatives to Tofu?

Based on documented capability coverage across the Agentic Index 14 point taxonomy, the closest alternatives to Tofu among GTM and revenue agents are Alta, FlashLabs, Warmly, Aomni. Each is graded on the same public evidence, and no vendor pays for placement.

Does Tofu support MCP and tool calling?

Partial under the Agentic Index verification standard. Tofu publishes some public evidence of tool calling and API or SDK extensibility, but narrower than full Model Context Protocol (MCP) support as defined by the Agentic Index taxonomy. Evidence basis: No public API documentation, no SDK, no developer portal, no docs subdomain and no MCP surface was found across four passes, including two that reached tofuhq.com product pages directly. PARTIAL RESTS ENTIRELY ON CUSTOM DATA CONNECTORS, whi. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Tofu have SOC 2, SSO and role based access control?

Partial under the Agentic Index verification standard. Tofu publishes some public evidence of security and identity controls, but narrower than the full set of SOC 2, single sign on (SSO) and role based access control (RBAC). Evidence basis: THE WEAKEST VERIFIABLE SECURITY POSTURE FOUND IN THE ENTIRE REGRADE, and worse than kommunicate, which at least asserted its certifications cleanly. Tofu publishes a privacy policy at tofuhq.com/privacy-policy. No trust centre, no security. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Tofu give you tracing and observability into agent runs?

Partial under the Agentic Index verification standard. Tofu publishes some public evidence of logging or observability, but narrower than full tracing and auditability across agent runs. Evidence basis: DOWNGRADED FROM THE UNSOURCED GRID, which carried Full. No audit log, agent action log, activity trail or reporting surface is documented anywhere on Tofu's own product pages. PARTIAL RESTS ON THE CHAMPIFY PRECEDENT RATHER THAN ON A TOFU FE. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

Tofu

Q: Who is Tofu for?

marketing teams, mid-market, enterprise

Visit site

GTM / revenue agent Verified 2026-06-23

Entry priceContact salesFull pricing detail

Agentic demand-gen and ABM with research, content, and launch agents grounded in a brand knowledge graph. Customers include Perplexity and Cursor.

Tofu runs agentic demand-gen and ABM through three named agents (Research, Content, and Launch) grounded in a brand knowledge graph (its Playbook), with the Launch Agent firing campaigns and updating the CRM. It integrates natively with HubSpot, Salesforce, Marketo, Outreach, Gong, and Salesloft; customers include Perplexity, Cursor, and RingCentral, sold on custom mid-market/enterprise pricing.

Vendor details

Canonical URL

https://tofuhq.com

Subcategory

Marketing — agentic demand gen / ABM

Use cases & customers

Primary use cases

demand generationABMcampaign launch

Target customers

marketing teamsmid-marketenterprise

Deployment options

SaaS

In practice

Your campaigns sound off-brand because the AI doesn't know your positioning. Tofu grounds its agents in a brand knowledge graph, so the content reflects your Playbook.

Research, content, and launching a campaign are three separate jobs. Tofu runs them as three agents, Research, Content, and Launch, so the campaign moves from insight to live.

Content that's written still has to get launched and logged. Tofu's Launch Agent fires the campaign and updates the CRM, closing the loop into Salesforce or HubSpot.

