Table of Contents

OpenTag Investment Report

OpenTag Investment Report

Category: AI workplace agent; team productivity and workflow automation

Company Stage: Pre-seed / accelerator stage; founded in 2026

Founder or Founders: Tony Kam, Shelden Shi, Wilson Nguyen

Headquarters: San Francisco, California, United States

Funding: Y Combinator S26; YC’s standard investment is $500,000. No other funding was publicly verified

Business Model: Usage-based B2B SaaS, with self-serve and enterprise plans

Product Hunt Launch Date: August 28, 2026

Report Date: August 31, 2026

Investment Metric	Assessment
Venture Potential	65/100
Unicorn Path	Conditional
Valuation Attractiveness	Not Assessable
Evidence Confidence	57/100
Final Decision	Watch

Executive Summary

OpenTag is an AI coworker embedded in Slack and Microsoft Teams. Users mention it in a channel, delegate work, and receive results in the same thread. The product is designed to acquire company context from approved channels, maintain an internal wiki, identify repetitive workflows, and execute work across connected applications with approval gates for consequential actions (official website; product documentation).

The initial target appears to be small and mid-sized knowledge-work teams that already operate primarily through Slack or Teams but do not want to configure individual automations or pay for AI on a per-seat basis. Pricing starts at $50 per workspace per month for 5,000 credits, with larger self-service plans reaching $1,000 per month and custom enterprise pricing (official pricing; plan documentation).

The strongest positive signal is the combination of a credible technical founding team, repeat Y Combinator experience, working product infrastructure, transparent pricing and fast access to a large collaboration-platform market. The founders report that early access began approximately two weeks before launch and that OpenTag was live with ten teams at the time of its YC launch. This is useful evidence of deployment but remains a company-reported, very small sample with no disclosed paid conversion or retention (Y Combinator profile).

The principal concern is intense and unusually direct competition. Anthropic’s Claude Tag provides substantially the same core interface—mentioning an AI teammate in Slack, granting it selected channel and tool access, delegating asynchronous work and building persistent context (Anthropic). Slack, Microsoft, Glean and Lindy also offer overlapping search, context and agent capabilities. OpenTag’s model routing and usage-based workspace pricing are differentiators, but neither is yet a demonstrated moat.

The final decision is Watch. OpenTag is credible enough to monitor, but ten company-reported deployments, unknown revenue and retention, incomplete security certification and unproven differentiation do not yet justify formal diligence. A move to DD would require evidence of paid adoption, repeat usage, retention, healthy gross margins and a defendable advantage beyond model selection.

Product Overview

OpenTag attempts to replace fragmented knowledge retrieval and manually configured workflow automation. Rather than asking employees to open a separate assistant, it sits inside team chat. A user can mention OpenTag, request a report, lookup, audit, draft or multi-application workflow, and receive the result in a shared thread.

The product’s core elements are:

Slack and Microsoft Teams interfaces.
Access limited to channels into which the agent is invited.
A continuously maintained company wiki.
Proactive suggestions for repetitive tasks.
Scheduled routines and multi-tool workflows.
Model routing through Conifer, with access to more than 80 models.
Human approval before external actions such as sending, submitting, purchasing or spending.
Per-person permissions and action logs (official website; security page).

OpenTag says each job operates in a sandboxed cloud machine and that the environment is removed after completion. Customer context and credentials otherwise reside in a dedicated customer or team environment managed by the company (product documentation; privacy policy).

Pricing is usage-based rather than per seat. The trial includes 10,000 non-expiring credits; Team starts at $50 per month for 5,000 monthly credits and scales to $1,000 per month for 100,000 credits. Annual billing receives a stated 15% discount. Enterprise adds volume pricing, SSO, SCIM, audit-log export, support and custom retention terms (plans).

According to OpenTag, simple tasks consume approximately 25–75 credits, recurring jobs 125–375 and larger projects 500–1,250. At $0.01 per credit on the monthly plans, this equates to approximately $0.25–$0.75, $1.25–$3.75 and $5–$12.50 of customer spend respectively—not the company’s underlying cost (credit documentation).

Product Quality Assessment: Thoughtful workflow positioning, usable pricing and sensible approval controls, but independent evidence of reliability and work quality is insufficient.

Founder and Team Assessment

Y Combinator identifies Tony Kam, Shelden Shi and Wilson Nguyen as OpenTag’s founders. The legal operator is Open Curiosity, Inc., according to the product’s privacy policy.

Kam and Shi previously co-founded Lilac Labs, a YC S24 voice-AI system for drive-through restaurants. Kam’s public profile lists software engineering experience at Tesla and an EECS degree from UC Berkeley; Shi’s lists engineering experience at Flatiron Health, Roche and CertiK. Nguyen was a founding engineer at Lilac Labs and previously worked at Vestible and Hewlett Packard Enterprise (Kam; Shi; Nguyen; Lilac Labs).