Agentic Index coverage score

7.5 / 14 capabilities · 54%

Integrations & Tool CallingBreadth across four distinct ecosystem classes rather than depth in one, which is what this axis rewards. Native integrations span CRM (Salesforce, HubSpot), marketing automation (Marketo, HubSpot), sales engagement (Outreach, Salesloft, Gong Engage), social and advertising (LinkedIn) and content management (the platform page states any CMS), plus custom data connectors. THE ARCHITECTURAL POINT MATTERS MORE THAN THE LIST: Tofu positions itself as running INSIDE the customer's existing martech stack rather than replacing it, pushing generated assets into the tools the team already operates, so integration is the delivery mechanism rather than a convenience feature. Same shape as regie-ai in this pass. Tofu platform page and product documentation 2026-07-30	Full
Workflow OrchestrationThree named agents that research accounts, create personalized content and launch campaigns. THE LAUNCH AGENT IS THE EVIDENCE THAT EARNS FULL: triggers are configured once, after which it sequences emails, stages leads, fires campaigns and updates the CRM without a human in the chain. That is unattended multi step execution ending in a sent campaign, which is materially more than autobound (does not send on the user's behalf) or champify (terminates at a human), both graded Partial in this pass. NOTABLE TENSION, RECORDED NOT PENALISED: Tofu's own category essay states that some platforms, INCLUDING TOFU, function as AI powered automation systems rather than true agents. That is the vendor UNDERSTATING itself in thought leadership while the product page documents autonomous execution, the inverse of the warmly trap. Graded on the product page. Tofu platform page and agentic marketing essay 2026-07-30	Full
Knowledge Grounding & RAGTHE PLAYBOOK IS A REAL STRUCTURED GROUNDING LAYER and the strongest part of the product. It ingests brand voice, ideal customer profile, personas, messaging frameworks and existing assets, and every generated asset is constrained by it, which is what makes per account output stay on brand at volume. On top of that a research step pulls per account context before writing, and follow ups are grounded in past objections, prior interactions and recorded lost reasons drawn from the connected CRM. Structurally this is the same shape as Jasper IQ, which earned Full in this pass: a persistent, customer owned brand and context layer applied as a constraint at generation time rather than a prompt library. Tofu platform page and product documentation 2026-07-30	Full
Human Oversight & GuardrailsTwo controls claimed, both on a first party surface but a weak one. Tofu states that ROLE BASED ACCESS CONTROLS AND CONTENT APPROVAL WORKFLOWS are built in for governance, and separately that customers run 100 plus marketing users collaborating in the system, which is the scale at which approval routing becomes necessary. The Playbook also functions as a pre generation constraint, since brand voice and messaging frameworks bound what the agent can produce before a human sees it, the same mechanism that earned Jasper a Full. WHAT KEEPS THIS AT PARTIAL: the approval workflow claim appears only on a self authored competitive comparison page, not in product documentation, and no per agent risk calibration, escalation logic or pre execution policy engine was found. Tofu first party comparison pages and platform page 2026-07-30	Partial
Security, Identity & GovernanceTHE WEAKEST VERIFIABLE SECURITY POSTURE FOUND IN THE ENTIRE REGRADE, and worse than kommunicate, which at least asserted its certifications cleanly. Tofu publishes a privacy policy at tofuhq.com/privacy-policy. No trust centre, no security page, no certification page, no subprocessor list and no downloadable report was found across four passes. THE ONLY SECURITY STATEMENT LOCATED IS ON A SELF AUTHORED COMPETITIVE COMPARISON POST AND IT CANNOT STATE ITS OWN CERTIFICATION STATUS, saying Tofu EITHER HAS OR IS IN THE PROCESS OF formal security attestations and that it has been able to satisfy InfoSec reviews. A vendor unable to say on its own domain whether it holds SOC 2 has made a non claim, and under the standing rule an unspecified or hedged attestation is treated as unverified. Partial rather than None rests on three real things: role based access controls, enterprise data encryption and GDPR compliance are each claimed on that same first party page, and named customers including publicly traded companies imply completed InfoSec reviews. CRITICAL COLLISION WARNING FOR ANY FUTURE EDITOR: trust.gotofu.com publishes ISO 27001:2022, PCI DSS, GDPR and AWS Frankfurt hosting, and gotofu.com publishes a security page citing Google Cloud certifications. THAT IS TOFU2 LIMITED OF HONG KONG, A DIFFERENT COMPANY. Do not credit any of it here. Tofu privacy policy and first party comparison pages 2026-07-30	Partial
Observability & AuditabilityDOWNGRADED FROM THE UNSOURCED GRID, which carried Full. No audit log, agent action log, activity trail or reporting surface is documented anywhere on Tofu's own product pages. PARTIAL RESTS ON THE CHAMPIFY PRECEDENT RATHER THAN ON A TOFU FEATURE: because the Launch Agent executes by writing into the customer's own CRM, marketing automation platform and sales engagement tools, the record of what it actually did lands in those systems and inherits their native audit trails and reporting. Content approval workflows produce a further record of who approved what. THAT IS INHERITED OBSERVABILITY, NOT SHIPPED OBSERVABILITY, and it means a buyer cannot answer what the agent did from inside Tofu. This is the thinnest Partial in the build and is FLAGGED FOR RE VERIFICATION. Tofu platform page and first party comparison pages 2026-07-30	Partial
Memory & State PersistenceDOWNGRADED FROM THE UNSOURCED GRID, which carried Full on no evidence, and downgraded for the same reason as demandbase in this pass. What exists is a PERSISTENT CONTEXT LAYER, not agent memory: the Playbook holds brand voice, ICP and personas durably across campaigns, and follow up generation draws on past objections, interactions and lost reasons. But that history lives in the connected CRM and the Playbook is configuration, so neither is state the agent maintains and reasons over. No cross session agent memory, no per agent state persistence and no retention or forgetting control at the agent level appeared on any first party surface across four passes. Separating a persistent knowledge layer from agent memory is the whole basis of this grade. Tofu platform page and product documentation 2026-07-30	Partial