The three founders previously worked together, reducing initial team-formation risk. Their backgrounds support technical execution, but there is no verified prior exit. The transition from Lilac Labs to a different product category also requires diligence: public evidence does not explain the prior company’s commercial outcome, asset ownership or whether any obligations carried into OpenTag.

The company’s LinkedIn page lists three employees, consistent with the three disclosed founders, although LinkedIn headcount is not authoritative. Full-time commitment appears likely from current profiles but is not fully verified for every founder.

Founder Assessment: Strong early technical cohesion and prior startup experience, but enterprise sales execution and the outcome of the previous venture remain unproven.

Market Opportunity

The narrow initial segment is Slack-centric startups and SMB knowledge-work teams that need cross-application research and workflow execution but lack dedicated automation staff.

Slack reports more than 200,000 paying customers and usage among 77 of the Fortune 100 (Slack company data). Applying OpenTag’s published pricing produces the following illustrative—not forecast—market scenarios:

10,000 workspaces at $1,200 annual revenue: $12 million ARR
20,000 workspaces at $5,000 annual revenue: $100 million ARR
5,000 enterprise customers at $20,000 annual revenue: $100 million ARR

The $1,200 and $5,000 assumptions fall within the current $600–$12,000 annual self-service range. The $20,000 enterprise figure is an analyst assumption; OpenTag does not disclose enterprise pricing.

Adjacent expansion opportunities include Microsoft Teams, function-specific agents for marketing, support and operations, additional integrations, regulated-industry deployments and an API or agent platform. International expansion is technically plausible, but current infrastructure is operated in the United States, which may constrain buyers with localization requirements (privacy policy).

The market can support venture-scale revenue. The issue is not market size but whether an independent agent can retain distribution and pricing power against Slack, Microsoft and model vendors.

Traction and Growth Signals

OpenTag’s founders reported ten teams in production after approximately two weeks of early access (YC launch page). Whether those teams were paying, active or retained was not disclosed.

The product launched on Product Hunt on August 28, 2026 and ranked fifth that day (Product Hunt leaderboard). Product Hunt also displayed approximately 429 followers and one review when researched. These indicate initial attention only—not product-market fit.

Additional positive signals include public pricing, a working signup flow, extensive product documentation, a trust center and support for self-service deployment. However, no named customer case studies, revenue figures, usage volumes, retention cohorts, expansion data or independently verified reviews were found.

Traction Assessment: Product deployment is credible, but commercial traction is extremely early and almost entirely company-reported.

Competitive Position

The most direct competitor is Claude Tag. Anthropic describes a Slack-based, multiplayer agent with selected channel access, persistent context, asynchronous tasks, proactive behavior, tool integrations, spend controls and activity logs. It is available in beta to Claude Team and Enterprise customers (Anthropic announcement).

Other competitors include:

Slack’s native AI and Slackbot, which summarize content, search connected sources and support workflow automation (Slack AI).
Microsoft 365 Copilot and agents, grounded in Microsoft 365 and business data (Microsoft).
Glean, offering permissions-aware enterprise search and workflow agents in Slack across more than 275 connectors (Glean).
Lindy, which offers a Slack-native teammate, persistent context, approvals, model selection and scheduled routines (Lindy pricing).
Manual alternatives such as Slack search, ChatGPT or Claude, Zapier/Make workflows and human operations staff.

OpenTag’s main differentiators are model-agnostic routing, one workspace-wide price instead of per-seat pricing, a self-maintaining wiki and proactive automation discovery. Its claimed 70% reduction in model spending is company-reported and not independently verified.

A separate concern is the existence of an unrelated, MIT-licensed project also called OpenTag, maintained by CopilotKit as a self-hosted Slack and Teams agent starter (GitHub). This creates potential brand, search-discovery and open-source substitution risk.

If the largest platform in this market launched the same feature within six months, why would customers continue using this product? The current answer is model neutrality, lower workspace-level pricing and accumulated workflow memory. Those advantages may help, but they do not yet establish substantial switching costs.

Defensibility Assessment: Low

Business Model and Economics

OpenTag combines recurring subscriptions with metered usage. This aligns revenue with workload and avoids the underutilized-seat problem. Enterprise controls and custom support could raise average contract value.

Variable costs include model inference, sandboxed cloud environments, storage, integration traffic, observability and support. Model routing may reduce cost by sending easier tasks to cheaper models, but the company’s “70% lower model spend” claim lacks an independently published methodology.

Gross margin is not publicly disclosed. A usage plan can support attractive margins only if credit consumption is priced consistently above model, browser-compute and infrastructure costs. Failed partial runs still consume credits, reducing OpenTag’s direct exposure, but potentially harming customer satisfaction (credit documentation).

The $50 entry price reduces adoption friction but limits revenue unless teams expand usage substantially. The company therefore needs strong credit expansion, enterprise conversion or both.