Deployment & Data ResidencyNo documented deployment or residency control. The single relevant statement is a passing mention of OPTIONS FOR DATA RESIDENCY on a self authored competitive comparison post, with no regions named, no mechanism described, no hosting provider identified and no corroboration anywhere else. Under the standing rule that hedged first party marketing claims without an artefact are unverified, that does not support even a Partial. No on premises, air gapped, single tenant or private VPC option was found, and Tofu is a 36 person US company at Series A, so an enterprise residency programme would be a surprising thing to leave undocumented. CRITICAL COLLISION WARNING: the AWS Frankfurt hosting and the Google Cloud Platform hosting statements that surface on a Tofu search BOTH BELONG TO GOTOFU.COM, Tofu2 Limited of Hong Kong, a different company, and one of them also contradicts the other. Neither describes this vendor. FLAGGED FOR RE VERIFICATION: if Tofu publishes a security or trust page naming regions, this moves to Partial. Tofu first party comparison pages 2026-07-30	Unable to verify
Prebuilt Agents, Templates & PacksThree prebuilt agents ship as the product rather than as a configurable framework: a research agent, a content agent and the Launch Agent. On top of that the PLAYBOOK IS THE TEMPLATE SYSTEM, a reusable structure holding brand voice, ICP, personas and messaging frameworks that instantiates entire campaigns from a single brief. Documented campaign types function as plays: lead nurture, event follow up, expansion, outbound, sales acceleration and re engagement. Output format coverage is unusually broad for this cluster, spanning emails, landing pages, microsites, ads, whitepapers, case studies, one pagers, decks and sales talk tracks. A single brief producing a full multi format campaign set per account is exactly what this axis measures. Tofu platform page and product documentation 2026-07-30	Full
Triggers & Channel CoverageChannel coverage is genuinely wide: email, landing pages, microsites, ads, social, whitepapers, case studies, one pagers, decks and sales talk tracks, with content tailored by industry, persona, account, signal and behaviour. Triggers are set once and fire the Launch Agent, and documented campaign types span lead nurture, event follow up, expansion, outbound, sales acceleration and re engagement. THE DEPENDENCY IS WORTH STATING PLAINLY AND TOFU STATES IT ITSELF: there is no native intent data and no account identification, so Tofu does not originate signals, it consumes them from the connected CRM, marketing automation platform or a paired intent vendor. The axis measures ability to trigger and to reach channels, both of which are present, so this is Full with the upstream dependency recorded. Tofu platform page and first party comparison pages 2026-07-30	Full
Model Flexibility & RoutingNo model choice, routing, bring your own model or bring your own key capability documented. Tofu describes itself as AI native and LLM native but names no model provider anywhere on its product pages, discloses no routing logic and offers the customer no selection. Four passes reached the platform page, the privacy policy and roughly a dozen first party posts without surfacing a single named model. THE DISTINCTION THAT MATTERS FOR THE BUYER, per the standing note on this axis: this is SINGLE PROVIDER DEPENDENCY WITH THE PROVIDER UNDISCLOSED, which is weaker than gorgias (dependency, provider known) and categorically different from omilia, pokee-ai or leena-ai, where a 0.0 reflects deliberate vertical integration that enables a certified or air gapped perimeter. Same score, opposite meaning, and for a content generation product the undisclosed provider is itself a procurement question. Tofu platform page and product documentation 2026-07-30	Unable to verify
APIs, SDKs & MCP ExtensibilityNo public API documentation, no SDK, no developer portal, no docs subdomain and no MCP surface was found across four passes, including two that reached tofuhq.com product pages directly. PARTIAL RESTS ENTIRELY ON CUSTOM DATA CONNECTORS, which Tofu lists alongside its native integrations, plus bidirectional sync with CRM and marketing automation that implies API access exists even though it is not published. THE HONEST READING: Tofu is architected to run inside other people's platforms rather than to be a platform others build on, which is a coherent product choice and the opposite of the demandbase posture graded in the previous build. THIS IS THE SINGLE MOST LIKELY CELL IN THIS GRID TO MOVE ON RE VERIFICATION, because Ext is the axis this pass has most often understated on first contact, and it is flagged accordingly. Tofu platform page and product documentation 2026-07-30	Partial
Testing, Debugging & OptimizationDOWNGRADED FROM THE UNSOURCED GRID, which carried Partial. No sandbox, evaluation harness, experimentation engine, A/B testing framework, regression suite or output scoring surface is documented on any Tofu product page. This differs from the Partial grades given to optimizely and demandbase in this pass, both of which at least ship a measurement product pointed at the outcomes their agents produce; Tofu ships none. THE ARCHITECTURAL REASON IS THE INTERESTING PART: because Tofu executes by pushing assets into the customer's CRM, marketing automation platform and sales engagement tools, campaign measurement necessarily happens in those systems rather than in Tofu. Running inside someone else's stack pushes evaluation out of the product along with observability, which is a structural consequence of the delivery model rather than neglect. Tofu platform page and first party comparison pages 2026-07-30	Unable to verify
Browser & Computer UseNo browser control, computer use, screen operation or robotic process automation capability of any kind documented. Tofu acts entirely through native integrations with CRM, marketing automation, sales engagement and content management systems, which is the correct architecture for a content generation and campaign execution product and not a gap in any practical sense. Four passes across the platform page, first party posts and third party profiles surfaced nothing in this class. Consistent with every other account based marketing vendor graded in this pass. Tofu platform page and product documentation 2026-07-30	Unable to verify

The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify 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

flat

What is public

Tofu (tofuhq.com - AI marketing platform: generates/personalizes/repurposes marketing content across channels (campaigns, ABM, web/email/ads); unifies brand + product data to produce on-brand assets) is custom/quote-based - sales-led, no public price tiers (mid-market/enterprise).

Billing mechanics

Custom subscription (sales-led, quote-based; scoped by usage/seats/content volume); not publicly itemized. No public figures.

Additional watchouts

Custom pricing only (no public figures); sales-led; value tied to content/personalization volume

Sales call required

Yes, required for paid access

Commercial notes

AI marketing-content / personalization platform; competes with Jasper, Writer, Copy.ai, Regie.ai, Mutiny (ABM personalization)

Support SLA / resale

Integrates with CRM/marketing stack (e.g., Salesforce, HubSpot, 6sense); content generation + personalization + ABM

Agentic Index verified 2026-06-25

Official pricing page

Related vendors

Alternatives to Tofu

The closest documented capability profiles to Tofu 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.

Alta8.5 / 14Adds documented Deployment & Data Residency
FlashLabs6.5 / 14A lighter documented profile than Tofu
Warmly8.5 / 14Adds documented Testing, Debugging & Optimization
Aomni6.0 / 14A lighter documented profile than Tofu
Artisan7.0 / 14Adds documented Testing, Debugging & Optimization
AutoRFP.ai7.0 / 14Adds documented Testing, Debugging & Optimization

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

Compare Tofu with other vendors →Compare all GTM and revenue agents →See where Tofu ranks →

Data confidence: high

Head to head

Tofu vs Demandbase Tofu vs Mutiny Tofu vs Jasper