Unicorn Path

A high-growth B2B software company with durable retention and acceptable margins might receive an illustrative 10× ARR multiple. Therefore:

\$1\text{ billion} \div 10 = \$100\text{ million ARR}

At current pricing, $100 million ARR would require approximately:

166,700 customers on the $50-per-month plan;
8,334 customers on the $1,000-per-month plan; or
5,000 enterprise customers at an assumed $20,000 ACV.

A more plausible mixed scenario would be 20,000 customers averaging $5,000 annually. That is 10% of Slack’s reported paid-customer base before considering Teams, but achieving it would require robust self-service conversion, international distribution and materially better retention than is currently evidenced.

OpenTag would also need deeper workflow integrations, security certification, enterprise administration, repeatable acquisition and proprietary workflow or organizational-memory advantages. Remaining merely a model router inside Slack is unlikely to sustain a unicorn outcome.

Unicorn Path: Conditional

Valuation Assessment

OpenTag is listed as a YC S26 company. YC’s published standard deal invests $500,000: $125,000 for 7% and $375,000 through an uncapped MFN SAFE (YC standard deal). This confirms the accelerator’s standard structure but does not establish a current company valuation.

No other round, valuation, SAFE cap, revenue multiple or fundraising terms were publicly verified.

Valuation Attractiveness: Not Assessable

Assessment requires current ARR, growth, retention, gross margin, burn, cash runway, cap table, outstanding SAFEs, proposed round size, valuation cap and liquidation terms.

Key Risks

Minimal validated traction: Ten teams do not demonstrate paid demand or retention.
Direct platform competition: Claude Tag closely matches OpenTag’s core user experience.
Low defensibility: Model routing and chat integrations can be replicated.
Security maturity: SOC 2 Type II is in progress, not completed (security page).
Sensitive-data exposure: The product processes Slack, Google Workspace and connected-application data through managed environments and model providers.
Unknown gross margin: Multi-step agents and sandboxed compute may be expensive.
Platform dependency: Slack or Teams policy, API and native-product changes could impair distribution.
Brand confusion: An unrelated open-source CopilotKit product uses the same OpenTag name.
Pivot history: The commercial outcome and liabilities of the founders’ previous company are unclear.

Final Assessment

Venture Potential: 65/100

Category	Score
Market Size and Expansion Potential	17/20
Traction and Growth Evidence	5/20
Founder and Team	12/15
Product Strength	8/10
Distribution Potential	10/15
Business Model and Economics	8/10
Defensibility	5/10
Total	65/100

The strongest elements are market size, technical execution and clear pricing. The weakest are commercially unverified traction and limited defensibility against platform owners and well-funded competitors.

Evidence Confidence: 57/100

Founder identity, legal entity, YC participation, pricing, product functionality and security posture are supported by first-party sources. The ten-team deployment figure is company-reported. Market and unicorn calculations are analyst scenarios.

Revenue, paid customers, retention, usage, gross margin, acquisition cost, burn, runway, customer references and current financing terms remain unavailable. Minor documentation inconsistencies—such as older Gini branding and terms referring only to Slack while current pages include Teams—also reduce confidence.

Final Decision: Watch

OpenTag is potentially venture-backable but too early for formal diligence based on public evidence. The exact overlap with Claude Tag and Lindy means rapid early traction is necessary; product quality alone is insufficient.

Upgrade Conditions

At least $500,000 ARR with verified monthly growth.
More than 100 paying workspaces, including referenceable customers.
Greater than 70% six-month logo retention.
Evidence of material expansion in monthly credit usage.
Gross margin above 70% after model and sandbox costs.
Completed SOC 2 Type II audit.
Repeatable acquisition beyond Product Hunt and YC exposure.
Evidence that model routing, company memory or workflow learning creates measurable switching costs.

Downgrade Conditions

Most early-access teams fail to convert or retain.
Anthropic, Slack or Microsoft eliminates the price or functionality advantage.
Gross margin remains structurally weak at current credit prices.
Material security, permissioning or data-loss incidents.
Continued product or brand confusion with the CopilotKit OpenTag project.
Loss of founder commitment or another major pivot without validated learning.

Questions for Further Diligence

How many of the ten reported teams are active paying customers?
What are current MRR, monthly growth and average revenue per workspace?
What percentage of trial workspaces convert to paid plans?
What are 30-, 90- and 180-day workspace retention rates?
How many tasks and credits does the median retained workspace consume monthly?
What is gross margin by plan after model, browser-compute and infrastructure costs?
Which acquisition channels have produced retained customers, and at what CAC?
How does output quality and cost compare experimentally with Claude Tag and Lindy?
Which workflow data or organizational memory remains proprietary and difficult to export?
What happened commercially to Lilac Labs, and which entity owns its assets and liabilities?
What is the current cap table, including both YC SAFEs and any prior or related-company investors?
What round is currently being raised, at what valuation, and with what runway and hiring plan?